SpamHam.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# SpamHam Text Classification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T12:32:00.752810Z",
     "start_time": "2018-08-30T12:32:00.746377Z"
    }
   },
   "source": [
    "<b> Imports</b>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:44:48.529010Z",
     "start_time": "2018-08-30T14:44:44.439307Z"
    }
   },
   "outputs": [],
   "source": [
    "import os\n",
    "from collections import Counter\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import random\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Making the data sets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T12:33:55.247744Z",
     "start_time": "2018-08-30T12:33:55.232767Z"
    }
   },
   "outputs": [],
   "source": [
    "# I've copied all spam and ham emails in singel directory 'emails', now making the data.\n",
    "def make_data():\n",
    "    direc = 'emails/'\n",
    "    files = os.listdir(direc)\n",
    "    emails = np.array([direc+email for email in files])\n",
    "    \n",
    "    random.shuffle(emails)\n",
    "    \n",
    "    email_col = np.array([[]])\n",
    "    label = np.array([[]])\n",
    "    c = len(emails)\n",
    "    for email in emails:\n",
    "        f = open(email, encoding='utf8', errors='ignore')\n",
    "        blob = f.read()\n",
    "        email_col = np.append(email_col, blob)\n",
    "        if 'ham' in email:\n",
    "            label = np.append(label, 'ham')\n",
    "        if 'spam' in email:\n",
    "            label = np.append(label, 'spam')\n",
    "        c-=1\n",
    "        print(c)\n",
    "        \n",
    "    return email_col, label\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T12:44:26.470520Z",
     "start_time": "2018-08-30T12:34:35.130608Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5171\n",
      "5170\n",
      "5169\n",
      "5168\n",
      "5167\n",
      "5166\n",
      "5165\n",
      "5164\n",
      "5163\n",
      "5162\n",
      "5161\n",
      "5160\n",
      "5159\n",
      "5158\n",
      "5157\n",
      "5156\n",
      "5155\n",
      "5154\n",
      "5153\n",
      "5152\n",
      "5151\n",
      "5150\n",
      "5149\n",
      "5148\n",
      "5147\n",
      "5146\n",
      "5145\n",
      "5144\n",
      "5143\n",
      "5142\n",
      "5141\n",
      "5140\n",
      "5139\n",
      "5138\n",
      "5137\n",
      "5136\n",
      "5135\n",
      "5134\n",
      "5133\n",
      "5132\n",
      "5131\n",
      "5130\n",
      "5129\n",
      "5128\n",
      "5127\n",
      "5126\n",
      "5125\n",
      "5124\n",
      "5123\n",
      "5122\n",
      "5121\n",
      "5120\n",
      "5119\n",
      "5118\n",
      "5117\n",
      "5116\n",
      "5115\n",
      "5114\n",
      "5113\n",
      "5112\n",
      "5111\n",
      "5110\n",
      "5109\n",
      "5108\n",
      "5107\n",
      "5106\n",
      "5105\n",
      "5104\n",
      "5103\n",
      "5102\n",
      "5101\n",
      "5100\n",
      "5099\n",
      "5098\n",
      "5097\n",
      "5096\n",
      "5095\n",
      "5094\n",
      "5093\n",
      "5092\n",
      "5091\n",
      "5090\n",
      "5089\n",
      "5088\n",
      "5087\n",
      "5086\n",
      "5085\n",
      "5084\n",
      "5083\n",
      "5082\n",
      "5081\n",
      "5080\n",
      "5079\n",
      "5078\n",
      "5077\n",
      "5076\n",
      "5075\n",
      "5074\n",
      "5073\n",
      "5072\n",
      "5071\n",
      "5070\n",
      "5069\n",
      "5068\n",
      "5067\n",
      "5066\n",
      "5065\n",
      "5064\n",
      "5063\n",
      "5062\n",
      "5061\n",
      "5060\n",
      "5059\n",
      "5058\n",
      "5057\n",
      "5056\n",
      "5055\n",
      "5054\n",
      "5053\n",
      "5052\n",
      "5051\n",
      "5050\n",
      "5049\n",
      "5048\n",
      "5047\n",
      "5046\n",
      "5045\n",
      "5044\n",
      "5043\n",
      "5042\n",
      "5041\n",
      "5040\n",
      "5039\n",
      "5038\n",
      "5037\n",
      "5036\n",
      "5035\n",
      "5034\n",
      "5033\n",
      "5032\n",
      "5031\n",
      "5030\n",
      "5029\n",
      "5028\n",
      "5027\n",
      "5026\n",
      "5025\n",
      "5024\n",
      "5023\n",
      "5022\n",
      "5021\n",
      "5020\n",
      "5019\n",
      "5018\n",
      "5017\n",
      "5016\n",
      "5015\n",
      "5014\n",
      "5013\n",
      "5012\n",
      "5011\n",
      "5010\n",
      "5009\n",
      "5008\n",
      "5007\n",
      "5006\n",
      "5005\n",
      "5004\n",
      "5003\n",
      "5002\n",
      "5001\n",
      "5000\n",
      "4999\n",
      "4998\n",
      "4997\n",
      "4996\n",
      "4995\n",
      "4994\n",
      "4993\n",
      "4992\n",
      "4991\n",
      "4990\n",
      "4989\n",
      "4988\n",
      "4987\n",
      "4986\n",
      "4985\n",
      "4984\n",
      "4983\n",
      "4982\n",
      "4981\n",
      "4980\n",
      "4979\n",
      "4978\n",
      "4977\n",
      "4976\n",
      "4975\n",
      "4974\n",
      "4973\n",
      "4972\n",
      "4971\n",
      "4970\n",
      "4969\n",
      "4968\n",
      "4967\n",
      "4966\n",
      "4965\n",
      "4964\n",
      "4963\n",
      "4962\n",
      "4961\n",
      "4960\n",
      "4959\n",
      "4958\n",
      "4957\n",
      "4956\n",
      "4955\n",
      "4954\n",
      "4953\n",
      "4952\n",
      "4951\n",
      "4950\n",
      "4949\n",
      "4948\n",
      "4947\n",
      "4946\n",
      "4945\n",
      "4944\n",
      "4943\n",
      "4942\n",
      "4941\n",
      "4940\n",
      "4939\n",
      "4938\n",
      "4937\n",
      "4936\n",
      "4935\n",
      "4934\n",
      "4933\n",
      "4932\n",
      "4931\n",
      "4930\n",
      "4929\n",
      "4928\n",
      "4927\n",
      "4926\n",
      "4925\n",
      "4924\n",
      "4923\n",
      "4922\n",
      "4921\n",
      "4920\n",
      "4919\n",
      "4918\n",
      "4917\n",
      "4916\n",
      "4915\n",
      "4914\n",
      "4913\n",
      "4912\n",
      "4911\n",
      "4910\n",
      "4909\n",
      "4908\n",
      "4907\n",
      "4906\n",
      "4905\n",
      "4904\n",
      "4903\n",
      "4902\n",
      "4901\n",
      "4900\n",
      "4899\n",
      "4898\n",
      "4897\n",
      "4896\n",
      "4895\n",
      "4894\n",
      "4893\n",
      "4892\n",
      "4891\n",
      "4890\n",
      "4889\n",
      "4888\n",
      "4887\n",
      "4886\n",
      "4885\n",
      "4884\n",
      "4883\n",
      "4882\n",
      "4881\n",
      "4880\n",
      "4879\n",
      "4878\n",
      "4877\n",
      "4876\n",
      "4875\n",
      "4874\n",
      "4873\n",
      "4872\n",
      "4871\n",
      "4870\n",
      "4869\n",
      "4868\n",
      "4867\n",
      "4866\n",
      "4865\n",
      "4864\n",
      "4863\n",
      "4862\n",
      "4861\n",
      "4860\n",
      "4859\n",
      "4858\n",
      "4857\n",
      "4856\n",
      "4855\n",
      "4854\n",
      "4853\n",
      "4852\n",
      "4851\n",
      "4850\n",
      "4849\n",
      "4848\n",
      "4847\n",
      "4846\n",
      "4845\n",
      "4844\n",
      "4843\n",
      "4842\n",
      "4841\n",
      "4840\n",
      "4839\n",
      "4838\n",
      "4837\n",
      "4836\n",
      "4835\n",
      "4834\n",
      "4833\n",
      "4832\n",
      "4831\n",
      "4830\n",
      "4829\n",
      "4828\n",
      "4827\n",
      "4826\n",
      "4825\n",
      "4824\n",
      "4823\n",
      "4822\n",
      "4821\n",
      "4820\n",
      "4819\n",
      "4818\n",
      "4817\n",
      "4816\n",
      "4815\n",
      "4814\n",
      "4813\n",
      "4812\n",
      "4811\n",
      "4810\n",
      "4809\n",
      "4808\n",
      "4807\n",
      "4806\n",
      "4805\n",
      "4804\n",
      "4803\n",
      "4802\n",
      "4801\n",
      "4800\n",
      "4799\n",
      "4798\n",
      "4797\n",
      "4796\n",
      "4795\n",
      "4794\n",
      "4793\n",
      "4792\n",
      "4791\n",
      "4790\n",
      "4789\n",
      "4788\n",
      "4787\n",
      "4786\n",
      "4785\n",
      "4784\n",
      "4783\n",
      "4782\n",
      "4781\n",
      "4780\n",
      "4779\n",
      "4778\n",
      "4777\n",
      "4776\n",
      "4775\n",
      "4774\n",
      "4773\n",
      "4772\n",
      "4771\n",
      "4770\n",
      "4769\n",
      "4768\n",
      "4767\n",
      "4766\n",
      "4765\n",
      "4764\n",
      "4763\n",
      "4762\n",
      "4761\n",
      "4760\n",
      "4759\n",
      "4758\n",
      "4757\n",
      "4756\n",
      "4755\n",
      "4754\n",
      "4753\n",
      "4752\n",
      "4751\n",
      "4750\n",
      "4749\n",
      "4748\n",
      "4747\n",
      "4746\n",
      "4745\n",
      "4744\n",
      "4743\n",
      "4742\n",
      "4741\n",
      "4740\n",
      "4739\n",
      "4738\n",
      "4737\n",
      "4736\n",
      "4735\n",
      "4734\n",
      "4733\n",
      "4732\n",
      "4731\n",
      "4730\n",
      "4729\n",
      "4728\n",
      "4727\n",
      "4726\n",
      "4725\n",
      "4724\n",
      "4723\n",
      "4722\n",
      "4721\n",
      "4720\n",
      "4719\n",
      "4718\n",
      "4717\n",
      "4716\n",
      "4715\n",
      "4714\n",
      "4713\n",
      "4712\n",
      "4711\n",
      "4710\n",
      "4709\n",
      "4708\n",
      "4707\n",
      "4706\n",
      "4705\n",
      "4704\n",
      "4703\n",
      "4702\n",
      "4701\n",
      "4700\n",
      "4699\n",
      "4698\n",
      "4697\n",
      "4696\n",
      "4695\n",
      "4694\n",
      "4693\n",
      "4692\n",
      "4691\n",
      "4690\n",
      "4689\n",
      "4688\n",
      "4687\n",
      "4686\n",
      "4685\n",
      "4684\n",
      "4683\n",
      "4682\n",
      "4681\n",
      "4680\n",
      "4679\n",
      "4678\n",
      "4677\n",
      "4676\n",
      "4675\n",
      "4674\n",
      "4673\n",
      "4672\n",
      "4671\n",
      "4670\n",
      "4669\n",
      "4668\n",
      "4667\n",
      "4666\n",
      "4665\n",
      "4664\n",
      "4663\n",
      "4662\n",
      "4661\n",
      "4660\n",
      "4659\n",
      "4658\n",
      "4657\n",
      "4656\n",
      "4655\n",
      "4654\n",
      "4653\n",
      "4652\n",
      "4651\n",
      "4650\n",
      "4649\n",
      "4648\n",
      "4647\n",
      "4646\n",
      "4645\n",
      "4644\n",
      "4643\n",
      "4642\n",
      "4641\n",
      "4640\n",
      "4639\n",
      "4638\n",
      "4637\n",
      "4636\n",
      "4635\n",
      "4634\n",
      "4633\n",
      "4632\n",
      "4631\n",
      "4630\n",
      "4629\n",
      "4628\n",
      "4627\n",
      "4626\n",
      "4625\n",
      "4624\n",
      "4623\n",
      "4622\n",
      "4621\n",
      "4620\n",
      "4619\n",
      "4618\n",
      "4617\n",
      "4616\n",
      "4615\n",
      "4614\n",
      "4613\n",
      "4612\n",
      "4611\n",
      "4610\n",
      "4609\n",
      "4608\n",
      "4607\n",
      "4606\n",
      "4605\n",
      "4604\n",
      "4603\n",
      "4602\n",
      "4601\n",
      "4600\n",
      "4599\n",
      "4598\n",
      "4597\n",
      "4596\n",
      "4595\n",
      "4594\n",
      "4593\n",
      "4592\n",
      "4591\n",
      "4590\n",
      "4589\n",
      "4588\n",
      "4587\n",
      "4586\n",
      "4585\n",
      "4584\n",
      "4583\n",
      "4582\n",
      "4581\n",
      "4580\n",
      "4579\n",
      "4578\n",
      "4577\n",
      "4576\n",
      "4575\n",
      "4574\n",
      "4573\n",
      "4572\n",
      "4571\n",
      "4570\n",
      "4569\n",
      "4568\n",
      "4567\n",
      "4566\n",
      "4565\n",
      "4564\n",
      "4563\n",
      "4562\n",
      "4561\n",
      "4560\n",
      "4559\n",
      "4558\n",
      "4557\n",
      "4556\n",
      "4555\n",
      "4554\n",
      "4553\n",
      "4552\n",
      "4551\n",
      "4550\n",
      "4549\n",
      "4548\n",
      "4547\n",
      "4546\n",
      "4545\n",
      "4544\n",
      "4543\n",
      "4542\n",
      "4541\n",
      "4540\n",
      "4539\n",
      "4538\n",
      "4537\n",
      "4536\n",
      "4535\n",
      "4534\n",
      "4533\n",
      "4532\n",
      "4531\n",
      "4530\n",
      "4529\n",
      "4528\n",
      "4527\n",
      "4526\n",
      "4525\n",
      "4524\n",
      "4523\n",
      "4522\n",
      "4521\n",
      "4520\n",
      "4519\n",
      "4518\n",
      "4517\n",
      "4516\n",
      "4515\n",
      "4514\n",
      "4513\n",
      "4512\n",
      "4511\n",
      "4510\n",
      "4509\n",
      "4508\n",
      "4507\n",
      "4506\n",
      "4505\n",
      "4504\n",
      "4503\n",
      "4502\n",
      "4501\n",
      "4500\n",
      "4499\n",
      "4498\n",
