988121project2.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.preprocessing import StandardScaler \n",
    "from sklearn.decomposition import PCA\n",
    "from scipy.spatial.distance import pdist, squareform\n",
    "import seaborn as sns\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.neighbors import KNeighborsClassifier \n",
    "from sklearn.metrics import accuracy_score\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.cluster import KMeans\n",
    "from scipy.spatial.distance import cdist\n",
    "from sklearn.metrics import mutual_info_score\n",
    "#to ignore the \"FutureWarning: Using a non-tuple sequence for multidimensional indexing is deprecated;\" \n",
    "#Google says it's a version problem, I have no idea, but Google tell me do add these 2 lines so I did\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.spatial.distance import pdist, squareform\n",
    "def VAT(R):\n",
    "    R = np.array(R)\n",
    "    N, M = R.shape\n",
    "    if N != M:\n",
    "        R = squareform(pdist(R))\n",
    "    J = list(range(0, N))\n",
    "    \n",
    "    y = np.max(R, axis=0)\n",
    "    i = np.argmax(R, axis=0)\n",
    "    j = np.argmax(y)\n",
    "    y = np.max(y)\n",
    "    \n",
    "    I = i[j]\n",
    "    del J[I]\n",
    "    \n",
    "    y = np.min(R[I,J], axis=0)\n",
    "    j = np.argmin(R[I,J], axis=0)\n",
    "    I = [I, J[j]]\n",
    "    J = [e for e in J if e != J[j]]\n",
    "    C = [1,1]\n",
    "\n",
    "    for r in range(2, N-1):\n",
    "        y = np.min(R[I,:][:,J], axis=0)\n",
    "        i = np.argmin(R[I,:][:,J], axis=0)\n",
    "        j = np.argmin(y)\n",
    "        y = np.min(y)\n",
    "        I.extend([J[j]])\n",
    "        J = [e for e in J if e != J[j]]\n",
    "        C.extend([i[j]])\n",
    "\n",
    "    y = np.min(R[I,:][:,J], axis=0)\n",
    "    i = np.argmin(R[I,:][:,J], axis=0)\n",
    "\n",
    "    I.extend(J)\n",
    "    C.extend(i)\n",
    "\n",
    "    RI = list(range(N))\n",
    "    for idx, val in enumerate(I):\n",
    "        RI[val] = idx\n",
    "\n",
    "    RV = R[I,:][:,I]\n",
    "    return RV.tolist(), C, I\n",
    "\n",
    "\n",
    "def entropy(probs):\n",
    "    \n",
    "    return -probs.dot(np.log2(probs))\n",
    "\n",
    "\n",
    "def mutual_info(df):\n",
    "    \n",
    "    Hx = entropy(df.iloc[:,0].value_counts(normalize=True, sort=False))\n",
    "    Hy = entropy(df.iloc[:,1].value_counts(normalize=True, sort=False))\n",
    "    \n",
    "    counts = df.groupby(list(df.columns.values)).size()\n",
    "    probs = counts/ counts.values.sum()\n",
    "    H_xy = entropy(probs)\n",
    "\n",
    "    # Mutual Information\n",
    "    I_xy = Hx + Hy - H_xy\n",
    "    MI = I_xy\n",
    "    NMI = I_xy/min(Hx,Hy) #I_xy/np.sqrt(H_x*H_y)\n",
    "    \n",
    "    return {'H_'+list(df)[0]:Hx,'H_'+list(df)[1]:Hy,'MI':MI,'NMI':NMI} \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "food =  pd.read_csv(\"food_nutrient_2011_13_AHS.csv\", header=0,low_memory=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "***\n",
      "Q1.c:foodscaled matrix details\n",
      "Number of rows: 5740\n",
      "Number of columns: 53\n",
      "Min: -2.0\n",
      "Max: 75.8\n",
      "Mean: 0.0\n",
      "Standard Deviation 1.0\n",
      "***\n"
     ]
    }
   ],
   "source": [
    "##Question 1  Feature Standardisation\n",
    "\n",
    "###1a\n",
    "#pick all the continuous features from food and store in a new dataframe food_with_con\n",
    "food_with_con = food.loc[:,'Energy, with dietary fibre (kJ)':'Total trans fatty acids (mg)']\n",
    "#transform all the data into the float type (cause there are some int type feature in the dataframe)\n",
    "food_with_con = food_with_con.astype(\"float\")\n",
    "\n",
    "###1b\n",
    "#normalise the data to have 0 mean and unit variance using the library functions. \n",
    "scaler = StandardScaler().fit(food_with_con)\n",
    "foodscaled=scaler.transform(food_with_con)\n",
    "\n",
    "#change the matrix into the dataframe form \n",
    "foodscaled = pd.DataFrame(foodscaled)\n",
    "\n",
    "\n",
    "###1c\n",
    "#find the number of rows and columns of the foodscaled\n",
    "nRows = foodscaled.shape[0]\n",
    "nCols = foodscaled.shape[1]\n",
    "\n",
    "#find the minimum and maximum number in the foodscaled \n",
    "minValue = round(foodscaled.min().min(), 1)\n",
    "maxValue = round(foodscaled.max().max(), 1)\n",
    "\n",
    "#find the mean and standard deviation of the foodscaled \n",
    "meanValue = round(foodscaled.mean().mean(), 1)\n",
    "stdValue = round(foodscaled.values.std(), 1)\n",
    "\n",
    "#print all the result out \n",
    "print(\"***\")\n",
    "print(\"Q1.c:foodscaled matrix details\")\n",
    "print(\"Number of rows: \"+str(nRows))\n",
    "print(\"Number of columns: \"+str(nCols))\n",
    "print(\"Min: \"+str(minValue))\n",
    "print(\"Max: \"+str(maxValue))\n",
    "print(\"Mean: \"+str(-meanValue)) # cause the float type problem, there is a negative sign before the mean value, so I just add a negative sign to make it more beautiful, it will not effect the result cause we normalize the mean value to 0.\n",
    "print(\"Standard Deviation \"+str(stdValue))\n",
    "print(\"***\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1a1d3574e0>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Question 2 Principal Components Analysis\n",
    "#reset the foodscaled at the beginning of the question. avoid some unexpected error\n",
    "#comment of these code is at the question 1, I just copy these from question 1, nothing new\n",
    "food_with_con = food.loc[:,'Energy, with dietary fibre (kJ)':'Total trans fatty acids (mg)']\n",
    "food_with_con = food_with_con.astype(\"float\")\n",
    "scaler = StandardScaler().fit(food_with_con)\n",
    "foodscaled=scaler.transform(food_with_con)\n",
    "foodscaled = pd.DataFrame(foodscaled)\n",
    "\n",
    "###2a\n",
    "#create the feature EnergyLevel which has value '1' if unstandardized Energy, with dietary fibre (kJ)is greater than 1000kj and '0' otherwise \n",