      "4497\n",
      "4496\n",
      "4495\n",
      "4494\n",
      "4493\n",
      "4492\n",
      "4491\n",
      "4490\n",
      "4489\n",
      "4488\n",
      "4487\n",
      "4486\n",
      "4485\n",
      "4484\n",
      "4483\n",
      "4482\n",
      "4481\n",
      "4480\n",
      "4479\n",
      "4478\n",
      "4477\n",
      "4476\n",
      "4475\n",
      "4474\n",
      "4473\n",
      "4472\n",
      "4471\n",
      "4470\n",
      "4469\n",
      "4468\n",
      "4467\n",
      "4466\n",
      "4465\n",
      "4464\n",
      "4463\n",
      "4462\n",
      "4461\n",
      "4460\n",
      "4459\n",
      "4458\n",
      "4457\n",
      "4456\n",
      "4455\n",
      "4454\n",
      "4453\n",
      "4452\n",
      "4451\n",
      "4450\n",
      "4449\n",
      "4448\n",
      "4447\n",
      "4446\n",
      "4445\n",
      "4444\n",
      "4443\n",
      "4442\n",
      "4441\n",
      "4440\n",
      "4439\n",
      "4438\n",
      "4437\n",
      "4436\n",
      "4435\n",
      "4434\n",
      "4433\n",
      "4432\n",
      "4431\n",
      "4430\n",
      "4429\n",
      "4428\n",
      "4427\n",
      "4426\n",
      "4425\n",
      "4424\n",
      "4423\n",
      "4422\n",
      "4421\n",
      "4420\n",
      "4419\n",
      "4418\n",
      "4417\n",
      "4416\n",
      "4415\n",
      "4414\n",
      "4413\n",
      "4412\n",
      "4411\n",
      "4410\n",
      "4409\n",
      "4408\n",
      "4407\n",
      "4406\n",
      "4405\n",
      "4404\n",
      "4403\n",
      "4402\n",
      "4401\n",
      "4400\n",
      "4399\n",
      "4398\n",
      "4397\n",
      "4396\n",
      "4395\n",
      "4394\n",
      "4393\n",
      "4392\n",
      "4391\n",
      "4390\n",
      "4389\n",
      "4388\n",
      "4387\n",
      "4386\n",
      "4385\n",
      "4384\n",
      "4383\n",
      "4382\n",
      "4381\n",
      "4380\n",
      "4379\n",
      "4378\n",
      "4377\n",
      "4376\n",
      "4375\n",
      "4374\n",
      "4373\n",
      "4372\n",
      "4371\n",
      "4370\n",
      "4369\n",
      "4368\n",
      "4367\n",
      "4366\n",
      "4365\n",
      "4364\n",
      "4363\n",
      "4362\n",
      "4361\n",
      "4360\n",
      "4359\n",
      "4358\n",
      "4357\n",
      "4356\n",
      "4355\n",
      "4354\n",
      "4353\n",
      "4352\n",
      "4351\n",
      "4350\n",
      "4349\n",
      "4348\n",
      "4347\n",
      "4346\n",
      "4345\n",
      "4344\n",
      "4343\n",
      "4342\n",
      "4341\n",
      "4340\n",
      "4339\n",
      "4338\n",
      "4337\n",
      "4336\n",
      "4335\n",
      "4334\n",
      "4333\n",
      "4332\n",
      "4331\n",
      "4330\n",
      "4329\n",
      "4328\n",
      "4327\n",
      "4326\n",
      "4325\n",
      "4324\n",
      "4323\n",
      "4322\n",
      "4321\n",
      "4320\n",
      "4319\n",
      "4318\n",
      "4317\n",
      "4316\n",
      "4315\n",
      "4314\n",
      "4313\n",
      "4312\n",
      "4311\n",
      "4310\n",
      "4309\n",
      "4308\n",
      "4307\n",
      "4306\n",
      "4305\n",
      "4304\n",
      "4303\n",
      "4302\n",
      "4301\n",
      "4300\n",
      "4299\n",
      "4298\n",
      "4297\n",
      "4296\n",
      "4295\n",
      "4294\n",
      "4293\n",
      "4292\n",
      "4291\n",
      "4290\n",
      "4289\n",
      "4288\n",
      "4287\n",
      "4286\n",
      "4285\n",
      "4284\n",
      "4283\n",
      "4282\n",
      "4281\n",
      "4280\n",
      "4279\n",
      "4278\n",
      "4277\n",
      "4276\n",
      "4275\n",
      "4274\n",
      "4273\n",
      "4272\n",
      "4271\n",
      "4270\n",
      "4269\n",
      "4268\n",
      "4267\n",
      "4266\n",
      "4265\n",
      "4264\n",
      "4263\n",
      "4262\n",
      "4261\n",
      "4260\n",
      "4259\n",
      "4258\n",
      "4257\n",
      "4256\n",
      "4255\n",
      "4254\n",
      "4253\n",
      "4252\n",
      "4251\n",
      "4250\n",
      "4249\n",
      "4248\n",
      "4247\n",
      "4246\n",
      "4245\n",
      "4244\n",
      "4243\n",
      "4242\n",
      "4241\n",
      "4240\n",
      "4239\n",
      "4238\n",
      "4237\n",
      "4236\n",
      "4235\n",
      "4234\n",
      "4233\n",
      "4232\n",
      "4231\n",
      "4230\n",
      "4229\n",
      "4228\n",
      "4227\n",
      "4226\n",
      "4225\n",
      "4224\n",
      "4223\n",
      "4222\n",
      "4221\n",
      "4220\n",
      "4219\n",
      "4218\n",
      "4217\n",
      "4216\n",
      "4215\n",
      "4214\n",
      "4213\n",
      "4212\n",
      "4211\n",
      "4210\n",
      "4209\n",
      "4208\n",
      "4207\n",
      "4206\n",
      "4205\n",
      "4204\n",
      "4203\n",
      "4202\n",
      "4201\n",
      "4200\n",
      "4199\n",
      "4198\n",
      "4197\n",
      "4196\n",
      "4195\n",
      "4194\n",
      "4193\n",
      "4192\n",
      "4191\n",
      "4190\n",
      "4189\n",
      "4188\n",
      "4187\n",
      "4186\n",
      "4185\n",
      "4184\n",
      "4183\n",
      "4182\n",
      "4181\n",
      "4180\n",
      "4179\n",
      "4178\n",
      "4177\n",
      "4176\n",
      "4175\n",
      "4174\n",
      "4173\n",
      "4172\n",
      "4171\n",
      "4170\n",
      "4169\n",
      "4168\n",
      "4167\n",
      "4166\n",
      "4165\n",
      "4164\n",
      "4163\n",
      "4162\n",
      "4161\n",
      "4160\n",
      "4159\n",
      "4158\n",
      "4157\n",
      "4156\n",
      "4155\n",
      "4154\n",
      "4153\n",
      "4152\n",
      "4151\n",
      "4150\n",
      "4149\n",
      "4148\n",
      "4147\n",
      "4146\n",
      "4145\n",
      "4144\n",
      "4143\n",
      "4142\n",
      "4141\n",
      "4140\n",
      "4139\n",
      "4138\n",
      "4137\n",
      "4136\n",
      "4135\n",
      "4134\n",
      "4133\n",
      "4132\n",
      "4131\n",
      "4130\n",
      "4129\n",
      "4128\n",
      "4127\n",
      "4126\n",
      "4125\n",
      "4124\n",
      "4123\n",
      "4122\n",
      "4121\n",
      "4120\n",
      "4119\n",
      "4118\n",
      "4117\n",
      "4116\n",
      "4115\n",
      "4114\n",
      "4113\n",
      "4112\n",
      "4111\n",
      "4110\n",
      "4109\n",
      "4108\n",
      "4107\n",
      "4106\n",
      "4105\n",
      "4104\n",
      "4103\n",
      "4102\n",
      "4101\n",
      "4100\n",
      "4099\n",
      "4098\n",
      "4097\n",
      "4096\n",
      "4095\n",
      "4094\n",
      "4093\n",
      "4092\n",
      "4091\n",
      "4090\n",
      "4089\n",
      "4088\n",
      "4087\n",
      "4086\n",
      "4085\n",
      "4084\n",
      "4083\n",
      "4082\n",
      "4081\n",
      "4080\n",
      "4079\n",
      "4078\n",
      "4077\n",
      "4076\n",
      "4075\n",
      "4074\n",
      "4073\n",
      "4072\n",
      "4071\n",
      "4070\n",
      "4069\n",
      "4068\n",
      "4067\n",
      "4066\n",
      "4065\n",
      "4064\n",
      "4063\n",
      "4062\n",
      "4061\n",
      "4060\n",
      "4059\n",
      "4058\n",
      "4057\n",
      "4056\n",
      "4055\n",
      "4054\n",
      "4053\n",
      "4052\n",
      "4051\n",
      "4050\n",
      "4049\n",
      "4048\n",
      "4047\n",
      "4046\n",
      "4045\n",
      "4044\n",
      "4043\n",
      "4042\n",
      "4041\n",
      "4040\n",
      "4039\n",
      "4038\n",
      "4037\n",
      "4036\n",
      "4035\n",
      "4034\n",
      "4033\n",
      "4032\n",
      "4031\n",
      "4030\n",
      "4029\n",
      "4028\n",
      "4027\n",
      "4026\n",
      "4025\n",
      "4024\n",
      "4023\n",
      "4022\n",
      "4021\n",
      "4020\n",
      "4019\n",
      "4018\n",
      "4017\n",
      "4016\n",
      "4015\n",
      "4014\n",
      "4013\n",
      "4012\n",
      "4011\n",
      "4010\n",
      "4009\n",
      "4008\n",
      "4007\n",
      "4006\n",
      "4005\n",
      "4004\n",
      "4003\n",
      "4002\n",
      "4001\n",
      "4000\n",
      "3999\n",
      "3998\n",
      "3997\n",
      "3996\n",
      "3995\n",
      "3994\n",
      "3993\n",
      "3992\n",
      "3991\n",
      "3990\n",
      "3989\n",
      "3988\n",
      "3987\n",
      "3986\n",
      "3985\n",
      "3984\n",
      "3983\n",
      "3982\n",
      "3981\n",
      "3980\n",
      "3979\n",
      "3978\n",
      "3977\n",
      "3976\n",
      "3975\n",
      "3974\n",
      "3973\n",
      "3972\n",
      "3971\n",
      "3970\n",
      "3969\n",
      "3968\n",
      "3967\n",
      "3966\n",
      "3965\n",
      "3964\n",
      "3963\n",
      "3962\n",
      "3961\n",
      "3960\n",
      "3959\n",
      "3958\n",
      "3957\n",
      "3956\n",
      "3955\n",
      "3954\n",
      "3953\n",
      "3952\n",
      "3951\n",
      "3950\n",
      "3949\n",
      "3948\n",
      "3947\n",
      "3946\n",
      "3945\n",
      "3944\n",
      "3943\n",
      "3942\n",
      "3941\n",
      "3940\n",
      "3939\n",
      "3938\n",
      "3937\n",
      "3936\n",
      "3935\n",
      "3934\n",
      "3933\n",
      "3932\n",
      "3931\n",
      "3930\n",
      "3929\n",
      "3928\n",
      "3927\n",
      "3926\n",
      "3925\n",
      "3924\n",
      "3923\n",
      "3922\n",
      "3921\n",
      "3920\n",
      "3919\n",
      "3918\n",
      "3917\n",
      "3916\n",
      "3915\n",
      "3914\n",
      "3913\n",
      "3912\n",
      "3911\n",
      "3910\n",
      "3909\n",
      "3908\n",
      "3907\n",
      "3906\n",
      "3905\n",
      "3904\n",
      "3903\n",
      "3902\n",
      "3901\n",
      "3900\n",
      "3899\n",
      "3898\n",
      "3897\n",
      "3896\n",
      "3895\n",
      "3894\n",
      "3893\n",
      "3892\n",
      "3891\n",
      "3890\n",
      "3889\n",
      "3888\n",
      "3887\n",
      "3886\n",
      "3885\n",
      "3884\n",
      "3883\n",
      "3882\n",
      "3881\n",
      "3880\n",
      "3879\n",
      "3878\n",
      "3877\n",
      "3876\n",
      "3875\n",
      "3874\n",
      "3873\n",
      "3872\n",
      "3871\n",
      "3870\n",
      "3869\n",
      "3868\n",
      "3867\n",
      "3866\n",
      "3865\n",
      "3864\n",
      "3863\n",
      "3862\n",
      "3861\n",
      "3860\n",
      "3859\n",
      "3858\n",
      "3857\n",
      "3856\n",
      "3855\n",
      "3854\n",
      "3853\n",
      "3852\n",
      "3851\n",
      "3850\n",
      "3849\n",
      "3848\n",
      "3847\n",
      "3846\n",
      "3845\n",
      "3844\n",
      "3843\n",
      "3842\n",
      "3841\n",
      "3840\n",
      "3839\n",
      "3838\n",
      "3837\n",
      "3836\n",
      "3835\n",
      "3834\n",
      "3833\n",
      "3832\n",
      "3831\n",
      "3830\n",
      "3829\n",
      "3828\n",
      "3827\n",
      "3826\n",
      "3825\n",
      "3824\n",
      "3823\n",
      "3822\n",
      "3821\n",
      "3820\n",
      "3819\n",
      "3818\n",
      "3817\n",
      "3816\n",
      "3815\n",
      "3814\n",
      "3813\n",
      "3812\n",
      "3811\n",
      "3810\n",
      "3809\n",
      "3808\n",
      "3807\n",
      "3806\n",
      "3805\n",
      "3804\n",
      "3803\n",
      "3802\n",
      "3801\n",
      "3800\n",
      "3799\n",
      "3798\n",
      "3797\n",
      "3796\n",
      "3795\n",
      "3794\n",
      "3793\n",
      "3792\n",
      "3791\n",
      "3790\n",
      "3789\n",
      "3788\n",
      "3787\n",
      "3786\n",
      "3785\n",
      "3784\n",
      "3783\n",
      "3782\n",
      "3781\n",
      "3780\n",
      "3779\n",
      "3778\n",
      "3777\n",
      "3776\n",
      "3775\n",
      "3774\n",
      "3773\n",
      "3772\n",
      "3771\n",
      "3770\n",
      "3769\n",
      "3768\n",
      "3767\n",
      "3766\n",
      "3765\n",
      "3764\n",
      "3763\n",
      "3762\n",
      "3761\n",
      "3760\n",
      "3759\n",
      "3758\n",
      "3757\n",
      "3756\n",
      "3755\n",
      "3754\n",
      "3753\n",
      "3752\n",
      "3751\n",
      "3750\n",
      "3749\n",
      "3748\n",
      "3747\n",
      "3746\n",
      "3745\n",
      "3744\n",
      "3743\n",
      "3742\n",
      "3741\n",
      "3740\n",
      "3739\n",
      "3738\n",
      "3737\n",
      "3736\n",
      "3735\n",
      "3734\n",
      "3733\n",
      "3732\n",
      "3731\n",
      "3730\n",
      "3729\n",
      "3728\n",
      "3727\n",
      "3726\n",
      "3725\n",
      "3724\n",
      "3723\n",
      "3722\n",
      "3721\n",
      "3720\n",
      "3719\n",
      "3718\n",
      "3717\n",
      "3716\n",
      "3715\n",
      "3714\n",
      "3713\n",
      "3712\n",
      "3711\n",
      "3710\n",
      "3709\n",