    "EnergyLevel= food['Energy, with dietary fibre (kJ)'].apply(lambda x: '1' if x > 1000 else '0')\n",
    "\n",
    "###2b\n",
    "#apply principal components analysis to foodscaled and set the number oc components equals to 2\n",
    "pca=PCA(n_components=2)\n",
    "#store the result in the foodreduced \n",
    "foodreduced=pca.fit_transform(foodscaled)\n",
    "#transform the foodreduced into the dataframe form \n",
    "foodreduced= pd.DataFrame(foodreduced)\n",
    "foodreduced.columns = [\"1st Principal Component\",\"2nd Principal Component\"]\n",
    "\n",
    "###2c\n",
    "#add the feature EnergyLevel in the foodreduced to draw the scatter plot\n",
    "foodreduced[\"EnergyLevel\"] = EnergyLevel\n",
    "\n",
    "#seperate that value into different color according to the energylevel\n",
    "blueElement = foodreduced[foodreduced[\"EnergyLevel\"]=='0']\n",
    "redElement = foodreduced[foodreduced[\"EnergyLevel\"]=='1']\n",
    "\n",
    "#plot low energylevel scatter plot(with color blue) first and add another data(with color red) at the same plot\n",
    "ax = blueElement.plot.scatter(x=\"1st Principal Component\",y=\"2nd Principal Component\", c='b', label='Low energy')\n",
    "redElement.plot.scatter(x=\"1st Principal Component\",y=\"2nd Principal Component\",c='r', label = 'High energy', ax=ax)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Question 2d \n",
    "\n",
    "1) The scatter plot is a projection that captures the largest amount of nutrition variance of different food, (we also seperate different level energy food by different color). \n",
    "\n",
    "2) The nutrition variance of low_energy food is less than high_energy one. and there is a significant differece in the nutrient variance between low_energy food and high_energy food \n",
    "\n",
    "3) \n",
    "Advantage: The original food dataset has 53 features, it is hard to visualize. The PCA can make the dataset into a 2-dimensional dataset to give a clear sense of variance. \n",
    "\n",
    "Disadvantage: On the process of dimension reduction, Some data will inevitably be lost. (eg.we cannot know which attribute casuse the biggest difference in food.)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Question 3\n",
    "#reset the foodscaled at the beginning of the question, avoid unexpected error\n",
    "#comment of these code is at the question 1, I just copy these from question 1, nothing new\n",
    "food_with_con = food.loc[:,'Energy, with dietary fibre (kJ)':'Total trans fatty acids (mg)']\n",
    "food_with_con = food_with_con.astype(\"float\")\n",
    "scaler = StandardScaler().fit(food_with_con)\n",
    "foodscaled=scaler.transform(food_with_con)\n",
    "foodscaled = pd.DataFrame(foodscaled)\n",
    "\n",
    "\n",
    "###3a\n",
    "# Convert the datatype of the Survey ID attribute into the string type\n",
    "food['Survey ID'] = food['Survey ID'].astype('str')\n",
    "\n",
    "#create the new attribute according to the survey ID\n",
    "foodCate = food['Survey ID'].str.slice(0, 2)\n",
    "\n",
    "\n",
    "###3b\n",
    "#insert the attribute into foodscaled\n",
    "foodscaled[\"Food category\"] = foodCate\n",
    "\n",
    "#find the the food with Food Category equals 13, 20, 24. and combine them into a new dataframe called foodscaledsample\n",
    "foodscaledsample = foodscaled.loc[foodscaled[\"Food category\"].isin([\"13\",\"20\",\"24\"])]\n",
    "#delete column \"Food category\" avoid it affecting the data\n",
    "del foodscaledsample[\"Food category\"]\n",
    "\n",
    "\n",
    "###3c\n",
    "\n",
    "#Apply VAT Algorithm to foodscaledsample dataset and visualise using heatmap\n",
    "RV, C, I = VAT(foodscaledsample)\n",
    "x=sns.heatmap(RV,cmap='viridis',xticklabels=False,yticklabels=False)\n",
    "x.set(xlabel='Objects', ylabel='Objects')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Question 3e \n",
    "\n",
    "1) There are 2 clusters in this heatmap. \n",
    "\n",
    "2) This is not expected, because there are 3 food category(13, 20, 24) in the dataset. \n",
    "\n",
    "3) When two of the food category are very close together and some heatmap colors will make them seems like there is  only one food category.\n",
    "\n",
    "4) One category's color is well separated from other two, and the two similar food category's colors is similar but still can tell they are different. (The greater the color difference, the better)\n",
    "\n",
    "5) Just some result after some experiments: \n",
    "\n",
    "I tried to add \"food category\" in the foodscaled, that shows me 2 obvious clusters or (3 cluster which is not so obvious). That shows the data is very close with each other, so that they will been effect so obvious by a discrete attribute(which is also being proved by question4). \n",
    "\n",
    "And I tried these 3 category one-to-one, and found there is nearly no difference betweem 20 and 24 food category. So they are the pair very close to each other. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Question 4\n",
    "#reset the foodscaled at the beginning of the question. avoid some unexpected error\n",
    "#comment of these code is at the question 1, I just copy these from question 1, nothing new\n",
    "food_with_con = food.loc[:,'Energy, with dietary fibre (kJ)':'Total trans fatty acids (mg)']\n",
    "food_with_con = food_with_con.astype(\"float\")\n",
    "scaler = StandardScaler().fit(food_with_con)\n",
    "foodscaled=scaler.transform(food_with_con)\n",
    "foodscaled = pd.DataFrame(foodscaled)\n",
    "\n",
    "#create a dictionary that will store number Of clusters and cooresponding SSE\n",
    "sse = {}\n",
    "\n",
    "#set the number of k from 2 to 25\n",
    "for k in range(2, 26):\n",
    "    #find the kmeans \n",
    "    kmeans = KMeans(n_clusters=k, max_iter=100).fit(foodscaled)\n",
    "    #and calculate the certain SSE\n",
    "    # Inertia: Sum of distances of samples to their closest cluste(definition in document)\n",
    "    # SSE(sum of squared errors): Sum of distances of objects from their cluster centroids(definition in workshop6 question5)\n",
    "    #they should be the same thing\n",
    "    sse[k] = kmeans.inertia_  \n",
    "\n",
    "#plot the sse out \n",
    "plt.figure()\n",
    "plt.plot(list(sse.keys()), list(sse.values()))\n",
    "plt.xlabel(\"Number of cluster\")\n",
    "plt.ylabel(\"SSE\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Question 4b\n",
    "\n",
    "Shape:\n",
    "\n",
    "The graph is a decreasing line.(nearly a linear decreasing line)\n",
    "\n",
    "\n",
    "Elbow:\n",
    "\n",
    "There is no \"obvious\" elbow in this graph, and every time I run my code, the small elbow will appear at different value(k-mean is random). While I found elbow at numOfCluster = 18 or numOfCluster = 19 is pretty stable.\n",
    "\n",
    "\n",
    "Analysis:\n",
    "\n",
    "This shape is expected cause as the cluster increase, the distance from each point to its cluster center will be decrese. so the shape should be decreasing. \n",
    "\n",
    "And for the elbow, There are 22 food category and there are the nutrition variance of some food category are close to each other so the elbow should exist between 15 to 22.\n",
    "\n",
    "At the same time there's no \"obvious\" elbow in the graph shows that these data are distributed evenly."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "###Question 5\n",
    "\n",
    "###5a\n",
    "# find out the corresponding features \n",
    "food_corr = food.loc[:,'Energy, with dietary fibre (kJ)':'Added sugars (g)']\n",
    "#calculate their pearson correlation\n",
    "correlationDF = food_corr.corr(method='pearson')\n",
    "\n",
    "#use heatmap to plot the matrix \n",
    "ax=sns.heatmap(correlationDF,cmap='YlGnBu')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#5b\n",
    "#take the attribute 'Energy, with dietary fibre (kJ)' and 'Protein (g)' out. \n",
    "Energy = food['Energy, with dietary fibre (kJ)']\n",
    "Protein = food['Protein (g)']\n",
    "\n",
    "#create a dictionary that will store bin-width and cooresponding Mutual Information\n",
    "MI = {}\n",
    "\n",
    "\n",
    "#use the cut function to cassify the data with bin-width = 2\n",
    "cuttedEnergy = pd.cut(Energy, 2, labels=False)\n",
    "cuttedProtein = pd.cut(Protein, 2, labels=False)\n",
    "#use the provided MI function to find the mutual information\n",
    "currentDataFrame = pd.DataFrame({'Energy':cuttedEnergy, 'Protein':cuttedProtein})\n",
    "result = mutual_info(currentDataFrame)\n",
    "#store the result in the MI dictionary\n",
    "MI[2] = result.get(\"MI\")\n",
    "\n",
    "\n",
    "#set the number of l from 10 to 200\n",
    "for l in range(10, 210, 10):\n",
    "    #use the cut function to cassify the data with certain bin-width\n",
    "    cuttedEnergy = pd.cut(Energy, l, labels=False)\n",
    "    cuttedProtein = pd.cut(Protein, l, labels=False)\n",
    "    currentDataFrame = pd.DataFrame({'Energy':cuttedEnergy, 'Protein':cuttedProtein})\n",
    "    #use the provided MI function to find the mutual information\n",
    "    result = mutual_info(currentDataFrame)\n",
    "    #store the result in the MI dictionary\n",
    "    MI[l] = result.get(\"MI\")\n",
    "    \n",
    "#plot the MI out\n",
    "plt.figure()\n",
    "plt.plot(list(MI.keys()), list(MI.values()))\n",
    "plt.xlabel(\"bin-width\")\n",
    "plt.ylabel(\"MI\")\n",
    "plt.show()    \n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##### Question 5b \n",
    "\n",
    "It is a increasing trend, the MI value increasing slowly before bin-width = 20. becomes a linear increasing line after bin-width=20.\n",
    "\n",
    "The formula for mutual information: MI(MI(X, Y) = H(X)-H(X|y).H(X) evaluate the entropy of X, and as the bin-width increase, the distribution of the values will be more uniform, so that the entropy value of X will be increase. So as the bin-width increase, we can see more clear that how X and Y are related. \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "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>feature1</th>\n",
       "      <th>feature2</th>\n",
       "      <th>Pearson Value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Alpha-tocopherol (mg)</td>\n",
       "      <td>Vitamin E (mg)</td>\n",
       "      <td>0.999528</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Energy, with dietary fibre (kJ)</td>\n",
       "      <td>Energy, without dietary fibre (kJ)</td>\n",
       "      <td>0.998608</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Beta-carotene (µg)</td>\n",
       "      <td>Provitamin A (b-carotene equivalents) (µg)</td>\n",
       "      <td>0.995961</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Available carbohydrates, with sugar alcohols (g)</td>\n",
       "      <td>Available carbohydrates, without sugar alcohol...</td>\n",
       "      <td>0.995466</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Free sugars (g)</td>\n",
       "      <td>Added sugars (g)</td>\n",
       "      <td>0.988894</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Total Folates  (µg)</td>\n",
       "      <td>Dietary folate equivalents  (µg)</td>\n",
       "      <td>0.985763</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Total polyunsaturated fat (g)</td>\n",
       "      <td>Linoleic acid (g)</td>\n",
       "      <td>0.973640</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Folic acid  (µg)</td>\n",
       "      <td>Dietary folate equivalents  (µg)</td>\n",
       "      <td>0.958112</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>Preformed vitamin A (retinol) (µg)</td>\n",
       "      <td>Vitamin A retinol equivalents (µg)</td>\n",
       "      <td>0.954100</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>C22:6w3 Docosahexaenoic (mg)</td>\n",
       "      <td>Total long chain omega 3 fatty acids (mg)</td>\n",
       "      <td>0.952762</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                            feature1  \\\n",
       "1                              Alpha-tocopherol (mg)   \n",
       "2                    Energy, with dietary fibre (kJ)   \n",
       "3                                 Beta-carotene (µg)   \n",
       "4   Available carbohydrates, with sugar alcohols (g)   \n",