      "3708\n",
      "3707\n",
      "3706\n",
      "3705\n",
      "3704\n",
      "3703\n",
      "3702\n",
      "3701\n",
      "3700\n",
      "3699\n",
      "3698\n",
      "3697\n",
      "3696\n",
      "3695\n",
      "3694\n",
      "3693\n",
      "3692\n",
      "3691\n",
      "3690\n",
      "3689\n",
      "3688\n",
      "3687\n",
      "3686\n",
      "3685\n",
      "3684\n",
      "3683\n",
      "3682\n",
      "3681\n",
      "3680\n",
      "3679\n",
      "3678\n",
      "3677\n",
      "3676\n",
      "3675\n",
      "3674\n",
      "3673\n",
      "3672\n",
      "3671\n",
      "3670\n",
      "3669\n",
      "3668\n",
      "3667\n",
      "3666\n",
      "3665\n",
      "3664\n",
      "3663\n",
      "3662\n",
      "3661\n",
      "3660\n",
      "3659\n",
      "3658\n",
      "3657\n",
      "3656\n",
      "3655\n",
      "3654\n",
      "3653\n",
      "3652\n",
      "3651\n",
      "3650\n",
      "3649\n",
      "3648\n",
      "3647\n",
      "3646\n",
      "3645\n",
      "3644\n",
      "3643\n",
      "3642\n",
      "3641\n",
      "3640\n",
      "3639\n",
      "3638\n",
      "3637\n",
      "3636\n",
      "3635\n",
      "3634\n",
      "3633\n",
      "3632\n",
      "3631\n",
      "3630\n",
      "3629\n",
      "3628\n",
      "3627\n",
      "3626\n",
      "3625\n",
      "3624\n",
      "3623\n",
      "3622\n",
      "3621\n",
      "3620\n",
      "3619\n",
      "3618\n",
      "3617\n",
      "3616\n",
      "3615\n",
      "3614\n",
      "3613\n",
      "3612\n",
      "3611\n",
      "3610\n",
      "3609\n",
      "3608\n",
      "3607\n",
      "3606\n",
      "3605\n",
      "3604\n",
      "3603\n",
      "3602\n",
      "3601\n",
      "3600\n",
      "3599\n",
      "3598\n",
      "3597\n",
      "3596\n",
      "3595\n",
      "3594\n",
      "3593\n",
      "3592\n",
      "3591\n",
      "3590\n",
      "3589\n",
      "3588\n",
      "3587\n",
      "3586\n",
      "3585\n",
      "3584\n",
      "3583\n",
      "3582\n",
      "3581\n",
      "3580\n",
      "3579\n",
      "3578\n",
      "3577\n",
      "3576\n",
      "3575\n",
      "3574\n",
      "3573\n",
      "3572\n",
      "3571\n",
      "3570\n",
      "3569\n",
      "3568\n",
      "3567\n",
      "3566\n",
      "3565\n",
      "3564\n",
      "3563\n",
      "3562\n",
      "3561\n",
      "3560\n",
      "3559\n",
      "3558\n",
      "3557\n",
      "3556\n",
      "3555\n",
      "3554\n",
      "3553\n",
      "3552\n",
      "3551\n",
      "3550\n",
      "3549\n",
      "3548\n",
      "3547\n",
      "3546\n",
      "3545\n",
      "3544\n",
      "3543\n",
      "3542\n",
      "3541\n",
      "3540\n",
      "3539\n",
      "3538\n",
      "3537\n",
      "3536\n",
      "3535\n",
      "3534\n",
      "3533\n",
      "3532\n",
      "3531\n",
      "3530\n",
      "3529\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3528\n",
      "3527\n",
      "3526\n",
      "3525\n",
      "3524\n",
      "3523\n",
      "3522\n",
      "3521\n",
      "3520\n",
      "3519\n",
      "3518\n",
      "3517\n",
      "3516\n",
      "3515\n",
      "3514\n",
      "3513\n",
      "3512\n",
      "3511\n",
      "3510\n",
      "3509\n",
      "3508\n",
      "3507\n",
      "3506\n",
      "3505\n",
      "3504\n",
      "3503\n",
      "3502\n",
      "3501\n",
      "3500\n",
      "3499\n",
      "3498\n",
      "3497\n",
      "3496\n",
      "3495\n",
      "3494\n",
      "3493\n",
      "3492\n",
      "3491\n",
      "3490\n",
      "3489\n",
      "3488\n",
      "3487\n",
      "3486\n",
      "3485\n",
      "3484\n",
      "3483\n",
      "3482\n",
      "3481\n",
      "3480\n",
      "3479\n",
      "3478\n",
      "3477\n",
      "3476\n",
      "3475\n",
      "3474\n",
      "3473\n",
      "3472\n",
      "3471\n",
      "3470\n",
      "3469\n",
      "3468\n",
      "3467\n",
      "3466\n",
      "3465\n",
      "3464\n",
      "3463\n",
      "3462\n",
      "3461\n",
      "3460\n",
      "3459\n",
      "3458\n",
      "3457\n",
      "3456\n",
      "3455\n",
      "3454\n",
      "3453\n",
      "3452\n",
      "3451\n",
      "3450\n",
      "3449\n",
      "3448\n",
      "3447\n",
      "3446\n",
      "3445\n",
      "3444\n",
      "3443\n",
      "3442\n",
      "3441\n",
      "3440\n",
      "3439\n",
      "3438\n",
      "3437\n",
      "3436\n",
      "3435\n",
      "3434\n",
      "3433\n",
      "3432\n",
      "3431\n",
      "3430\n",
      "3429\n",
      "3428\n",
      "3427\n",
      "3426\n",
      "3425\n",
      "3424\n",
      "3423\n",
      "3422\n",
      "3421\n",
      "3420\n",
      "3419\n",
      "3418\n",
      "3417\n",
      "3416\n",
      "3415\n",
      "3414\n",
      "3413\n",
      "3412\n",
      "3411\n",
      "3410\n",
      "3409\n",
      "3408\n",
      "3407\n",
      "3406\n",
      "3405\n",
      "3404\n",
      "3403\n",
      "3402\n",
      "3401\n",
      "3400\n",
      "3399\n",
      "3398\n",
      "3397\n",
      "3396\n",
      "3395\n",
      "3394\n",
      "3393\n",
      "3392\n",
      "3391\n",
      "3390\n",
      "3389\n",
      "3388\n",
      "3387\n",
      "3386\n",
      "3385\n",
      "3384\n",
      "3383\n",
      "3382\n",
      "3381\n",
      "3380\n",
      "3379\n",
      "3378\n",
      "3377\n",
      "3376\n",
      "3375\n",
      "3374\n",
      "3373\n",
      "3372\n",
      "3371\n",
      "3370\n",
      "3369\n",
      "3368\n",
      "3367\n",
      "3366\n",
      "3365\n",
      "3364\n",
      "3363\n",
      "3362\n",
      "3361\n",
      "3360\n",
      "3359\n",
      "3358\n",
      "3357\n",
      "3356\n",
      "3355\n",
      "3354\n",
      "3353\n",
      "3352\n",
      "3351\n",
      "3350\n",
      "3349\n",
      "3348\n",
      "3347\n",
      "3346\n",
      "3345\n",
      "3344\n",
      "3343\n",
      "3342\n",
      "3341\n",
      "3340\n",
      "3339\n",
      "3338\n",
      "3337\n",
      "3336\n",
      "3335\n",
      "3334\n",
      "3333\n",
      "3332\n",
      "3331\n",
      "3330\n",
      "3329\n",
      "3328\n",
      "3327\n",
      "3326\n",
      "3325\n",
      "3324\n",
      "3323\n",
      "3322\n",
      "3321\n",
      "3320\n",
      "3319\n",
      "3318\n",
      "3317\n",
      "3316\n",
      "3315\n",
      "3314\n",
      "3313\n",
      "3312\n",
      "3311\n",
      "3310\n",
      "3309\n",
      "3308\n",
      "3307\n",
      "3306\n",
      "3305\n",
      "3304\n",
      "3303\n",
      "3302\n",
      "3301\n",
      "3300\n",
      "3299\n",
      "3298\n",
      "3297\n",
      "3296\n",
      "3295\n",
      "3294\n",
      "3293\n",
      "3292\n",
      "3291\n",
      "3290\n",
      "3289\n",
      "3288\n",
      "3287\n",
      "3286\n",
      "3285\n",
      "3284\n",
      "3283\n",
      "3282\n",
      "3281\n",
      "3280\n",
      "3279\n",
      "3278\n",
      "3277\n",
      "3276\n",
      "3275\n",
      "3274\n",
      "3273\n",
      "3272\n",
      "3271\n",
      "3270\n",
      "3269\n",
      "3268\n",
      "3267\n",
      "3266\n",
      "3265\n",
      "3264\n",
      "3263\n",
      "3262\n",
      "3261\n",
      "3260\n",
      "3259\n",
      "3258\n",
      "3257\n",
      "3256\n",
      "3255\n",
      "3254\n",
      "3253\n",
      "3252\n",
      "3251\n",
      "3250\n",
      "3249\n",
      "3248\n",
      "3247\n",
      "3246\n",
      "3245\n",
      "3244\n",
      "3243\n",
      "3242\n",
      "3241\n",
      "3240\n",
      "3239\n",
      "3238\n",
      "3237\n",
      "3236\n",
      "3235\n",
      "3234\n",
      "3233\n",
      "3232\n",
      "3231\n",
      "3230\n",
      "3229\n",
      "3228\n",
      "3227\n",
      "3226\n",
      "3225\n",
      "3224\n",
      "3223\n",
      "3222\n",
      "3221\n",
      "3220\n",
      "3219\n",
      "3218\n",
      "3217\n",
      "3216\n",
      "3215\n",
      "3214\n",
      "3213\n",
      "3212\n",
      "3211\n",
      "3210\n",
      "3209\n",
      "3208\n",
      "3207\n",
      "3206\n",
      "3205\n",
      "3204\n",
      "3203\n",
      "3202\n",
      "3201\n",
      "3200\n",
      "3199\n",
      "3198\n",
      "3197\n",
      "3196\n",
      "3195\n",
      "3194\n",
      "3193\n",
      "3192\n",
      "3191\n",
      "3190\n",
      "3189\n",
      "3188\n",
      "3187\n",
      "3186\n",
      "3185\n",
      "3184\n",
      "3183\n",
      "3182\n",
      "3181\n",
      "3180\n",
      "3179\n",
      "3178\n",
      "3177\n",
      "3176\n",
      "3175\n",
      "3174\n",
      "3173\n",
      "3172\n",
      "3171\n",
      "3170\n",
      "3169\n",
      "3168\n",
      "3167\n",
      "3166\n",
      "3165\n",
      "3164\n",
      "3163\n",
      "3162\n",
      "3161\n",
      "3160\n",
      "3159\n",
      "3158\n",
      "3157\n",
      "3156\n",
      "3155\n",
      "3154\n",
      "3153\n",
      "3152\n",
      "3151\n",
      "3150\n",
      "3149\n",
      "3148\n",
      "3147\n",
      "3146\n",
      "3145\n",
      "3144\n",
      "3143\n",
      "3142\n",
      "3141\n",
      "3140\n",
      "3139\n",
      "3138\n",
      "3137\n",
      "3136\n",
      "3135\n",
      "3134\n",
      "3133\n",
      "3132\n",
      "3131\n",
      "3130\n",
      "3129\n",
      "3128\n",
      "3127\n",
      "3126\n",
      "3125\n",
      "3124\n",
      "3123\n",
      "3122\n",
      "3121\n",
      "3120\n",
      "3119\n",
      "3118\n",
      "3117\n",
      "3116\n",
      "3115\n",
      "3114\n",
      "3113\n",
      "3112\n",
      "3111\n",
      "3110\n",
      "3109\n",
      "3108\n",
      "3107\n",
      "3106\n",
      "3105\n",
      "3104\n",
      "3103\n",
      "3102\n",
      "3101\n",
      "3100\n",
      "3099\n",
      "3098\n",
      "3097\n",
      "3096\n",
      "3095\n",
      "3094\n",
      "3093\n",
      "3092\n",
      "3091\n",
      "3090\n",
      "3089\n",
      "3088\n",
      "3087\n",
      "3086\n",
      "3085\n",
      "3084\n",
      "3083\n",
      "3082\n",
      "3081\n",
      "3080\n",
      "3079\n",
      "3078\n",
      "3077\n",
      "3076\n",
      "3075\n",
      "3074\n",
      "3073\n",
      "3072\n",
      "3071\n",
      "3070\n",
      "3069\n",
      "3068\n",
      "3067\n",
      "3066\n",
      "3065\n",
      "3064\n",
      "3063\n",
      "3062\n",
      "3061\n",
      "3060\n",
      "3059\n",
      "3058\n",
      "3057\n",
      "3056\n",
      "3055\n",
      "3054\n",
      "3053\n",
      "3052\n",
      "3051\n",
      "3050\n",
      "3049\n",
      "3048\n",
      "3047\n",
      "3046\n",
      "3045\n",
      "3044\n",
      "3043\n",
      "3042\n",
      "3041\n",
      "3040\n",
      "3039\n",
      "3038\n",
      "3037\n",
      "3036\n",
      "3035\n",
      "3034\n",
      "3033\n",
      "3032\n",
      "3031\n",
      "3030\n",
      "3029\n",
      "3028\n",
      "3027\n",
      "3026\n",
      "3025\n",
      "3024\n",
      "3023\n",
      "3022\n",
      "3021\n",
      "3020\n",
      "3019\n",
      "3018\n",
      "3017\n",
      "3016\n",
      "3015\n",
      "3014\n",
      "3013\n",
      "3012\n",
      "3011\n",
      "3010\n",
      "3009\n",
      "3008\n",
      "3007\n",
      "3006\n",
      "3005\n",
      "3004\n",
      "3003\n",
      "3002\n",
      "3001\n",
      "3000\n",
      "2999\n",
      "2998\n",
      "2997\n",
      "2996\n",
      "2995\n",
      "2994\n",
      "2993\n",
      "2992\n",
      "2991\n",
      "2990\n",
      "2989\n",
      "2988\n",
      "2987\n",
      "2986\n",
      "2985\n",
      "2984\n",
      "2983\n",
      "2982\n",
      "2981\n",
      "2980\n",
      "2979\n",
      "2978\n",
      "2977\n",
      "2976\n",
      "2975\n",
      "2974\n",
      "2973\n",
      "2972\n",
      "2971\n",
      "2970\n",
      "2969\n",
      "2968\n",
      "2967\n",
      "2966\n",
      "2965\n",
      "2964\n",
      "2963\n",
      "2962\n",
      "2961\n",
      "2960\n",
      "2959\n",
      "2958\n",
      "2957\n",
      "2956\n",
      "2955\n",
      "2954\n",
      "2953\n",
      "2952\n",
      "2951\n",
      "2950\n",
      "2949\n",
      "2948\n",
      "2947\n",
      "2946\n",
      "2945\n",
      "2944\n",
      "2943\n",
      "2942\n",
      "2941\n",
      "2940\n",
      "2939\n",
      "2938\n",
      "2937\n",
      "2936\n",
      "2935\n",
      "2934\n",
      "2933\n",
      "2932\n",
      "2931\n",
      "2930\n",
      "2929\n",
      "2928\n",
      "2927\n",
      "2926\n",
      "2925\n",
      "2924\n",
      "2923\n",
      "2922\n",
      "2921\n",
      "2920\n",
      "2919\n",
      "2918\n",
      "2917\n",
      "2916\n",
      "2915\n",
      "2914\n",
      "2913\n",
      "2912\n",
      "2911\n",
      "2910\n",
      "2909\n",
      "2908\n",
      "2907\n",
      "2906\n",
      "2905\n",
      "2904\n",
      "2903\n",
      "2902\n",
      "2901\n",
      "2900\n",
      "2899\n",
      "2898\n",
      "2897\n",
      "2896\n",
      "2895\n",
      "2894\n",
      "2893\n",
      "2892\n",
      "2891\n",
      "2890\n",
      "2889\n",
      "2888\n",
      "2887\n",
      "2886\n",
      "2885\n",
      "2884\n",
      "2883\n",
      "2882\n",
      "2881\n",
      "2880\n",
      "2879\n",
      "2878\n",
      "2877\n",
      "2876\n",
      "2875\n",
      "2874\n",
      "2873\n",
      "2872\n",
      "2871\n",
      "2870\n",
      "2869\n",
      "2868\n",
      "2867\n",
      "2866\n",
      "2865\n",
      "2864\n",
      "2863\n",
      "2862\n",
      "2861\n",
      "2860\n",
      "2859\n",
      "2858\n",
      "2857\n",
      "2856\n",
      "2855\n",
      "2854\n",
      "2853\n",
      "2852\n",
      "2851\n",
      "2850\n",
      "2849\n",
      "2848\n",
      "2847\n",
      "2846\n",
      "2845\n",
      "2844\n",
      "2843\n",
      "2842\n",
      "2841\n",
      "2840\n",
      "2839\n",
      "2838\n",
      "2837\n",
      "2836\n",
      "2835\n",
      "2834\n",
      "2833\n",
      "2832\n",
      "2831\n",
      "2830\n",
      "2829\n",
      "2828\n",
      "2827\n",
      "2826\n",
      "2825\n",
      "2824\n",
      "2823\n",
      "2822\n",
      "2821\n",
      "2820\n",
      "2819\n",
      "2818\n",
      "2817\n",
      "2816\n",
      "2815\n",
      "2814\n",
      "2813\n",
      "2812\n",
      "2811\n",
      "2810\n",
      "2809\n",
      "2808\n",
      "2807\n",
      "2806\n",
      "2805\n",
      "2804\n",
      "2803\n",
      "2802\n",
      "2801\n",
      "2800\n",
      "2799\n",
      "2798\n",
      "2797\n",
      "2796\n",
      "2795\n",
      "2794\n",
      "2793\n",
      "2792\n",
      "2791\n",
      "2790\n",
      "2789\n",
      "2788\n",
      "2787\n",
      "2786\n",
      "2785\n",
      "2784\n",
      "2783\n",
      "2782\n",
      "2781\n",
      "2780\n",
      "2779\n",
      "2778\n",
      "2777\n",
      "2776\n",
      "2775\n",
      "2774\n",
      "2773\n",
      "2772\n",
      "2771\n",
      "2770\n",
      "2769\n",
      "2768\n",
      "2767\n",
      "2766\n",
      "2765\n",
      "2764\n",
      "2763\n",
      "2762\n",
      "2761\n",
      "2760\n",
      "2759\n",
      "2758\n",
      "2757\n",
      "2756\n",
      "2755\n",
      "2754\n",
      "2753\n",
      "2752\n",
      "2751\n",
      "2750\n",
      "2749\n",
      "2748\n",
      "2747\n",
      "2746\n",
      "2745\n",
      "2744\n",
      "2743\n",
      "2742\n",
      "2741\n",
      "2740\n",
      "2739\n",
      "2738\n",
      "2737\n",
      "2736\n",
      "2735\n",
      "2734\n",
      "2733\n",
      "2732\n",
      "2731\n",
      "2730\n",
      "2729\n",
      "2728\n",
      "2727\n",
      "2726\n",
      "2725\n",
      "2724\n",
      "2723\n",
      "2722\n",
      "2721\n",
      "2720\n",
      "2719\n",
      "2718\n",
      "2717\n",
      "2716\n",
      "2715\n",
      "2714\n",
      "2713\n",
      "2712\n",
      "2711\n",
      "2710\n",
      "2709\n",
      "2708\n",
      "2707\n",
      "2706\n",
      "2705\n",
      "2704\n",
      "2703\n",
      "2702\n",
      "2701\n",
      "2700\n",
      "2699\n",
      "2698\n",
      "2697\n",
      "2696\n",
      "2695\n",
      "2694\n",
      "2693\n",
      "2692\n",
      "2691\n",
      "2690\n",
      "2689\n",
      "2688\n",
      "2687\n",
      "2686\n",
      "2685\n",
      "2684\n",
      "2683\n",
      "2682\n",
      "2681\n",
      "2680\n",
      "2679\n",
      "2678\n",
      "2677\n",
      "2676\n",
      "2675\n",
      "2674\n",
      "2673\n",
      "2672\n",
      "2671\n",
      "2670\n",
      "2669\n",
      "2668\n",
      "2667\n",
      "2666\n",
      "2665\n",
      "2664\n",
      "2663\n",
      "2662\n",
      "2661\n",
      "2660\n",
      "2659\n",
      "2658\n",
      "2657\n",
      "2656\n",
      "2655\n",
      "2654\n",
      "2653\n",
      "2652\n",
      "2651\n",
      "2650\n",
      "2649\n",
      "2648\n",
      "2647\n",
      "2646\n",
      "2645\n",
      "2644\n",
      "2643\n",
      "2642\n",
      "2641\n",
      "2640\n",
      "2639\n",
      "2638\n",
      "2637\n",
      "2636\n",
      "2635\n",
      "2634\n",
      "2633\n",
      "2632\n",
      "2631\n",
      "2630\n",
      "2629\n",
      "2628\n",
      "2627\n",
      "2626\n",
      "2625\n",
      "2624\n",
      "2623\n",
      "2622\n",
      "2621\n",
      "2620\n",
      "2619\n",
      "2618\n",
      "2617\n",
      "2616\n",
      "2615\n",
      "2614\n",
      "2613\n",
      "2612\n",
      "2611\n",
      "2610\n",
      "2609\n",
      "2608\n",
      "2607\n",
      "2606\n",
      "2605\n",
      "2604\n",
      "2603\n",
      "2602\n",
      "2601\n",
      "2600\n",
      "2599\n",
      "2598\n",
      "2597\n",
      "2596\n",
      "2595\n",
      "2594\n",
      "2593\n",
      "2592\n",
      "2591\n",
      "2590\n",
      "2589\n",
      "2588\n",
      "2587\n",
      "2586\n",
      "2585\n",
      "2584\n",
      "2583\n",
      "2582\n",
      "2581\n",
      "2580\n",
      "2579\n",
      "2578\n",
      "2577\n",
      "2576\n",
      "2575\n",
      "2574\n",
      "2573\n",
      "2572\n",
      "2571\n",
      "2570\n",
      "2569\n",
      "2568\n",
      "2567\n",
      "2566\n",
      "2565\n",
      "2564\n",
      "2563\n",
      "2562\n",
      "2561\n",
      "2560\n",
      "2559\n",
      "2558\n",
      "2557\n",
      "2556\n",
      "2555\n",
      "2554\n",
      "2553\n",
      "2552\n",
      "2551\n",
      "2550\n",
      "2549\n",
      "2548\n",
      "2547\n",
      "2546\n",
      "2545\n",
      "2544\n",
      "2543\n",
      "2542\n",
      "2541\n",
      "2540\n",
      "2539\n",
      "2538\n",
      "2537\n",
      "2536\n",
      "2535\n",
      "2534\n",
      "2533\n",
      "2532\n",
      "2531\n",
      "2530\n",
      "2529\n",
      "2528\n",
      "2527\n",
      "2526\n",
      "2525\n",
      "2524\n",
      "2523\n",
      "2522\n",
      "2521\n",
      "2520\n",
      "2519\n",
      "2518\n",
      "2517\n",
      "2516\n",
      "2515\n",
      "2514\n",
      "2513\n",
      "2512\n",
      "2511\n",
      "2510\n",
      "2509\n",
      "2508\n",
      "2507\n",
      "2506\n",
      "2505\n",
      "2504\n",
      "2503\n",
      "2502\n",
      "2501\n",
      "2500\n",
      "2499\n",
      "2498\n",
      "2497\n",
      "2496\n",
      "2495\n",
      "2494\n",
      "2493\n",
      "2492\n",
      "2491\n",
      "2490\n",
      "2489\n",
      "2488\n",
      "2487\n",
      "2486\n",
      "2485\n",
      "2484\n",
      "2483\n",
      "2482\n",
      "2481\n",
      "2480\n",
      "2479\n",
      "2478\n",
      "2477\n",
      "2476\n",
      "2475\n",
      "2474\n",
      "2473\n",
      "2472\n",
      "2471\n",
      "2470\n",
      "2469\n",
      "2468\n",
      "2467\n",
      "2466\n",
      "2465\n",
      "2464\n",
      "2463\n",
      "2462\n",
      "2461\n",
      "2460\n",
      "2459\n",
      "2458\n",
      "2457\n",
      "2456\n",
      "2455\n",
      "2454\n",
      "2453\n",
      "2452\n",
      "2451\n",
      "2450\n",
      "2449\n",
      "2448\n",
      "2447\n",
      "2446\n",
      "2445\n",
      "2444\n",
      "2443\n",
      "2442\n",
      "2441\n",
      "2440\n",
      "2439\n",
      "2438\n",
      "2437\n",
      "2436\n",
      "2435\n",
      "2434\n",
      "2433\n",
      "2432\n",
      "2431\n",
      "2430\n",
      "2429\n",
      "2428\n",
      "2427\n",
      "2426\n",
      "2425\n",
      "2424\n",
      "2423\n",
      "2422\n",
      "2421\n",
      "2420\n",
      "2419\n",
      "2418\n",
      "2417\n",
      "2416\n",
      "2415\n",
      "2414\n",
      "2413\n",
      "2412\n",
      "2411\n",
      "2410\n",
      "2409\n",
      "2408\n",
      "2407\n",
      "2406\n",
      "2405\n",
      "2404\n",
      "2403\n",
      "2402\n",
      "2401\n",
      "2400\n",
      "2399\n",
      "2398\n",
      "2397\n",
      "2396\n",
      "2395\n",
      "2394\n",
      "2393\n",
      "2392\n",
      "2391\n",
      "2390\n",
      "2389\n",
      "2388\n",
      "2387\n",
      "2386\n",
      "2385\n",
      "2384\n",
      "2383\n",
      "2382\n",
      "2381\n",
      "2380\n",
      "2379\n",
      "2378\n",
      "2377\n",
      "2376\n",
      "2375\n",
      "2374\n",
      "2373\n",
      "2372\n",
      "2371\n",
      "2370\n",
      "2369\n",
      "2368\n",
      "2367\n",
      "2366\n",
      "2365\n",
      "2364\n",
      "2363\n",
      "2362\n",
      "2361\n",
      "2360\n",
      "2359\n",
      "2358\n",
      "2357\n",
      "2356\n",
      "2355\n",
      "2354\n",
      "2353\n",
      "2352\n",
      "2351\n",
      "2350\n",
      "2349\n",
      "2348\n",
      "2347\n",
      "2346\n",
      "2345\n",
      "2344\n",
      "2343\n",
      "2342\n",
      "2341\n",
      "2340\n",
      "2339\n",
      "2338\n",
      "2337\n",
      "2336\n",
      "2335\n",
      "2334\n",
      "2333\n",
      "2332\n",
      "2331\n",
      "2330\n",
      "2329\n",
      "2328\n",
      "2327\n",
      "2326\n",
      "2325\n",
      "2324\n",
      "2323\n",
      "2322\n",
      "2321\n",
      "2320\n",
      "2319\n",
      "2318\n",
      "2317\n",
      "2316\n",
      "2315\n",
      "2314\n",
      "2313\n",
      "2312\n",
      "2311\n",
      "2310\n",
      "2309\n",
      "2308\n",
      "2307\n",
      "2306\n",
      "2305\n",
      "2304\n",
      "2303\n",
      "2302\n",
      "2301\n",
      "2300\n",
      "2299\n",
      "2298\n",
      "2297\n",
      "2296\n",
      "2295\n",
      "2294\n",
      "2293\n",
      "2292\n",
      "2291\n",
      "2290\n",
      "2289\n",
      "2288\n",
      "2287\n",
      "2286\n",
      "2285\n",
      "2284\n",
      "2283\n",
      "2282\n",
      "2281\n",
      "2280\n",
      "2279\n",
      "2278\n",
      "2277\n",
      "2276\n",
      "2275\n",
      "2274\n",
      "2273\n",
      "2272\n",
      "2271\n",
      "2270\n",
      "2269\n",
      "2268\n",
      "2267\n",
      "2266\n",
      "2265\n",
      "2264\n",
      "2263\n",
      "2262\n",
      "2261\n",
      "2260\n",
      "2259\n",
      "2258\n",
      "2257\n",
      "2256\n",
      "2255\n",
      "2254\n",
      "2253\n",
      "2252\n",
      "2251\n",
      "2250\n",
      "2249\n",
      "2248\n",
      "2247\n",
      "2246\n",
      "2245\n",
      "2244\n",
      "2243\n",
      "2242\n",
      "2241\n",
      "2240\n",
      "2239\n",
      "2238\n",
      "2237\n",
      "2236\n",
      "2235\n",
      "2234\n",
      "2233\n",
      "2232\n",
      "2231\n",
      "2230\n",
      "2229\n",
      "2228\n",
      "2227\n",
      "2226\n",
      "2225\n",
      "2224\n",
      "2223\n",
      "2222\n",
      "2221\n",
      "2220\n",
      "2219\n",
      "2218\n",
      "2217\n",
      "2216\n",
      "2215\n",
      "2214\n",
      "2213\n",
      "2212\n",
      "2211\n",
      "2210\n",
      "2209\n",
      "2208\n",
      "2207\n",
      "2206\n",
      "2205\n",
      "2204\n",
      "2203\n",
      "2202\n",
      "2201\n",
      "2200\n",
      "2199\n",
      "2198\n",
      "2197\n",
      "2196\n",
      "2195\n",
      "2194\n",
      "2193\n",
      "2192\n",
      "2191\n",
      "2190\n",
      "2189\n",
      "2188\n",
      "2187\n",
      "2186\n",
      "2185\n",
      "2184\n",
      "2183\n",
      "2182\n",
      "2181\n",
      "2180\n",
      "2179\n",
      "2178\n",
      "2177\n",
      "2176\n",
      "2175\n",
      "2174\n",
      "2173\n",
      "2172\n",
      "2171\n",
      "2170\n",
      "2169\n",
      "2168\n",
      "2167\n",
      "2166\n",
      "2165\n",
      "2164\n",
      "2163\n",
      "2162\n",
      "2161\n",
      "2160\n",
      "2159\n",
      "2158\n",
      "2157\n",
      "2156\n",
      "2155\n",
      "2154\n",
      "2153\n",
      "2152\n",
      "2151\n",
      "2150\n",
      "2149\n",
      "2148\n",
      "2147\n",
      "2146\n",
      "2145\n",
      "2144\n",
      "2143\n",
      "2142\n",
      "2141\n",
      "2140\n",
      "2139\n",
      "2138\n",
      "2137\n",
      "2136\n",
      "2135\n",
      "2134\n",
      "2133\n",