       "5                                    Free sugars (g)   \n",
       "6                                Total Folates  (µg)   \n",
       "7                      Total polyunsaturated fat (g)   \n",
       "8                                   Folic acid  (µg)   \n",
       "9                 Preformed vitamin A (retinol) (µg)   \n",
       "10                      C22:6w3 Docosahexaenoic (mg)   \n",
       "\n",
       "                                             feature2  Pearson Value  \n",
       "1                                      Vitamin E (mg)       0.999528  \n",
       "2                  Energy, without dietary fibre (kJ)       0.998608  \n",
       "3          Provitamin A (b-carotene equivalents) (µg)       0.995961  \n",
       "4   Available carbohydrates, without sugar alcohol...       0.995466  \n",
       "5                                    Added sugars (g)       0.988894  \n",
       "6                    Dietary folate equivalents  (µg)       0.985763  \n",
       "7                                   Linoleic acid (g)       0.973640  \n",
       "8                    Dietary folate equivalents  (µg)       0.958112  \n",
       "9                  Vitamin A retinol equivalents (µg)       0.954100  \n",
       "10          Total long chain omega 3 fatty acids (mg)       0.952762  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "###5c \n",
    "\n",
    "#Pearson correlation \n",
    "\n",
    "#pick all the continuous features from food and store in a new dataframe food_with_con\n",
    "food_with_con = food.loc[:,'Energy, with dietary fibre (kJ)':'Total trans fatty acids (mg)']\n",
    "#transform all the data into the float type (cause there are some int type feature in the dataframe)\n",
    "food_with_con = food_with_con.astype(\"float\")\n",
    "\n",
    "#calculate 53 pairs of Pearson correlation\n",
    "correlationDF = food_with_con.corr(method='pearson')\n",
    "#change all values on the diagonal to zero \n",
    "correlationDF.values[[np.arange(len(correlationDF.index))]*2] = 0\n",
    "#find the top 10 feature-pairs \n",
    "sortedCorrelation = correlationDF.max().sort_values(ascending = False)\n",
    "\n",
    "\n",
    "\n",
    "#store the feature and Pearson value in the TopPearsonCor dataframe\n",
    "#feature1 is the one of the feature pair and feature 2 is another\n",
    "feature1 = []\n",
    "feature2 = []\n",
    "# Pearson_value is the Pearson correlation Pearson value corresponding to the feature1 and feature2\n",
    "Pearson_Value = []\n",
    "\n",
    "#cause the series has already sorted, so we only need to find top 10 by the order of the series \n",
    "for index in sortedCorrelation.index:\n",
    "    # when there are already 10 values, then stop the loop\n",
    "    if(len(Pearson_Value)==10):\n",
    "        break\n",
    "        \n",
    "    #when there are no value in the list or this value is not repeated value of last pair(feature1 VS feature2 and feature2 VS feature1 are repeated)\n",
    "    if(len(Pearson_Value)==0 or sortedCorrelation[index]!=Pearson_Value[-1]):\n",
    "        #store the features and Pearson correlation value into the list\n",
    "        feature1.append(index)\n",
    "        Pearson_Value.append(sortedCorrelation[index])\n",
    "        corfeature = correlationDF[correlationDF[index]==sortedCorrelation[index]].index.tolist()[0]\n",
    "        feature2.append(corfeature)\n",
    "        \n",
    "#store all these information together \n",
    "data = {\"feature1\":feature1, \"feature2\":feature2, \"Pearson Value\":Pearson_Value}\n",
    "ToptenPearsoncor = pd.DataFrame(data)\n",
    "newIndex = list(range(1, 11))\n",
    "ToptenPearsoncor.index  = newIndex\n",
    "#print the dataframe out\n",
    "ToptenPearsoncor\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "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>feature1</th>\n",
       "      <th>feature2</th>\n",
       "      <th>Mutual information</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Energy, with dietary fibre (kJ)</td>\n",
       "      <td>Energy, without dietary fibre (kJ)</td>\n",
       "      <td>3.140250</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Available carbohydrates, with sugar alcohols (g)</td>\n",
       "      <td>Available carbohydrates, without sugar alcohol...</td>\n",
       "      <td>3.070813</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Energy, with dietary fibre (kJ)</td>\n",
       "      <td>Moisture (g)</td>\n",
       "      <td>1.880160</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Energy, without dietary fibre (kJ)</td>\n",
       "      <td>Moisture (g)</td>\n",
       "      <td>1.791029</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Tryptophan (mg)</td>\n",
       "      <td>Protein (g)</td>\n",
       "      <td>1.359985</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Free sugars (g)</td>\n",
       "      <td>Added sugars (g)</td>\n",
       "      <td>1.253871</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Starch (g)</td>\n",
       "      <td>Available carbohydrates, without sugar alcohol...</td>\n",
       "      <td>1.101232</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Starch (g)</td>\n",
       "      <td>Available carbohydrates, with sugar alcohols (g)</td>\n",
       "      <td>1.095449</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>Free sugars (g)</td>\n",
       "      <td>Total sugars (g)</td>\n",
       "      <td>0.978975</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Energy, with dietary fibre (kJ)</td>\n",
       "      <td>Total fat (g)</td>\n",
       "      <td>0.959014</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                            feature1  \\\n",
       "1                    Energy, with dietary fibre (kJ)   \n",
       "2   Available carbohydrates, with sugar alcohols (g)   \n",
       "3                    Energy, with dietary fibre (kJ)   \n",
       "4                 Energy, without dietary fibre (kJ)   \n",
       "5                                    Tryptophan (mg)   \n",
       "6                                    Free sugars (g)   \n",
       "7                                         Starch (g)   \n",
       "8                                         Starch (g)   \n",
       "9                                    Free sugars (g)   \n",
       "10                   Energy, with dietary fibre (kJ)   \n",
       "\n",
       "                                             feature2  Mutual information  \n",
       "1                  Energy, without dietary fibre (kJ)            3.140250  \n",
       "2   Available carbohydrates, without sugar alcohol...            3.070813  \n",
       "3                                        Moisture (g)            1.880160  \n",
       "4                                        Moisture (g)            1.791029  \n",
       "5                                         Protein (g)            1.359985  \n",
       "6                                    Added sugars (g)            1.253871  \n",
       "7   Available carbohydrates, without sugar alcohol...            1.101232  \n",
       "8    Available carbohydrates, with sugar alcohols (g)            1.095449  \n",
       "9                                    Total sugars (g)            0.978975  \n",
       "10                                      Total fat (g)            0.959014  "
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#Mutual Information\n",
    "#pick all the continuous features from food and store in a new dataframe food_with_con\n",
    "food_with_con = food.loc[:,'Energy, with dietary fibre (kJ)':'Total trans fatty acids (mg)']\n",
    "#transform all the data into the float type (cause there are some int type feature in the dataframe)\n",
    "food_with_con = food_with_con.astype(\"float\")\n",
    "\n",
    "#create empty cloumn feature pair and Mutual Information for the dataframe \n",
    "#attribute1 is the one of the feature pair and attribute 2 is another\n",
    "attribute1 =[]\n",
    "attribute2 =[]\n",
    "#Mutual_information is the corresponding MI for the certain pairs\n",
    "Mutual_information = []\n",
    "\n",
    "# go through every pair in the continuous features\n",
    "for index1 in food_with_con.columns:\n",
    "    for index2 in food_with_con.columns:\n",
    "        if index1==index2:\n",
    "            pass\n",
    "        else:\n",
    "            #store the certain feature name in the attributes list \n",
    "            attribute1.append(index1)\n",
    "            attribute2.append(index2)\n",
    "            \n",
    "            #find the Mutual information by the MI function with bin-width 20\n",
    "            feature1= food_with_con[index1]\n",
    "            feature2= food_with_con[index2]\n",
    "            cuttedfeature1 = pd.cut(feature1, 20, labels=False)\n",
    "            cuttedfeature2 = pd.cut(feature2, 20, labels=False)\n",
    "            currentDataFrame = pd.DataFrame({'feature1':cuttedfeature1, 'Protein':cuttedfeature2})\n",
    "            result = mutual_info(currentDataFrame)\n",
    "            #store the result in the MI list \n",
    "            Mutual_information.append(result[\"MI\"])\n",
    "            \n",
    "#store all the information together as a dataframe\n",
    "data = {\"feature1\":attribute1, \"feature2\":attribute2, \"Mutual information\":Mutual_information}\n",
    "MIlist = pd.DataFrame(data)\n",
    "#sort the dataframe by the value of mutual information \n",
    "sortedMI = MIlist.sort_values(by=\"Mutual information\", ascending = False)    \n",
    "\n",
    "# then find the top 10 MI value \n",
    "toptenMI = sortedMI.iloc[0:20, :]\n",
    "newIndex = list(range(0, 20))\n",
    "toptenMI.index  = newIndex\n",
    "for i in range(0, 20):\n",
    "    if(i%2 == 1):\n",
    "        toptenMI = toptenMI.drop(i)\n",
    "newIndex = list(range(1, 11))\n",
    "toptenMI.index  = newIndex\n",
    "\n",
    "\n",
    "# print the top ten pairs out\n",
    "toptenMI"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Question 5c \n",
    "\n",
    "\n",
    "There are some top10 feature-pairs by Pearson occur in the top 10 feature_pairs by Mutual Information.\n",
    "\n",
    "Pearson value is for assessing linear correlation, the top 10 feature-pairs by Pearson are the top 10 feature-pairs has most strong linear correlation. \n",
    "\n",
    "And Mutual Information is a measure of correlation the amount of information about one attribute we gain by another.\n",
    "\n",
    "So the linear correlation can also being measure by Mutual Information but not as accurate as Pearson.\n",
    "\n",
    "That's why some of top10 feature-pairs by Pearson occur in the top 10 feature_pairs by Mutual Information. \n",
    "\n",
    "And other top 10 feature_pairs by Mutual Information(that is not in Pearson) are the ones got other \n",
    "correlation(eg.quadratic correlation) that is not defect by the Pearson"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "***\n",
      "X_train matrix:  (4592, 53)\n",
      "y_train labels:  (4592, 1)\n",
      "X_test matrix:  (1148, 53)\n",
      "y_test labels:  (1148, 1)\n",
      "***\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "##Question 6\n",
    "#reset the foodscaled at the beginning of the question\n",
    "food_with_con = food.loc[:,'Energy, with dietary fibre (kJ)':'Total trans fatty acids (mg)']\n",
    "food_with_con = food_with_con.astype(\"float\")\n",
    "scaler = StandardScaler().fit(food_with_con)\n",
    "foodscaled=scaler.transform(food_with_con)\n",
    "foodscaled = pd.DataFrame(foodscaled)\n",
    "\n",
    "#reset the food category attribute\n",
    "food['Survey ID'] = food['Survey ID'].astype('str')\n",
    "foodCate = food['Survey ID'].str.slice(0, 2)\n",
    "\n",
    "###6a\n",
    "#randomly select 80% of instances to be training and the rest to be testing \n",
    "X_train, X_test, y_train, y_test = train_test_split(foodscaled, foodCate,train_size=0.80, test_size=0.20, random_state=0)\n",
    "\n",
    "#cause there is only 1 attribute in the y_train and y_test, so ther is no values on shape[1], so we add 1 manully\n",
    "y_train_tuple = (y_train.shape[0], 1)\n",