      "2132\n",
      "2131\n",
      "2130\n",
      "2129\n",
      "2128\n",
      "2127\n",
      "2126\n",
      "2125\n",
      "2124\n",
      "2123\n",
      "2122\n",
      "2121\n",
      "2120\n",
      "2119\n",
      "2118\n",
      "2117\n",
      "2116\n",
      "2115\n",
      "2114\n",
      "2113\n",
      "2112\n",
      "2111\n",
      "2110\n",
      "2109\n",
      "2108\n",
      "2107\n",
      "2106\n",
      "2105\n",
      "2104\n",
      "2103\n",
      "2102\n",
      "2101\n",
      "2100\n",
      "2099\n",
      "2098\n",
      "2097\n",
      "2096\n",
      "2095\n",
      "2094\n",
      "2093\n",
      "2092\n",
      "2091\n",
      "2090\n",
      "2089\n",
      "2088\n",
      "2087\n",
      "2086\n",
      "2085\n",
      "2084\n",
      "2083\n",
      "2082\n",
      "2081\n",
      "2080\n",
      "2079\n",
      "2078\n",
      "2077\n",
      "2076\n",
      "2075\n",
      "2074\n",
      "2073\n",
      "2072\n",
      "2071\n",
      "2070\n",
      "2069\n",
      "2068\n",
      "2067\n",
      "2066\n",
      "2065\n",
      "2064\n",
      "2063\n",
      "2062\n",
      "2061\n",
      "2060\n",
      "2059\n",
      "2058\n",
      "2057\n",
      "2056\n",
      "2055\n",
      "2054\n",
      "2053\n",
      "2052\n",
      "2051\n",
      "2050\n",
      "2049\n",
      "2048\n",
      "2047\n",
      "2046\n",
      "2045\n",
      "2044\n",
      "2043\n",
      "2042\n",
      "2041\n",
      "2040\n",
      "2039\n",
      "2038\n",
      "2037\n",
      "2036\n",
      "2035\n",
      "2034\n",
      "2033\n",
      "2032\n",
      "2031\n",
      "2030\n",
      "2029\n",
      "2028\n",
      "2027\n",
      "2026\n",
      "2025\n",
      "2024\n",
      "2023\n",
      "2022\n",
      "2021\n",
      "2020\n",
      "2019\n",
      "2018\n",
      "2017\n",
      "2016\n",
      "2015\n",
      "2014\n",
      "2013\n",
      "2012\n",
      "2011\n",
      "2010\n",
      "2009\n",
      "2008\n",
      "2007\n",
      "2006\n",
      "2005\n",
      "2004\n",
      "2003\n",
      "2002\n",
      "2001\n",
      "2000\n",
      "1999\n",
      "1998\n",
      "1997\n",
      "1996\n",
      "1995\n",
      "1994\n",
      "1993\n",
      "1992\n",
      "1991\n",
      "1990\n",
      "1989\n",
      "1988\n",
      "1987\n",
      "1986\n",
      "1985\n",
      "1984\n",
      "1983\n",
      "1982\n",
      "1981\n",
      "1980\n",
      "1979\n",
      "1978\n",
      "1977\n",
      "1976\n",
      "1975\n",
      "1974\n",
      "1973\n",
      "1972\n",
      "1971\n",
      "1970\n",
      "1969\n",
      "1968\n",
      "1967\n",
      "1966\n",
      "1965\n",
      "1964\n",
      "1963\n",
      "1962\n",
      "1961\n",
      "1960\n",
      "1959\n",
      "1958\n",
      "1957\n",
      "1956\n",
      "1955\n",
      "1954\n",
      "1953\n",
      "1952\n",
      "1951\n",
      "1950\n",
      "1949\n",
      "1948\n",
      "1947\n",
      "1946\n",
      "1945\n",
      "1944\n",
      "1943\n",
      "1942\n",
      "1941\n",
      "1940\n",
      "1939\n",
      "1938\n",
      "1937\n",
      "1936\n",
      "1935\n",
      "1934\n",
      "1933\n",
      "1932\n",
      "1931\n",
      "1930\n",
      "1929\n",
      "1928\n",
      "1927\n",
      "1926\n",
      "1925\n",
      "1924\n",
      "1923\n",
      "1922\n",
      "1921\n",
      "1920\n",
      "1919\n",
      "1918\n",
      "1917\n",
      "1916\n",
      "1915\n",
      "1914\n",
      "1913\n",
      "1912\n",
      "1911\n",
      "1910\n",
      "1909\n",
      "1908\n",
      "1907\n",
      "1906\n",
      "1905\n",
      "1904\n",
      "1903\n",
      "1902\n",
      "1901\n",
      "1900\n",
      "1899\n",
      "1898\n",
      "1897\n",
      "1896\n",
      "1895\n",
      "1894\n",
      "1893\n",
      "1892\n",
      "1891\n",
      "1890\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1889\n",
      "1888\n",
      "1887\n",
      "1886\n",
      "1885\n",
      "1884\n",
      "1883\n",
      "1882\n",
      "1881\n",
      "1880\n",
      "1879\n",
      "1878\n",
      "1877\n",
      "1876\n",
      "1875\n",
      "1874\n",
      "1873\n",
      "1872\n",
      "1871\n",
      "1870\n",
      "1869\n",
      "1868\n",
      "1867\n",
      "1866\n",
      "1865\n",
      "1864\n",
      "1863\n",
      "1862\n",
      "1861\n",
      "1860\n",
      "1859\n",
      "1858\n",
      "1857\n",
      "1856\n",
      "1855\n",
      "1854\n",
      "1853\n",
      "1852\n",
      "1851\n",
      "1850\n",
      "1849\n",
      "1848\n",
      "1847\n",
      "1846\n",
      "1845\n",
      "1844\n",
      "1843\n",
      "1842\n",
      "1841\n",
      "1840\n",
      "1839\n",
      "1838\n",
      "1837\n",
      "1836\n",
      "1835\n",
      "1834\n",
      "1833\n",
      "1832\n",
      "1831\n",
      "1830\n",
      "1829\n",
      "1828\n",
      "1827\n",
      "1826\n",
      "1825\n",
      "1824\n",
      "1823\n",
      "1822\n",
      "1821\n",
      "1820\n",
      "1819\n",
      "1818\n",
      "1817\n",
      "1816\n",
      "1815\n",
      "1814\n",
      "1813\n",
      "1812\n",
      "1811\n",
      "1810\n",
      "1809\n",
      "1808\n",
      "1807\n",
      "1806\n",
      "1805\n",
      "1804\n",
      "1803\n",
      "1802\n",
      "1801\n",
      "1800\n",
      "1799\n",
      "1798\n",
      "1797\n",
      "1796\n",
      "1795\n",
      "1794\n",
      "1793\n",
      "1792\n",
      "1791\n",
      "1790\n",
      "1789\n",
      "1788\n",
      "1787\n",
      "1786\n",
      "1785\n",
      "1784\n",
      "1783\n",
      "1782\n",
      "1781\n",
      "1780\n",
      "1779\n",
      "1778\n",
      "1777\n",
      "1776\n",
      "1775\n",
      "1774\n",
      "1773\n",
      "1772\n",
      "1771\n",
      "1770\n",
      "1769\n",
      "1768\n",
      "1767\n",
      "1766\n",
      "1765\n",
      "1764\n",
      "1763\n",
      "1762\n",
      "1761\n",
      "1760\n",
      "1759\n",
      "1758\n",
      "1757\n",
      "1756\n",
      "1755\n",
      "1754\n",
      "1753\n",
      "1752\n",
      "1751\n",
      "1750\n",
      "1749\n",
      "1748\n",
      "1747\n",
      "1746\n",
      "1745\n",
      "1744\n",
      "1743\n",
      "1742\n",
      "1741\n",
      "1740\n",
      "1739\n",
      "1738\n",
      "1737\n",
      "1736\n",
      "1735\n",
      "1734\n",
      "1733\n",
      "1732\n",
      "1731\n",
      "1730\n",
      "1729\n",
      "1728\n",
      "1727\n",
      "1726\n",
      "1725\n",
      "1724\n",
      "1723\n",
      "1722\n",
      "1721\n",
      "1720\n",
      "1719\n",
      "1718\n",
      "1717\n",
      "1716\n",
      "1715\n",
      "1714\n",
      "1713\n",
      "1712\n",
      "1711\n",
      "1710\n",
      "1709\n",
      "1708\n",
      "1707\n",
      "1706\n",
      "1705\n",
      "1704\n",
      "1703\n",
      "1702\n",
      "1701\n",
      "1700\n",
      "1699\n",
      "1698\n",
      "1697\n",
      "1696\n",
      "1695\n",
      "1694\n",
      "1693\n",
      "1692\n",
      "1691\n",
      "1690\n",
      "1689\n",
      "1688\n",
      "1687\n",
      "1686\n",
      "1685\n",
      "1684\n",
      "1683\n",
      "1682\n",
      "1681\n",
      "1680\n",
      "1679\n",
      "1678\n",
      "1677\n",
      "1676\n",
      "1675\n",
      "1674\n",
      "1673\n",
      "1672\n",
      "1671\n",
      "1670\n",
      "1669\n",
      "1668\n",
      "1667\n",
      "1666\n",
      "1665\n",
      "1664\n",
      "1663\n",
      "1662\n",
      "1661\n",
      "1660\n",
      "1659\n",
      "1658\n",
      "1657\n",
      "1656\n",
      "1655\n",
      "1654\n",
      "1653\n",
      "1652\n",
      "1651\n",
      "1650\n",
      "1649\n",
      "1648\n",
      "1647\n",
      "1646\n",
      "1645\n",
      "1644\n",
      "1643\n",
      "1642\n",
      "1641\n",
      "1640\n",
      "1639\n",
      "1638\n",
      "1637\n",
      "1636\n",
      "1635\n",
      "1634\n",
      "1633\n",
      "1632\n",
      "1631\n",
      "1630\n",
      "1629\n",
      "1628\n",
      "1627\n",
      "1626\n",
      "1625\n",
      "1624\n",
      "1623\n",
      "1622\n",
      "1621\n",
      "1620\n",
      "1619\n",
      "1618\n",
      "1617\n",
      "1616\n",
      "1615\n",
      "1614\n",
      "1613\n",
      "1612\n",
      "1611\n",
      "1610\n",
      "1609\n",
      "1608\n",
      "1607\n",
      "1606\n",
      "1605\n",
      "1604\n",
      "1603\n",
      "1602\n",
      "1601\n",
      "1600\n",
      "1599\n",
      "1598\n",
      "1597\n",
      "1596\n",
      "1595\n",
      "1594\n",
      "1593\n",
      "1592\n",
      "1591\n",
      "1590\n",
      "1589\n",
      "1588\n",
      "1587\n",
      "1586\n",
      "1585\n",
      "1584\n",
      "1583\n",
      "1582\n",
      "1581\n",
      "1580\n",
      "1579\n",
      "1578\n",
      "1577\n",
      "1576\n",
      "1575\n",
      "1574\n",
      "1573\n",
      "1572\n",
      "1571\n",
      "1570\n",
      "1569\n",
      "1568\n",
      "1567\n",
      "1566\n",
      "1565\n",
      "1564\n",
      "1563\n",
      "1562\n",
      "1561\n",
      "1560\n",
      "1559\n",
      "1558\n",
      "1557\n",
      "1556\n",
      "1555\n",
      "1554\n",
      "1553\n",
      "1552\n",
      "1551\n",
      "1550\n",
      "1549\n",
      "1548\n",
      "1547\n",
      "1546\n",
      "1545\n",
      "1544\n",
      "1543\n",
      "1542\n",
      "1541\n",
      "1540\n",
      "1539\n",
      "1538\n",
      "1537\n",
      "1536\n",
      "1535\n",
      "1534\n",
      "1533\n",
      "1532\n",
      "1531\n",
      "1530\n",
      "1529\n",
      "1528\n",
      "1527\n",
      "1526\n",
      "1525\n",
      "1524\n",
      "1523\n",
      "1522\n",
      "1521\n",
      "1520\n",
      "1519\n",
      "1518\n",
      "1517\n",
      "1516\n",
      "1515\n",
      "1514\n",
      "1513\n",
      "1512\n",
      "1511\n",
      "1510\n",
      "1509\n",
      "1508\n",
      "1507\n",
      "1506\n",
      "1505\n",
      "1504\n",
      "1503\n",
      "1502\n",
      "1501\n",
      "1500\n",
      "1499\n",
      "1498\n",
      "1497\n",
      "1496\n",
      "1495\n",
      "1494\n",
      "1493\n",
      "1492\n",
      "1491\n",
      "1490\n",
      "1489\n",
      "1488\n",
      "1487\n",
      "1486\n",
      "1485\n",
      "1484\n",
      "1483\n",
      "1482\n",
      "1481\n",
      "1480\n",
      "1479\n",
      "1478\n",
      "1477\n",
      "1476\n",
      "1475\n",
      "1474\n",
      "1473\n",
      "1472\n",
      "1471\n",
      "1470\n",
      "1469\n",
      "1468\n",
      "1467\n",
      "1466\n",
      "1465\n",
      "1464\n",
      "1463\n",
      "1462\n",
      "1461\n",
      "1460\n",
      "1459\n",
      "1458\n",
      "1457\n",
      "1456\n",
      "1455\n",
      "1454\n",
      "1453\n",
      "1452\n",
      "1451\n",
      "1450\n",
      "1449\n",
      "1448\n",
      "1447\n",
      "1446\n",
      "1445\n",
      "1444\n",
      "1443\n",
      "1442\n",
      "1441\n",
      "1440\n",
      "1439\n",
      "1438\n",
      "1437\n",
      "1436\n",
      "1435\n",
      "1434\n",
      "1433\n",
      "1432\n",
      "1431\n",
      "1430\n",
      "1429\n",
      "1428\n",
      "1427\n",
      "1426\n",
      "1425\n",
      "1424\n",
      "1423\n",
      "1422\n",
      "1421\n",
      "1420\n",
      "1419\n",
      "1418\n",
      "1417\n",
      "1416\n",
      "1415\n",
      "1414\n",
      "1413\n",
      "1412\n",
      "1411\n",
      "1410\n",
      "1409\n",
      "1408\n",
      "1407\n",
      "1406\n",
      "1405\n",
      "1404\n",
      "1403\n",
      "1402\n",
      "1401\n",
      "1400\n",
      "1399\n",
      "1398\n",
      "1397\n",
      "1396\n",
      "1395\n",
      "1394\n",
      "1393\n",
      "1392\n",
      "1391\n",
      "1390\n",
      "1389\n",
      "1388\n",
      "1387\n",
      "1386\n",
      "1385\n",
      "1384\n",
      "1383\n",
      "1382\n",
      "1381\n",
      "1380\n",
      "1379\n",
      "1378\n",
      "1377\n",
      "1376\n",
      "1375\n",
      "1374\n",
      "1373\n",
      "1372\n",
      "1371\n",
      "1370\n",
      "1369\n",
      "1368\n",
      "1367\n",
      "1366\n",
      "1365\n",
      "1364\n",
      "1363\n",
      "1362\n",
      "1361\n",
      "1360\n",
      "1359\n",
      "1358\n",
      "1357\n",
      "1356\n",
      "1355\n",
      "1354\n",
      "1353\n",
      "1352\n",
      "1351\n",
      "1350\n",
      "1349\n",
      "1348\n",
      "1347\n",