    "y_test_tuple = (y_test.shape[0], 1)\n",
    "\n",
    "#print out all the result \n",
    "print(\"***\")\n",
    "print(\"X_train matrix: \", str(X_train.shape))\n",
    "print(\"y_train labels: \", str(y_train_tuple))\n",
    "print(\"X_test matrix: \", str(X_test.shape))\n",
    "print(\"y_test labels: \", str(y_test_tuple))\n",
    "print(\"***\")\n",
    "\n",
    "\n",
    "###6b\n",
    "# create a dictionary to \n",
    "accuracy = {}\n",
    "for k in range(1, 41):\n",
    "    #train the decision tree by useing training data, and set the max_depth as k\n",
    "    clf = DecisionTreeClassifier(criterion = \"entropy\", random_state= 1, max_depth=k)\n",
    "    clf.fit(X_train, y_train)\n",
    "    #calculate the accuracy score for using on the testing data\n",
    "    y_pred = clf.predict(X_test)\n",
    "    accuracy[k] = accuracy_score(y_test, y_pred)\n",
    "\n",
    "#Plot the accuracy dictionary out     \n",
    "plt.figure()\n",
    "plt.plot(list(accuracy.keys()), list(accuracy.values()))\n",
    "plt.xlabel(\"Maximum Depth\")\n",
    "plt.ylabel(\"Accuracy\")\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Question 6c \n",
    "The shape of the graph shows that \n",
    "\n",
    "      accuracy increase rapidly as the maximum-depth increase, 0 <= maximum-depth <= 12\n",
    "      \n",
    "      accuracy reach its maximum value 0.85 , maximum-depth = 11\n",
    "      \n",
    "      accuracy drop a little bit and become and constant value , maximum-depth > 11\n",
    "\n",
    "max-depth is how depth that allow the tree to grow to, and at the same time the model will be more complex. It will much fitter the training data, so that the accuracy will increase. this is why graph act like this at the beginning. As max-depth continue increasing, There are not so many features affect the result, so the change of accuracy becomes small and tend to be a specific value.\n",
    "\n",
    "\"If set max_depth too high, then the decision tree might simply overfit the training data without capturing useful patterns as we would like; this will cause testing error to increase.while if set it too low, that is not good as well\" (quote from https://stackoverflow.com/questions/49289187/decision-tree-sklearn-depth-of-tree-and-accuracy). According to our graph,  maximum-depth = 11 is the best choice."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "***\n",
      "Q7a: Food category prediction using k-NN(k=1)\n",
      "Train accuracy: 100.0%\n",
      "Test accuracy: 89.5%\n",
      "***\n",
      "***\n",
      "Q7b: Food category prediction using k-NN(k=3)\n",
      "Train accuracy: 93.3%\n",
      "Test accuracy: 88.0%\n",
      "***\n"
     ]
    }
   ],
   "source": [
    "#Question 7\n",
    "\n",
    "#reset the foodscaled at the beginning of the question\n",
    "food_with_con = food.loc[:,'Energy, with dietary fibre (kJ)':'Total trans fatty acids (mg)']\n",
    "food_with_con = food_with_con.astype(\"float\")\n",
    "scaler = StandardScaler().fit(food_with_con)\n",
    "foodscaled=scaler.transform(food_with_con)\n",
    "foodscaled = pd.DataFrame(foodscaled)\n",
    "\n",
    "#reset the food category attribute\n",
    "food['Survey ID'] = food['Survey ID'].astype('str')\n",
    "foodCate = food['Survey ID'].str.slice(0, 2)\n",
    "\n",
    "###7a\n",
    "\n",
    "#randomly select 80% of instances to be training and the rest to be testing \n",
    "X_train, X_test, y_train, y_test = train_test_split(foodscaled, foodCate, test_size=0.2)\n",
    "\n",
    "# using the training data to predict the result by 1-neighbur-classifier\n",
    "neigh = KNeighborsClassifier(n_neighbors=1)\n",
    "neigh.fit(X_train, y_train) \n",
    "\n",
    "#calculate the accuracy for both training data and testing data \n",
    "acc1 = neigh.score(X_train, y_train) \n",
    "acc2 = neigh.score(X_test, y_test) \n",
    "\n",
    "#print the result out \n",
    "print(\"***\")\n",
    "print(\"Q7a: Food category prediction using k-NN(k=1)\")\n",
    "print(\"Train accuracy: \"+str(round(acc1*100, 1))+\"%\")\n",
    "print(\"Test accuracy: \"+str(round(acc2*100, 1))+\"%\")\n",
    "print(\"***\")\n",
    "\n",
    "\n",
    "\n",
    "##7b\n",
    "#randomly select 80% of instances to be training and the rest to be testing \n",
    "X_train, X_test, y_train, y_test = train_test_split(foodscaled, foodCate, test_size=0.2, random_state=42)\n",
    "\n",
    "# using the training data to predict the result by 3-neighbur-classifier\n",
    "neigh2 = KNeighborsClassifier(n_neighbors=3)\n",
    "neigh2.fit(X_train, y_train) \n",
    "\n",
    "#calculate the accuracy for both training data and testing data\n",
    "acc1 = neigh2.score(X_train, y_train) \n",
    "acc2 = neigh2.score(X_test, y_test) \n",
    "\n",
    "#print the result out \n",
    "print(\"***\")\n",
    "print(\"Q7b: Food category prediction using k-NN(k=3)\")\n",
    "print(\"Train accuracy: \"+str(round(acc1*100, 1))+\"%\")\n",
    "print(\"Test accuracy: \"+str(round(acc2*100, 1))+\"%\")\n",
    "print(\"***\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Question 7c \n",
    "\n",
    "Similarity:\n",
    "\n",
    "1) The test accuracy is pretty similar for k=1 and k=3 \n",
    "\n",
    "2) The training accuracy is always higher than the testing accuracy \n",
    "\n",
    "Difference:\n",
    "\n",
    "1) The train accuracy for k=1 and k=3 is different\n",
    "\n",
    "Analysis: \n",
    "\n",
    "For k=1, the data is trained with its nearest point, so its accuracy should be 100%, (just like what it being trained use into the test, they are the same thing). while for k=3, the data's uncertainty increase, so even for the training data itselt, the accuracy is decreased.   \n",
    "\n",
    "\n",
    "Question 7d \n",
    "\n",
    "The accurancy would be biased, since the testing data was looked at when doing feature selection process that should not have been seen. Consequenctly the model that was trained using the results from the feature selection was developed on information that should not have been seen. The accuracy will thus likely be over optimistic \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Question 8"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# The Interaction Term  Pairs Parts "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "#First Method to generate the generate the new feature\n",