      "1346\n",
      "1345\n",
      "1344\n",
      "1343\n",
      "1342\n",
      "1341\n",
      "1340\n",
      "1339\n",
      "1338\n",
      "1337\n",
      "1336\n",
      "1335\n",
      "1334\n",
      "1333\n",
      "1332\n",
      "1331\n",
      "1330\n",
      "1329\n",
      "1328\n",
      "1327\n",
      "1326\n",
      "1325\n",
      "1324\n",
      "1323\n",
      "1322\n",
      "1321\n",
      "1320\n",
      "1319\n",
      "1318\n",
      "1317\n",
      "1316\n",
      "1315\n",
      "1314\n",
      "1313\n",
      "1312\n",
      "1311\n",
      "1310\n",
      "1309\n",
      "1308\n",
      "1307\n",
      "1306\n",
      "1305\n",
      "1304\n",
      "1303\n",
      "1302\n",
      "1301\n",
      "1300\n",
      "1299\n",
      "1298\n",
      "1297\n",
      "1296\n",
      "1295\n",
      "1294\n",
      "1293\n",
      "1292\n",
      "1291\n",
      "1290\n",
      "1289\n",
      "1288\n",
      "1287\n",
      "1286\n",
      "1285\n",
      "1284\n",
      "1283\n",
      "1282\n",
      "1281\n",
      "1280\n",
      "1279\n",
      "1278\n",
      "1277\n",
      "1276\n",
      "1275\n",
      "1274\n",
      "1273\n",
      "1272\n",
      "1271\n",
      "1270\n",
      "1269\n",
      "1268\n",
      "1267\n",
      "1266\n",
      "1265\n",
      "1264\n",
      "1263\n",
      "1262\n",
      "1261\n",
      "1260\n",
      "1259\n",
      "1258\n",
      "1257\n",
      "1256\n",
      "1255\n",
      "1254\n",
      "1253\n",
      "1252\n",
      "1251\n",
      "1250\n",
      "1249\n",
      "1248\n",
      "1247\n",
      "1246\n",
      "1245\n",
      "1244\n",
      "1243\n",
      "1242\n",
      "1241\n",
      "1240\n",
      "1239\n",
      "1238\n",
      "1237\n",
      "1236\n",
      "1235\n",
      "1234\n",
      "1233\n",
      "1232\n",
      "1231\n",
      "1230\n",
      "1229\n",
      "1228\n",
      "1227\n",
      "1226\n",
      "1225\n",
      "1224\n",
      "1223\n",
      "1222\n",
      "1221\n",
      "1220\n",
      "1219\n",
      "1218\n",
      "1217\n",
      "1216\n",
      "1215\n",
      "1214\n",
      "1213\n",
      "1212\n",
      "1211\n",
      "1210\n",
      "1209\n",
      "1208\n",
      "1207\n",
      "1206\n",
      "1205\n",
      "1204\n",
      "1203\n",
      "1202\n",
      "1201\n",
      "1200\n",
      "1199\n",
      "1198\n",
      "1197\n",
      "1196\n",
      "1195\n",
      "1194\n",
      "1193\n",
      "1192\n",
      "1191\n",
      "1190\n",
      "1189\n",
      "1188\n",
      "1187\n",
      "1186\n",
      "1185\n",
      "1184\n",
      "1183\n",
      "1182\n",
      "1181\n",
      "1180\n",
      "1179\n",
      "1178\n",
      "1177\n",
      "1176\n",
      "1175\n",
      "1174\n",
      "1173\n",
      "1172\n",
      "1171\n",
      "1170\n",
      "1169\n",
      "1168\n",
      "1167\n",
      "1166\n",
      "1165\n",
      "1164\n",
      "1163\n",
      "1162\n",
      "1161\n",
      "1160\n",
      "1159\n",
      "1158\n",
      "1157\n",
      "1156\n",
      "1155\n",
      "1154\n",
      "1153\n",
      "1152\n",
      "1151\n",
      "1150\n",
      "1149\n",
      "1148\n",
      "1147\n",
      "1146\n",
      "1145\n",
      "1144\n",
      "1143\n",
      "1142\n",
      "1141\n",
      "1140\n",
      "1139\n",
      "1138\n",
      "1137\n",
      "1136\n",
      "1135\n",
      "1134\n",
      "1133\n",
      "1132\n",
      "1131\n",
      "1130\n",
      "1129\n",
      "1128\n",
      "1127\n",
      "1126\n",
      "1125\n",
      "1124\n",
      "1123\n",
      "1122\n",
      "1121\n",
      "1120\n",
      "1119\n",
      "1118\n",
      "1117\n",
      "1116\n",
      "1115\n",
      "1114\n",
      "1113\n",
      "1112\n",
      "1111\n",
      "1110\n",
      "1109\n",
      "1108\n",
      "1107\n",
      "1106\n",
      "1105\n",
      "1104\n",
      "1103\n",
      "1102\n",
      "1101\n",
      "1100\n",
      "1099\n",
      "1098\n",
      "1097\n",
      "1096\n",
      "1095\n",
      "1094\n",
      "1093\n",
      "1092\n",
      "1091\n",
      "1090\n",
      "1089\n",
      "1088\n",
      "1087\n",
      "1086\n",
      "1085\n",
      "1084\n",
      "1083\n",
      "1082\n",
      "1081\n",
      "1080\n",
      "1079\n",
      "1078\n",
      "1077\n",
      "1076\n",
      "1075\n",
      "1074\n",
      "1073\n",
      "1072\n",
      "1071\n",
      "1070\n",
      "1069\n",
      "1068\n",
      "1067\n",
      "1066\n",
      "1065\n",
      "1064\n",
      "1063\n",
      "1062\n",
      "1061\n",
      "1060\n",
      "1059\n",
      "1058\n",
      "1057\n",
      "1056\n",
      "1055\n",
      "1054\n",
      "1053\n",
      "1052\n",
      "1051\n",
      "1050\n",
      "1049\n",
      "1048\n",
      "1047\n",
      "1046\n",
      "1045\n",
      "1044\n",
      "1043\n",
      "1042\n",
      "1041\n",
      "1040\n",
      "1039\n",
      "1038\n",
      "1037\n",
      "1036\n",
      "1035\n",
      "1034\n",
      "1033\n",
      "1032\n",
      "1031\n",
      "1030\n",
      "1029\n",
      "1028\n",
      "1027\n",
      "1026\n",
      "1025\n",
      "1024\n",
      "1023\n",
      "1022\n",
      "1021\n",
      "1020\n",
      "1019\n",
      "1018\n",
      "1017\n",
      "1016\n",
      "1015\n",
      "1014\n",
      "1013\n",
      "1012\n",
      "1011\n",
      "1010\n",
      "1009\n",
      "1008\n",
      "1007\n",
      "1006\n",
      "1005\n",
      "1004\n",
      "1003\n",
      "1002\n",
      "1001\n",
      "1000\n",
      "999\n",
      "998\n",
      "997\n",
      "996\n",
      "995\n",
      "994\n",
      "993\n",
      "992\n",
      "991\n",
      "990\n",
      "989\n",
      "988\n",
      "987\n",
      "986\n",
      "985\n",
      "984\n",
      "983\n",
      "982\n",
      "981\n",
      "980\n",
      "979\n",
      "978\n",
      "977\n",
      "976\n",
      "975\n",
      "974\n",
      "973\n",
      "972\n",
      "971\n",
      "970\n",
      "969\n",
      "968\n",
      "967\n",
      "966\n",
      "965\n",
      "964\n",
      "963\n",
      "962\n",
      "961\n",
      "960\n",
      "959\n",
      "958\n",
      "957\n",
      "956\n",
      "955\n",
      "954\n",
      "953\n",
      "952\n",
      "951\n",
      "950\n",
      "949\n",
      "948\n",
      "947\n",
      "946\n",
      "945\n",
      "944\n",
      "943\n",
      "942\n",
      "941\n",
      "940\n",
      "939\n",
      "938\n",
      "937\n",
      "936\n",
      "935\n",
      "934\n",
      "933\n",
      "932\n",
      "931\n",
      "930\n",
      "929\n",
      "928\n",
      "927\n",
      "926\n",
      "925\n",
      "924\n",
      "923\n",
      "922\n",
      "921\n",
      "920\n",
      "919\n",
      "918\n",
      "917\n",
      "916\n",
      "915\n",
      "914\n",
      "913\n",
      "912\n",
      "911\n",
      "910\n",
      "909\n",
      "908\n",
      "907\n",
      "906\n",
      "905\n",
      "904\n",
      "903\n",
      "902\n",
      "901\n",
      "900\n",
      "899\n",
      "898\n",
      "897\n",
      "896\n",
      "895\n",
      "894\n",
      "893\n",
      "892\n",
      "891\n",
      "890\n",
      "889\n",
      "888\n",
      "887\n",
      "886\n",
      "885\n",
      "884\n",
      "883\n",
      "882\n",
      "881\n",
      "880\n",
      "879\n",
      "878\n",
      "877\n",
      "876\n",
      "875\n",
      "874\n",
      "873\n",
      "872\n",
      "871\n",
      "870\n",
      "869\n",
      "868\n",
      "867\n",
      "866\n",
      "865\n",
      "864\n",
      "863\n",
      "862\n",
      "861\n",
      "860\n",
      "859\n",
      "858\n",
      "857\n",
      "856\n",
      "855\n",
      "854\n",
      "853\n",
      "852\n",
      "851\n",
      "850\n",
      "849\n",
      "848\n",
      "847\n",
      "846\n",
      "845\n",
      "844\n",
      "843\n",
      "842\n",
      "841\n",
      "840\n",
      "839\n",
      "838\n",
      "837\n",
      "836\n",
      "835\n",
      "834\n",
      "833\n",
      "832\n",
      "831\n",
      "830\n",
      "829\n",
      "828\n",
      "827\n",
      "826\n",
      "825\n",
      "824\n",
      "823\n",
      "822\n",
      "821\n",
      "820\n",
      "819\n",
      "818\n",
      "817\n",
      "816\n",
      "815\n",
      "814\n",
      "813\n",
      "812\n",
      "811\n",
      "810\n",
      "809\n",
      "808\n",
      "807\n",
      "806\n",
      "805\n",
      "804\n",
      "803\n",
      "802\n",
      "801\n",
      "800\n",
      "799\n",
      "798\n",
      "797\n",
      "796\n",
      "795\n",
      "794\n",
      "793\n",
      "792\n",
      "791\n",
      "790\n",
      "789\n",
      "788\n",
      "787\n",
      "786\n",
      "785\n",
      "784\n",
      "783\n",
      "782\n",
      "781\n",
      "780\n",
      "779\n",
      "778\n",
      "777\n",
      "776\n",
      "775\n",
      "774\n",
      "773\n",
      "772\n",
      "771\n",
      "770\n",
      "769\n",
      "768\n",
      "767\n",
      "766\n",
      "765\n",
      "764\n",
      "763\n",
      "762\n",
      "761\n",
      "760\n",
      "759\n",
      "758\n",
      "757\n",
      "756\n",
      "755\n",
      "754\n",
      "753\n",
      "752\n",
      "751\n",
      "750\n",
      "749\n",
      "748\n",
      "747\n",
      "746\n",
      "745\n",
      "744\n",
      "743\n",
      "742\n",
      "741\n",
      "740\n",
      "739\n",
      "738\n",
      "737\n",
      "736\n",
      "735\n",
      "734\n",
      "733\n",
      "732\n",
      "731\n",
      "730\n",
      "729\n",
      "728\n",
      "727\n",
      "726\n",
      "725\n",
      "724\n",
      "723\n",
      "722\n",
      "721\n",
      "720\n",
      "719\n",
      "718\n",
      "717\n",
      "716\n",
      "715\n",
      "714\n",
      "713\n",
      "712\n",
      "711\n",
      "710\n",
      "709\n",
      "708\n",
      "707\n",
      "706\n",
      "705\n",
      "704\n",
      "703\n",
      "702\n",
      "701\n",
      "700\n",
      "699\n",
      "698\n",
      "697\n",
      "696\n",
      "695\n",
      "694\n",
      "693\n",
      "692\n",
      "691\n",
      "690\n",
      "689\n",
      "688\n",
      "687\n",
      "686\n",
      "685\n",
      "684\n",
      "683\n",
      "682\n",
      "681\n",
      "680\n",
      "679\n",
      "678\n",
      "677\n",
      "676\n",
      "675\n",
      "674\n",
      "673\n",
      "672\n",
      "671\n",
      "670\n",
      "669\n",
      "668\n",
      "667\n",
      "666\n",
      "665\n",
      "664\n",
      "663\n",
      "662\n",
      "661\n",
      "660\n",
      "659\n",
      "658\n",
      "657\n",
      "656\n",
      "655\n",
      "654\n",
      "653\n",
      "652\n",
      "651\n",
      "650\n",
      "649\n",
      "648\n",
      "647\n",
      "646\n",
      "645\n",
      "644\n",
      "643\n",
      "642\n",
      "641\n",
      "640\n",
      "639\n",
      "638\n",
      "637\n",
      "636\n",
      "635\n",
      "634\n",
      "633\n",
      "632\n",
      "631\n",
      "630\n",
      "629\n",
      "628\n",
      "627\n",
      "626\n",
      "625\n",
      "624\n",
      "623\n",
      "622\n",
      "621\n",
      "620\n",
      "619\n",
      "618\n",
      "617\n",
      "616\n",
      "615\n",
      "614\n",
      "613\n",
      "612\n",
      "611\n",
      "610\n",
      "609\n",
      "608\n",
      "607\n",
      "606\n",
      "605\n",
      "604\n",
      "603\n",
      "602\n",
      "601\n",
      "600\n",
      "599\n",
      "598\n",
      "597\n",
      "596\n",
      "595\n",
      "594\n",
      "593\n",
      "592\n",
      "591\n",
      "590\n",
      "589\n",
      "588\n",
      "587\n",
      "586\n",
      "585\n",
      "584\n",
      "583\n",
      "582\n",
      "581\n",
      "580\n",
      "579\n",
      "578\n",
      "577\n",
      "576\n",
      "575\n",
      "574\n",
      "573\n",
      "572\n",
      "571\n",
      "570\n",
      "569\n",
      "568\n",
      "567\n",
      "566\n",
      "565\n",
      "564\n",
      "563\n",
      "562\n",
      "561\n",
      "560\n",
      "559\n",
      "558\n",
      "557\n",
      "556\n",
      "555\n",
      "554\n",
      "553\n",
      "552\n",
      "551\n",
      "550\n",
      "549\n",
      "548\n",
      "547\n",
      "546\n",
      "545\n",
      "544\n",
      "543\n",
      "542\n",
      "541\n",
      "540\n",
      "539\n",
      "538\n",
      "537\n",
      "536\n",
      "535\n",
      "534\n",
      "533\n",
      "532\n",
      "531\n",
      "530\n",
      "529\n",
      "528\n",
      "527\n",
      "526\n",
      "525\n",
      "524\n",
      "523\n",
      "522\n",
      "521\n",
      "520\n",
      "519\n",
      "518\n",
      "517\n",
      "516\n",
      "515\n",
      "514\n",
      "513\n",
      "512\n",
      "511\n",
      "510\n",
      "509\n",
      "508\n",
      "507\n",
      "506\n",
      "505\n",
      "504\n",
      "503\n",
      "502\n",
      "501\n",
      "500\n",
      "499\n",
      "498\n",
      "497\n",
      "496\n",
      "495\n",
      "494\n",
      "493\n",
      "492\n",
      "491\n",
      "490\n",
      "489\n",
      "488\n",
      "487\n",
      "486\n",
      "485\n",
      "484\n",
      "483\n",
      "482\n",
      "481\n",
      "480\n",
      "479\n",
      "478\n",
      "477\n",
      "476\n",
      "475\n",
      "474\n",
      "473\n",
      "472\n",
      "471\n",
      "470\n",
      "469\n",
      "468\n",
      "467\n",
      "466\n",
      "465\n",
      "464\n",
      "463\n",
      "462\n",
      "461\n",
      "460\n",
      "459\n",
      "458\n",
      "457\n",
      "456\n",
      "455\n",
      "454\n",
      "453\n",
      "452\n",
      "451\n",
      "450\n",
      "449\n",
      "448\n",
      "447\n",
      "446\n",
      "445\n",
      "444\n",
      "443\n",
      "442\n",
      "441\n",
      "440\n",
      "439\n",
      "438\n",
      "437\n",
      "436\n",
      "435\n",
      "434\n",
      "433\n",
      "432\n",
      "431\n",
      "430\n",
      "429\n",
      "428\n",
      "427\n",
      "426\n",
      "425\n",
      "424\n",
      "423\n",
      "422\n",
      "421\n",
      "420\n",
      "419\n",
      "418\n",
      "417\n",
      "416\n",
      "415\n",
      "414\n",
      "413\n",
      "412\n",
      "411\n",
      "410\n",
      "409\n",
      "408\n",
      "407\n",
      "406\n",
      "405\n",
      "404\n",
      "403\n",
      "402\n",
      "401\n",
      "400\n",
      "399\n",
      "398\n",
      "397\n",
      "396\n",
      "395\n",
      "394\n",
      "393\n",
      "392\n",
      "391\n",
      "390\n",
      "389\n",
      "388\n",
      "387\n",
      "386\n",
      "385\n",
      "384\n",
      "383\n",
      "382\n",
      "381\n",
      "380\n",
      "379\n",
      "378\n",
      "377\n",
      "376\n",
      "375\n",
      "374\n",
      "373\n",
      "372\n",
      "371\n",
      "370\n",
      "369\n",
      "368\n",
      "367\n",
      "366\n",
      "365\n",
      "364\n",
      "363\n",
      "362\n",
      "361\n",
      "360\n",
      "359\n",
      "358\n",
      "357\n",
      "356\n",
      "355\n",
      "354\n",
      "353\n",
      "352\n",
      "351\n",
      "350\n",
      "349\n",
      "348\n",
      "347\n",
      "346\n",
      "345\n",
      "344\n",
      "343\n",
      "342\n",
      "341\n",
      "340\n",
      "339\n",
      "338\n",
      "337\n",
      "336\n",
      "335\n",
      "334\n",
      "333\n",
      "332\n",
      "331\n",
      "330\n",
      "329\n",
      "328\n",
      "327\n",
      "326\n",
      "325\n",
      "324\n",
      "323\n",
      "322\n",
      "321\n",
      "320\n",
      "319\n",
      "318\n",
      "317\n",
      "316\n",
      "315\n",
      "314\n",
      "313\n",
      "312\n",
      "311\n",
      "310\n",
      "309\n",
      "308\n",
      "307\n",
      "306\n",
      "305\n",
      "304\n",
      "303\n",
      "302\n",
      "301\n",
      "300\n",
      "299\n",
      "298\n",
      "297\n",
      "296\n",
      "295\n",
      "294\n",
      "293\n",
      "292\n",
      "291\n",
      "290\n",
      "289\n",
      "288\n",
      "287\n",
      "286\n",
      "285\n",
      "284\n",
      "283\n",
      "282\n",
      "281\n",
      "280\n",
      "279\n",
      "278\n",
      "277\n",
      "276\n",
      "275\n",
      "274\n",
      "273\n",
      "272\n",
      "271\n",
      "270\n",
      "269\n",
      "268\n",
      "267\n",
      "266\n",
      "265\n",
      "264\n",
      "263\n",
      "262\n",
      "261\n",
      "260\n",
      "259\n",
      "258\n",
      "257\n",
      "256\n",
      "255\n",
      "254\n",
      "253\n",
      "252\n",
      "251\n",
      "250\n",
      "249\n",
      "248\n",
      "247\n",
      "246\n",
      "245\n",
      "244\n",
      "243\n",
      "242\n",
      "241\n",
      "240\n",
      "239\n",
      "238\n",
      "237\n",
      "236\n",
      "235\n",
      "234\n",
      "233\n",
      "232\n",
      "231\n",
      "230\n",
      "229\n",
      "228\n",
      "227\n",
      "226\n",
      "225\n",
      "224\n",
      "223\n",
      "222\n",
      "221\n",
      "220\n",
      "219\n",
      "218\n",
      "217\n",
      "216\n",
      "215\n",
      "214\n",
      "213\n",
      "212\n",
      "211\n",
      "210\n",
      "209\n",
      "208\n",
      "207\n",
      "206\n",
      "205\n",
      "204\n",
      "203\n",
      "202\n",
      "201\n",
      "200\n",
      "199\n",
      "198\n",
      "197\n",
      "196\n",
      "195\n",
      "194\n",
      "193\n",
      "192\n",
      "191\n",
      "190\n",
      "189\n",
      "188\n",
      "187\n",
      "186\n",
      "185\n",
      "184\n",
      "183\n",
      "182\n",
      "181\n",
      "180\n",
      "179\n",
      "178\n",
      "177\n",
      "176\n",
      "175\n",
      "174\n",
      "173\n",
      "172\n",
      "171\n",
      "170\n",
      "169\n",
      "168\n",
      "167\n",
      "166\n",
      "165\n",
      "164\n",
      "163\n",
      "162\n",
      "161\n",
      "160\n",
      "159\n",
      "158\n",
      "157\n",
      "156\n",
      "155\n",
      "154\n",
      "153\n",
      "152\n",
      "151\n",
      "150\n",
      "149\n",
      "148\n",
      "147\n",
      "146\n",
      "145\n",
      "144\n",
      "143\n",
      "142\n",
      "141\n",
      "140\n",
      "139\n",
      "138\n",
      "137\n",
      "136\n",
      "135\n",
      "134\n",
      "133\n",
      "132\n",
      "131\n",
      "130\n",
      "129\n",
      "128\n",
      "127\n",
      "126\n",
      "125\n",
      "124\n",
      "123\n",
      "122\n",
      "121\n",
      "120\n",
      "119\n",
      "118\n",
      "117\n",
      "116\n",
      "115\n",
      "114\n",
      "113\n",
      "112\n",
      "111\n",
      "110\n",
      "109\n",
      "108\n",
      "107\n",
      "106\n",
      "105\n",
      "104\n",
      "103\n",
      "102\n",
      "101\n",
      "100\n",
      "99\n",
      "98\n",
      "97\n",
      "96\n",
      "95\n",
      "94\n",
      "93\n",
      "92\n",
      "91\n",
      "90\n",
      "89\n",
      "88\n",
      "87\n",
      "86\n",
      "85\n",
      "84\n",
      "83\n",
      "82\n",
      "81\n",
      "80\n",
      "79\n",
      "78\n",
      "77\n",
      "76\n",
      "75\n",
      "74\n",
      "73\n",
      "72\n",
      "71\n",
      "70\n",
      "69\n",
      "68\n",
      "67\n",
      "66\n",
      "65\n",
      "64\n",
      "63\n",
      "62\n",
      "61\n",
      "60\n",
      "59\n",
      "58\n",
      "57\n",
      "56\n",
      "55\n",
      "54\n",
      "53\n",
      "52\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "51\n",
      "50\n",
      "49\n",
      "48\n",
      "47\n",
      "46\n",
      "45\n",
      "44\n",
      "43\n",
      "42\n",
      "41\n",
      "40\n",
      "39\n",
      "38\n",
      "37\n",
      "36\n",
      "35\n",
      "34\n",
      "33\n",
      "32\n",
      "31\n",
      "30\n",
      "29\n",
      "28\n",
      "27\n",
      "26\n",
      "25\n",
      "24\n",
      "23\n",
      "22\n",
      "21\n",
      "20\n",
      "19\n",
      "18\n",
      "17\n",
      "16\n",
      "15\n",
      "14\n",
      "13\n",
      "12\n",
      "11\n",
      "10\n",
      "9\n",
      "8\n",
      "7\n",
      "6\n",
      "5\n",
      "4\n",
      "3\n",
      "2\n",
      "1\n",
      "0\n"
     ]
    }
   ],
   "source": [
    "emails, labels = make_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T12:44:26.493835Z",
     "start_time": "2018-08-30T12:44:26.481794Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((5172,), (5172,))"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "emails.shape, labels.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T12:44:26.624065Z",
     "start_time": "2018-08-30T12:44:26.502447Z"
    }
   },
   "outputs": [],
   "source": [
    "# making them dimension ready\n",
    "emails = emails[:, np.newaxis]\n",
    "labels = labels[:, np.newaxis]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:04:00.467535Z",
     "start_time": "2018-08-30T14:04:00.457704Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((5172, 1), (5172, 1))"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "emails.shape, labels.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:04:27.186000Z",
     "start_time": "2018-08-30T14:04:26.644471Z"
    }
   },
   "outputs": [],
   "source": [
    "df = pd.DataFrame({'message': emails.flatten(), 'labels':labels.flatten()}, index=np.arange(5172))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:04:30.380097Z",
     "start_time": "2018-08-30T14:04:30.359244Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>labels</th>\n",
       "      <th>message</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>spam</td>\n",
       "      <td>Subject: special promotion - get a las vegas v...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ham</td>\n",
       "      <td>Subject: re : enron / hpl actuals for august 1...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ham</td>\n",
       "      <td>Subject: re : enron industrial markets - super...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ham</td>\n",
       "      <td>Subject: revision # 1 - enron / hpl actuals fo...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>spam</td>\n",
       "      <td>Subject: legal operating systems for a third o...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  labels                                            message\n",
       "0   spam  Subject: special promotion - get a las vegas v...\n",
       "1    ham  Subject: re : enron / hpl actuals for august 1...\n",
       "2    ham  Subject: re : enron industrial markets - super...\n",
       "3    ham  Subject: revision # 1 - enron / hpl actuals fo...\n",
       "4   spam  Subject: legal operating systems for a third o..."
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "preprocess email data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:05:46.490804Z",
     "start_time": "2018-08-30T14:05:24.302468Z"
    }
   },
   "outputs": [],
   "source": [
    "from bs4 import BeautifulSoup\n",
    "from nltk import word_tokenize, WordNetLemmatizer\n",
    "lemmatizer = WordNetLemmatizer()\n",
    "def purify_text(message):\n",
    "    soup = BeautifulSoup(message)\n",
    "    text =  soup.get_text()\n",
    "    text = text.replace(\"\\n\", \" \").replace(\"/\", \"\").replace(\"|\", \"\").replace(\"http\", \"\").replace(':', \"\").replace('\\t', \"\").replace(\"Subject\", \"\").replace('re', \"\").strip()\n",
    "    tokens = word_tokenize(text)\n",
    "    temp = [lemmatizer.lemmatize(word.lower()) for word in tokens]\n",
    "    return ' '.join(temp)\n",
    "\n",
    "df['message'] = df['message'].apply(purify_text)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:05:46.520371Z",
     "start_time": "2018-08-30T14:05:46.510484Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>labels</th>\n",
       "      <th>message</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>spam</td>\n",
       "      <td>special promotion - get a la vega vacation for...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ham</td>\n",
       "      <td>enron hpl actuals for august 11 - 13 , 2000 vi...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ham</td>\n",
       "      <td>enron industrial market - super saturday cruit...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ham</td>\n",
       "      <td>vision # 1 - enron hpl actuals for july 5 , 20...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>spam</td>\n",
       "      <td>legal operating system for a third of the pric...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  labels                                            message\n",
       "0   spam  special promotion - get a la vega vacation for...\n",
       "1    ham  enron hpl actuals for august 11 - 13 , 2000 vi...\n",
       "2    ham  enron industrial market - super saturday cruit...\n",
       "3    ham  vision # 1 - enron hpl actuals for july 5 , 20...\n",
       "4   spam  legal operating system for a third of the pric..."