    "#Interaction term pairs\n",
    "#fpair = f1*f2\n",
    "food_with_con = food.loc[:,'Energy, with dietary fibre (kJ)':'Total trans fatty acids (mg)']\n",
    "food_with_con = food_with_con.astype(\"float\")\n",
    "scaler = StandardScaler().fit(food_with_con)\n",
    "foodscaled=scaler.transform(food_with_con)\n",
    "foodscaled = pd.DataFrame(foodscaled)\n",
    "\n",
    "# for features in foodscaled.columns:\n",
    "#     print(foodscaled[features])\n",
    "\n",
    "foodscaled\n",
    "for i in range(53):\n",
    "    for j in range(i, 53):\n",
    "        if i == j:\n",
    "            pass\n",
    "        else:\n",
    "            newfeatureName =\"\" + str(i)+ \"*\" + str(j)\n",
    "            foodscaled[newfeatureName] = foodscaled[i]*foodscaled[j]\n",
    "foodwithmul = foodscaled\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "scrolled": false
   },
   "outputs": [],
   "source": [
    "#calculate the MI for foodwithmul\n",
    "food['Survey ID'] = food['Survey ID'].astype('str')\n",
    "foodCate = food['Survey ID'].str.slice(0, 2)\n",
    "foodCate = foodCate.astype('int')\n",
    "attribute1 =[]\n",
    "Mutual_information = []\n",
    "\n",
    "cuttedfeature2 = pd.cut(foodCate, 20, labels=False)\n",
    "\n",
    "for index1 in foodwithmul.columns:\n",
    "\n",
    "    attribute1.append(index1)\n",
    "            \n",
    "    feature1= foodwithmul[index1]\n",
    "    \n",
    "    cuttedfeature1 = pd.cut(feature1, 20, labels=False)\n",
    "\n",
    "    currentDataFrame = pd.DataFrame({'feature1':cuttedfeature1, 'Food category':cuttedfeature2})\n",
    "    \n",
    "    result = mutual_info(currentDataFrame)\n",
    "    \n",
    "    Mutual_information.append(result[\"MI\"])\n",
    "            \n",
    "data = {\"feature1\":attribute1,\"Mutual information\":Mutual_information}\n",
    "MIlist = pd.DataFrame(data)\n",
    "\n",
    "sortedMI = MIlist.sort_values(by=\"Mutual information\", ascending = False)       \n",
    "\n",
    "newIndex = list(range(sortedMI.shape[0]))\n",
    "sortedMI.index  = newIndex\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#select the different numbers of features\n",
    "food['Survey ID'] = food['Survey ID'].astype('str')\n",
    "foodCate = food['Survey ID'].str.slice(0, 2)\n",
    "\n",
    "mulaccuracy = {}\n",
    "for n in range(2, 100):\n",
    "    data = pd.DataFrame(columns=())\n",
    "    features = sortedMI.loc[0:n,[\"feature1\"]]\n",
    "    for i in features.values:\n",
    "        data[i] = foodwithmul[i]\n",
    "\n",
    "    X_train, X_test, y_train, y_test = train_test_split(data,foodCate,test_size=0.2)\n",
    "    neigh = KNeighborsClassifier(n_neighbors=1)\n",
    "    neigh.fit(X_train, y_train) \n",
    "    acc2 = neigh.score(X_test, y_test) \n",
    "    mulaccuracy[n] = acc2\n",
    "\n",
    "#plot the result\n",
    "plt.figure()\n",
    "plt.plot(list(mulaccuracy.keys()), list(mulaccuracy.values()))\n",
    "plt.xlabel(\"number of features\")\n",
    "plt.ylabel(\"Accuracy\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "#First Method to generate the generate the new feature\n",
    "#Interaction term pairs\n",
    "#fpair = f1/f2\n",
    "food_with_con = food.loc[:,'Energy, with dietary fibre (kJ)':'Total trans fatty acids (mg)']\n",
    "food_with_con = food_with_con.astype(\"float\")\n",
    "scaler = StandardScaler().fit(food_with_con)\n",
    "foodscaled=scaler.transform(food_with_con)\n",
    "foodscaled = pd.DataFrame(foodscaled)\n",
    "\n",
    "# for features in foodscaled.columns:\n",
    "#     print(foodscaled[features])\n",
    "\n",
    "foodscaled\n",
    "for i in range(53):\n",
    "    for j in range(i, 53):\n",
    "        if i == j:\n",
    "            pass\n",
    "        else:\n",
    "            newfeatureName =\"\" + str(i)+ \"/\" + str(j)\n",
    "            foodscaled[newfeatureName] = foodscaled[i]/foodscaled[j]\n",
    "foodwithfract = foodscaled  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "#calculate the MI for foodwithfract\n",
    "food['Survey ID'] = food['Survey ID'].astype('str')\n",
    "foodCate = food['Survey ID'].str.slice(0, 2)\n",
    "foodCate = foodCate.astype('int')\n",
    "attribute1 =[]\n",
    "Mutual_information = []\n",
    "\n",
    "cuttedfeature2 = pd.cut(foodCate, 20, labels=False)\n",
    "\n",
    "for index1 in foodwithfract.columns:\n",
    "\n",
    "    attribute1.append(index1)\n",
    "            \n",
    "    feature1= foodwithfract[index1]\n",
    "    \n",
    "    cuttedfeature1 = pd.cut(feature1, 20, labels=False)\n",
    "\n",
    "    currentDataFrame = pd.DataFrame({'feature1':cuttedfeature1, 'Food category':cuttedfeature2})\n",
    "    \n",
    "    result = mutual_info(currentDataFrame)\n",
    "    \n",
    "    Mutual_information.append(result[\"MI\"])\n",
    "            \n",
    "data = {\"feature1\":attribute1,\"Mutual information\":Mutual_information}\n",
    "MIlist = pd.DataFrame(data)\n",
    "\n",
    "sortedMI = MIlist.sort_values(by=\"Mutual information\", ascending = False)       \n",
    "\n",
    "newIndex = list(range(sortedMI.shape[0]))\n",
    "sortedMI.index  = newIndex"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#select the different numbers of features\n",
    "food['Survey ID'] = food['Survey ID'].astype('str')\n",
    "foodCate = food['Survey ID'].str.slice(0, 2)\n",
    "\n",
    "\n",
    "fracaccuracy = {}\n",
    "for n in range(2, 100):\n",
    "    data = pd.DataFrame(columns=())\n",
    "    features = sortedMI.loc[0:n,[\"feature1\"]]\n",
    "    for i in features.values:\n",
    "        data[i] = foodwithfract[i]\n",
    "\n",
    "    X_train, X_test, y_train, y_test = train_test_split(data,foodCate,test_size=0.2)\n",
    "    neigh = KNeighborsClassifier(n_neighbors=1)\n",
    "    neigh.fit(X_train, y_train) \n",
    "    acc2 = neigh.score(X_test, y_test) \n",