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Adding sms spam data to this dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:07:48.005993Z",
     "start_time": "2018-08-30T14:07:47.983194Z"
    }
   },
   "outputs": [],
   "source": [
    "df_sms = pd.read_csv('smsspamcollection/SMSSpamCollection', sep='\\t', names=['labels', 'message'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:07:48.777545Z",
     "start_time": "2018-08-30T14:07:48.754759Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>labels</th>\n",
       "      <th>message</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ham</td>\n",
       "      <td>Go until jurong point, crazy.. Available only ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ham</td>\n",
       "      <td>Ok lar... Joking wif u oni...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>spam</td>\n",
       "      <td>Free entry in 2 a wkly comp to win FA Cup fina...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ham</td>\n",
       "      <td>U dun say so early hor... U c already then say...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>ham</td>\n",
       "      <td>Nah I don't think he goes to usf, he lives aro...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  labels                                            message\n",
       "0    ham  Go until jurong point, crazy.. Available only ...\n",
       "1    ham                      Ok lar... Joking wif u oni...\n",
       "2   spam  Free entry in 2 a wkly comp to win FA Cup fina...\n",
       "3    ham  U dun say so early hor... U c already then say...\n",
       "4    ham  Nah I don't think he goes to usf, he lives aro..."
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_sms.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:08:29.531312Z",
     "start_time": "2018-08-30T14:08:26.144401Z"
    }
   },
   "outputs": [],
   "source": [
    "df_sms['message'] = df_sms['message'].apply(purify_text)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:08:31.178710Z",
     "start_time": "2018-08-30T14:08:31.157271Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>labels</th>\n",
       "      <th>message</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ham</td>\n",
       "      <td>go until jurong point , crazy.. available only...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ham</td>\n",
       "      <td>ok lar ... joking wif u oni ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>spam</td>\n",
       "      <td>fe entry in 2 a wkly comp to win fa cup final ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ham</td>\n",
       "      <td>u dun say so early hor ... u c alady then say ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>ham</td>\n",
       "      <td>nah i do n't think he go to usf , he life arou...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  labels                                            message\n",
       "0    ham  go until jurong point , crazy.. available only...\n",
       "1    ham                    ok lar ... joking wif u oni ...\n",
       "2   spam  fe entry in 2 a wkly comp to win fa cup final ...\n",
       "3    ham  u dun say so early hor ... u c alady then say ...\n",
       "4    ham  nah i do n't think he go to usf , he life arou..."
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_sms.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Merging the two datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:09:53.817254Z",
     "start_time": "2018-08-30T14:09:53.798425Z"
    }
   },
   "outputs": [],
   "source": [
    "df_final = df.merge(df_sms, how='outer')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:09:54.318719Z",
     "start_time": "2018-08-30T14:09:54.303988Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>labels</th>\n",
       "      <th>message</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>spam</td>\n",
       "      <td>special promotion - get a la vega vacation for...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>ham</td>\n",
       "      <td>enron hpl actuals for august 11 - 13 , 2000 vi...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>ham</td>\n",
       "      <td>enron industrial market - super saturday cruit...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>ham</td>\n",
       "      <td>vision # 1 - enron hpl actuals for july 5 , 20...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>spam</td>\n",
       "      <td>legal operating system for a third of the pric...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  labels                                            message\n",
       "0   spam  special promotion - get a la vega vacation for...\n",
       "1    ham  enron hpl actuals for august 11 - 13 , 2000 vi...\n",
       "2    ham  enron industrial market - super saturday cruit...\n",
       "3    ham  vision # 1 - enron hpl actuals for july 5 , 20...\n",
       "4   spam  legal operating system for a third of the pric..."
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_final.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:10:31.261367Z",
     "start_time": "2018-08-30T14:10:31.253009Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>labels</th>\n",
       "      <th>message</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>10739</th>\n",
       "      <td>spam</td>\n",
       "      <td>this is the 2nd time we have tried 2 contact u...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10740</th>\n",
       "      <td>ham</td>\n",
       "      <td>will ü b going to esplanade fr home ?</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10741</th>\n",
       "      <td>ham</td>\n",
       "      <td>pity , * wa in mood for that . so ... any othe...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10742</th>\n",
       "      <td>ham</td>\n",
       "      <td>the guy did some bitching but i acted like i '...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10743</th>\n",
       "      <td>ham</td>\n",
       "      <td>rofl . it true to it name</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      labels                                            message\n",
       "10739   spam  this is the 2nd time we have tried 2 contact u...\n",
       "10740    ham              will ü b going to esplanade fr home ?\n",
       "10741    ham  pity , * wa in mood for that . so ... any othe...\n",
       "10742    ham  the guy did some bitching but i acted like i '...\n",
       "10743    ham                          rofl . it true to it name"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_final.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:12:33.337792Z",
     "start_time": "2018-08-30T14:12:33.292509Z"
    }
   },
   "outputs": [],
   "source": [
    "df_final = df_final.sample(frac=1).reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:44:07.790073Z",
     "start_time": "2018-08-30T14:44:07.776827Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>labels</th>\n",
       "      <th>message</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>10739</th>\n",
       "      <td>ham</td>\n",
       "      <td>mm feeling sleepy . today itself i shall get t...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10740</th>\n",
       "      <td>ham</td>\n",
       "      <td>and also i 've sorta blown him off a couple ti...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10741</th>\n",
       "      <td>ham</td>\n",
       "      <td>duke exchange deal feb . 2000 prod . dan , i b...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10742</th>\n",
       "      <td>spam</td>\n",
       "      <td>send out five motor att client id 4551 ferring...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10743</th>\n",
       "      <td>ham</td>\n",
       "      <td>panenergy marketing march 2000 production deal...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "      labels                                            message\n",
       "10739    ham  mm feeling sleepy . today itself i shall get t...\n",
       "10740    ham  and also i 've sorta blown him off a couple ti...\n",
       "10741    ham  duke exchange deal feb . 2000 prod . dan , i b...\n",
       "10742   spam  send out five motor att client id 4551 ferring...\n",
       "10743    ham  panenergy marketing march 2000 production deal..."
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_final.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:44:21.767717Z",
     "start_time": "2018-08-30T14:44:21.761274Z"
    }
   },
   "outputs": [],
   "source": [
    "df = df_final.copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualizing the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:44:56.874748Z",
     "start_time": "2018-08-30T14:44:56.461526Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7fc928369128>"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fc940b15be0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.countplot(df['labels'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:46:01.032671Z",
     "start_time": "2018-08-30T14:46:00.780726Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7fc9261ad390>"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fc926253240>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df['length'] = df['message'].apply(len)\n",
    "sns.distplot(df['length'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:48:26.210631Z",
     "start_time": "2018-08-30T14:48:26.198009Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ham     260\n",
       "spam    190\n",
       "Name: labels, dtype: int64"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[df['length']>2500]['labels'].value_counts() #.count()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T14:48:51.604016Z",
     "start_time": "2018-08-30T14:48:51.004289Z"
    }
   },
   "outputs": [],
   "source": [
    "for index in df[df['length']>2500].index.tolist():\n",
    "    df.drop(index=index, axis=1, inplace=True)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T15:01:50.676551Z",
     "start_time": "2018-08-30T15:01:50.408179Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7fc92608e160>"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fc9260b54a8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df['length'] = df['message'].apply(len)\n",
    "sns.distplot(df['length'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T15:02:36.981087Z",
     "start_time": "2018-08-30T15:02:36.400687Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([<matplotlib.axes._subplots.AxesSubplot object at 0x7fc9261680b8>,\n",
       "       <matplotlib.axes._subplots.AxesSubplot object at 0x7fc92b2ac550>],\n",
       "      dtype=object)"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fc925ff5240>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df.hist('length', by='labels', figsize=(15, 10), bins=60)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Applying NLP; BOW, TF-IDF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T15:02:59.230524Z",
     "start_time": "2018-08-30T15:02:59.217712Z"
    }
   },
   "outputs": [],
   "source": [
    "import string\n",
    "from nltk.corpus import stopwords\n",
    "def text_process(message):\n",
    "    noPunc = [char for char in message if char not in string.punctuation]\n",
    "    noPunc = ''.join(noPunc)\n",
    "    \n",
    "    return [word for word in noPunc.split() if word not in stopwords.words('english')]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T15:04:17.619753Z",
     "start_time": "2018-08-30T15:03:15.250051Z"
    }
   },
   "outputs": [],
   "source": [
    "df['message'] = df['message'].apply(text_process)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T15:04:43.611474Z",
     "start_time": "2018-08-30T15:04:43.581422Z"
    }
   },
   "outputs": [],
   "source": [
    "df['message'] = df['message'].apply(lambda x: ' '.join(x))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T15:05:47.901529Z",
     "start_time": "2018-08-30T15:05:47.891614Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0                                                  yup\n",
       "1    rolex fover genuine plica watch thank expssing...\n",
       "2    watch lor saw swatch one thk quite ok ard 116 ...\n",
       "3    methanol plant status appears methanol plant m...\n",
       "5    fw crosstex energy driscoll ranch 1 3 meter 98...\n",
       "Name: message, dtype: object"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df['message'].head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T15:06:25.627067Z",
     "start_time": "2018-08-30T15:06:25.166441Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import CountVectorizer\n",
    "bow_transformer = CountVectorizer().fit(df['message'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T15:06:37.104456Z",
     "start_time": "2018-08-30T15:06:37.094033Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "37648\n"
     ]
    }
   ],
   "source": [
    "print(len(bow_transformer.vocabulary_))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T15:06:47.805923Z",
     "start_time": "2018-08-30T15:06:47.429742Z"
    }
   },
   "outputs": [],
   "source": [
    "message_bow = bow_transformer.transform(df['message'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T15:06:59.465086Z",
     "start_time": "2018-08-30T15:06:59.460453Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sparsity 0.07220618615096115\n"
     ]
    }
   ],
   "source": [
    "sparsity = (100.0 * message_bow.nnz/(message_bow.shape[0] * message_bow.shape[1]))\n",
    "print('sparsity {}'.format(sparsity))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T15:07:14.019923Z",
     "start_time": "2018-08-30T15:07:14.009345Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import TfidfTransformer\n",
    "tfidf_transformer_bow = TfidfTransformer().fit(message_bow)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T15:11:06.926378Z",
     "start_time": "2018-08-30T15:11:06.906414Z"
    }
   },
   "outputs": [],
   "source": [
    "message_tfidf = tfidf_transformer_bow.transform(message_bow)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T15:11:25.017380Z",
     "start_time": "2018-08-30T15:11:25.001887Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.naive_bayes import MultinomialNB\n",
    "\n",
    "def buildClassifier(alpha):\n",
    "    text_classifier = MultinomialNB(alpha=alpha)\n",
    "    return text_classifier"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Making the pipeline, making it production ready!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T15:14:58.924440Z",
     "start_time": "2018-08-30T15:14:58.917861Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.pipeline import Pipeline\n",
    "pipeline = Pipeline([('vec', CountVectorizer()),\n",
    "                   ('tfidf', TfidfTransformer()),\n",
    "                   ('clf', MultinomialNB(alpha=0.01))])\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<b> Grid Search for best parameters. </b>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T15:14:59.263144Z",
     "start_time": "2018-08-30T15:14:59.258164Z"
    }
   },
   "outputs": [],
   "source": [
    "from sklearn.model_selection import GridSearchCV"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T19:58:39.588238Z",
     "start_time": "2018-08-30T19:58:39.583385Z"
    }
   },
   "outputs": [],
   "source": [
    "parameters = {'clf__alpha': (0.00001, 0.0001, 0.001, 0.01, 0.1, 0.8, 0.9, 1)}\n",
    "grid_search = GridSearchCV(estimator=pipeline, param_grid=parameters, scoring='accuracy', cv=10, verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T19:59:39.505044Z",
     "start_time": "2018-08-30T19:58:42.275759Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 10 folds for each of 8 candidates, totalling 80 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=1)]: Done  80 out of  80 | elapsed:   56.7s finished\n"
     ]
    }
   ],
   "source": [
    "grid_search.fit(df['message'], df['labels'])\n",
    "best_parameters = grid_search.best_params_\n",
    "best_accuracy = grid_search.best_score_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T20:09:36.638101Z",
     "start_time": "2018-08-30T20:09:36.633771Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'clf__alpha': 0.1}"
      ]
     },
     "execution_count": 115,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "best_parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T20:09:40.945460Z",
     "start_time": "2018-08-30T20:09:40.935122Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.965125315717894"
      ]
     },
     "execution_count": 116,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "best_accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T19:41:32.561731Z",
     "start_time": "2018-08-30T19:41:32.556139Z"
    }
   },
   "outputs": [],
   "source": [
    "pipeline = Pipeline([('vec', CountVectorizer()),\n",
    "                   ('tfidf', TfidfTransformer()),\n",
    "                   ('clf', MultinomialNB(alpha=0.1))])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T19:41:53.082546Z",
     "start_time": "2018-08-30T19:41:52.611425Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Pipeline(memory=None,\n",
       "     steps=[('vec', CountVectorizer(analyzer='word', binary=False, decode_error='strict',\n",
       "        dtype=<class 'numpy.int64'>, encoding='utf-8', input='content',\n",
       "        lowercase=True, max_df=1.0, max_features=None, min_df=1,\n",
       "        ngram_range=(1, 1), preprocessor=None, stop_words=None,\n",
       "        strip_...inear_tf=False, use_idf=True)), ('clf', MultinomialNB(alpha=0.1, class_prior=None, fit_prior=True))])"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pipeline.fit(df['message'], df['labels'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T19:54:14.128995Z",
     "start_time": "2018-08-30T19:54:14.115849Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['spam'], dtype='<U4')"
      ]
     },
     "execution_count": 111,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "input_word = ['Download this app for free music']\n",
    "input_word = text_process(input_word)\n",
    "input_word = [' '.join(input_word)]\n",
    "pipeline.predict(input_word)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Saving the Pipeline to be integrated in our web-app"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T19:50:32.685591Z",
     "start_time": "2018-08-30T19:50:32.681042Z"
    }
   },
   "outputs": [],
   "source": [
    "import pickle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2018-08-30T19:51:47.919976Z",
     "start_time": "2018-08-30T19:51:47.798417Z"
    }
   },
   "outputs": [],
   "source": [
    "filename = 'textClf.pkl'\n",
    "pickle.dump(pipeline, open(filename, 'wb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.4"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": false
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}