    "    fracaccuracy[n] = acc2\n",
    "\n",
    "#plot the result\n",
    "plt.figure()\n",
    "plt.plot(list(fracaccuracy.keys()), list(fracaccuracy.values()))\n",
    "plt.xlabel(\"number of features\")\n",
    "plt.ylabel(\"Accuracy\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# The Clustering Labels Parts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#######################################################\n",
    "\n",
    "food_with_con = food.loc[:,'Energy, with dietary fibre (kJ)':'Total trans fatty acids (mg)']\n",
    "food_with_con = food_with_con.astype(\"float\")\n",
    "scaler = StandardScaler().fit(food_with_con)\n",
    "foodscaled=scaler.transform(food_with_con)\n",
    "foodscaled = pd.DataFrame(foodscaled)\n",
    "\n",
    "food['Survey ID'] = food['Survey ID'].astype('str')\n",
    "foodCate = food['Survey ID'].str.slice(0, 2)\n",
    "\n",
    "kaccuracy = {}\n",
    "for k in range(2, 40):\n",
    "    #generates the cluster labels with certain k\n",
    "    kmeans = KMeans(n_clusters=k, max_iter=100).fit(foodscaled)\n",
    "    labels= kmeans.labels_\n",
    "    labels = pd.Series(labels)\n",
    "    labels.value_counts()\n",
    "    #add the generated attributes in the dataframe\n",
    "    foodscaled[\"clusterlabel\"] = labels\n",
    "    foodwithcluster = foodscaled\n",
    "    \n",
    "    #use the new dataframe to do the prediction \n",
    "    #randomly select 80% of instances to be training and the rest to be testing \n",
    "    X_train, X_test, y_train, y_test = train_test_split(foodwithcluster , foodCate, test_size=0.2)\n",
    "    neigh = KNeighborsClassifier(n_neighbors=1)\n",
    "    neigh.fit(X_train, y_train) \n",
    "    acc2 = neigh.score(X_test, y_test) \n",
    "    kaccuracy[k] = acc2\n",
    "\n",
    "#plot the graph out \n",
    "plt.figure()\n",
    "plt.plot(list(kaccuracy.keys()), list(kaccuracy.values()), \"bo\")\n",
    "plt.xlabel(\"k-mean\")\n",
    "plt.ylabel(\"Accuracy\")\n",
    "plt.show()\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "food_with_con = food.loc[:,'Energy, with dietary fibre (kJ)':'Total trans fatty acids (mg)']\n",
    "food_with_con = food_with_con.astype(\"float\")\n",
    "scaler = StandardScaler().fit(food_with_con)\n",
    "foodscaled=scaler.transform(food_with_con)\n",
    "foodscaled = pd.DataFrame(foodscaled)\n",
    "\n",
    "food['Survey ID'] = food['Survey ID'].astype('str')\n",
    "foodCate = food['Survey ID'].str.slice(0, 2)\n",
    "\n",
    "kaccuracy = {}\n",
    "for k in range(2, 40):\n",
    "    #generates the cluster labels with certain k\n",
    "    kmeans = KMeans(n_clusters=k, max_iter=100).fit(foodscaled)\n",
    "    labels= kmeans.labels_\n",
    "    labels = pd.Series(labels)\n",
    "    labels.value_counts()\n",
    "    #add the generated attributes in the dataframe\n",
    "    foodscaled[\"clusterlabel\"] = labels\n",
    "    foodwithcluster = foodscaled\n",
    "    \n",
    "    #use the new dataframe to do the prediction \n",
    "    #randomly select 80% of instances to be training and the rest to be testing \n",
    "    X_train, X_test, y_train, y_test = train_test_split(foodwithcluster , foodCate, test_size=0.2)\n",
    "    neigh = KNeighborsClassifier(n_neighbors=3)\n",
    "    neigh.fit(X_train, y_train) \n",
    "    acc2 = neigh.score(X_test, y_test) \n",
    "    kaccuracy[k] = acc2\n",
    "\n",
    "#plot the graph out \n",
    "plt.figure()\n",
    "plt.plot(list(kaccuracy.keys()), list(kaccuracy.values()), \"bo\")\n",
    "plt.xlabel(\"k-mean\")\n",
    "plt.ylabel(\"Accuracy\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "food_with_con = food.loc[:,'Energy, with dietary fibre (kJ)':'Total trans fatty acids (mg)']\n",
    "food_with_con = food_with_con.astype(\"float\")\n",
    "scaler = StandardScaler().fit(food_with_con)\n",
    "foodscaled=scaler.transform(food_with_con)\n",
    "foodscaled = pd.DataFrame(foodscaled)\n",
    "\n",
    "food['Survey ID'] = food['Survey ID'].astype('str')\n",
    "foodCate = food['Survey ID'].str.slice(0, 2)\n",
    "\n",
    "kaccuracy = {}\n",
    "for k in range(2, 40):\n",
    "    #generates the cluster labels with certain k\n",
    "    kmeans = KMeans(n_clusters=k, max_iter=100).fit(foodscaled)\n",
    "    labels= kmeans.labels_\n",
    "    labels = pd.Series(labels)\n",
    "    labels.value_counts()\n",
    "    #add the generated attributes in the dataframe\n",
    "    foodscaled[\"clusterlabel\"] = labels\n",
    "    foodwithcluster = foodscaled\n",
    "    \n",
    "    #use the new dataframe to do the prediction \n",
    "    #randomly select 80% of instances to be training and the rest to be testing \n",
    "    X_train, X_test, y_train, y_test = train_test_split(foodwithcluster , foodCate, test_size=0.2)\n",
    "    neigh = KNeighborsClassifier(n_neighbors=5)\n",
    "    neigh.fit(X_train, y_train) \n",
    "    acc2 = neigh.score(X_test, y_test) \n",
    "    kaccuracy[k] = acc2\n",
    "\n",
    "#plot the graph out \n",
    "plt.figure()\n",
    "plt.plot(list(kaccuracy.keys()), list(kaccuracy.values()), \"bo\")\n",
    "plt.xlabel(\"k-mean\")\n",
    "plt.ylabel(\"Accuracy\")\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Comment for question 8"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Question 8 \n",
    "\n",
    "comments for the Interation term pairs\n",
    "\n",
    "For both fpair = f1*f2 and fpair = f1/f2, the accuracy is similar, and it seems it cannot deliver an accuracy. But more MI-high feature we select the more accuracy the result get. \n",
    "\n",
    "comments for the clustering labels part\n",
    "\n",
    "The clustering labels method can deliver an accuracy boost. But it only contribute to K-NN when k = 1 but not for higher k values\n",
    "\n",
    "When we didn't generate new features, the accuracy has already being very high for the foodscaled and food category, so the contribute for the generate the new features seems not so obvious in this condition."
   ]
  }
 ],
 "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": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}