plot_no_components.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_2635227/1665957966.py:5: MatplotlibDeprecationWarning: The seaborn styles shipped by Matplotlib are deprecated since 3.6, as they no longer correspond to the styles shipped by seaborn. However, they will remain available as 'seaborn-v0_8-<style>'. Alternatively, directly use the seaborn API instead.\n",
      "  plt.style.use('seaborn-whitegrid')\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib\n",
    "import numpy as np\n",
    "plt.style.use('seaborn-whitegrid')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "result_dir = \"experiment_results/vanilla_fix_bug_compute_loss_encodev2\"\n",
    "pattern = r\".*\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "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>dataset</th>\n",
       "      <th>algo</th>\n",
       "      <th>use_contrastive</th>\n",
       "      <th>seed</th>\n",
       "      <th>agent_id</th>\n",
       "      <th>no_components</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>kmnist</td>\n",
       "      <td>modular</td>\n",
       "      <td>True</td>\n",
       "      <td>seed_4</td>\n",
       "      <td>agent_1</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>kmnist</td>\n",
       "      <td>modular</td>\n",
       "      <td>True</td>\n",
       "      <td>seed_4</td>\n",
       "      <td>agent_2</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>kmnist</td>\n",
       "      <td>modular</td>\n",
       "      <td>True</td>\n",
       "      <td>seed_4</td>\n",
       "      <td>agent_6</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>kmnist</td>\n",
       "      <td>modular</td>\n",
       "      <td>True</td>\n",
       "      <td>seed_4</td>\n",
       "      <td>agent_7</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>kmnist</td>\n",
       "      <td>modular</td>\n",
       "      <td>True</td>\n",
       "      <td>seed_4</td>\n",
       "      <td>agent_4</td>\n",
       "      <td>10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  dataset     algo  use_contrastive    seed agent_id  no_components\n",
       "0  kmnist  modular             True  seed_4  agent_1             10\n",
       "1  kmnist  modular             True  seed_4  agent_2              8\n",
       "2  kmnist  modular             True  seed_4  agent_6             10\n",
       "3  kmnist  modular             True  seed_4  agent_7              9\n",
       "4  kmnist  modular             True  seed_4  agent_4             10"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv(f\"{result_dir}_no_components.csv\")\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "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>dataset</th>\n",
       "      <th>algo</th>\n",
       "      <th>use_contrastive</th>\n",
       "      <th>no_components</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>fashionmnist</td>\n",
       "      <td>modular</td>\n",
       "      <td>False</td>\n",
       "      <td>7.015625</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>fashionmnist</td>\n",
       "      <td>modular</td>\n",
       "      <td>True</td>\n",
       "      <td>7.437500</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>kmnist</td>\n",
       "      <td>modular</td>\n",
       "      <td>False</td>\n",
       "      <td>8.765625</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>kmnist</td>\n",
       "      <td>modular</td>\n",
       "      <td>True</td>\n",
       "      <td>9.171875</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>mnist</td>\n",
       "      <td>modular</td>\n",
       "      <td>False</td>\n",
       "      <td>8.140625</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>mnist</td>\n",
       "      <td>modular</td>\n",
       "      <td>True</td>\n",
       "      <td>8.500000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        dataset     algo  use_contrastive  no_components\n",
       "0  fashionmnist  modular            False       7.015625\n",
       "1  fashionmnist  modular             True       7.437500\n",
       "2        kmnist  modular            False       8.765625\n",
       "3        kmnist  modular             True       9.171875\n",
       "4         mnist  modular            False       8.140625\n",
       "5         mnist  modular             True       8.500000"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "res = df.groupby(\n",
    "    [\"dataset\", \"algo\", \"use_contrastive\"])[\"no_components\"].mean().reset_index()\n",
    "res"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# create a bar plot\n",
    "datasets = res[\"dataset\"].unique()\n",
    "# for each dataset, plot two bars: use_contrastive=True and use_contrastive=False\n",
    "grouped = res.groupby(['dataset', 'use_contrastive'])['no_components'].mean().unstack()\n",
    "\n",
    "fig, ax = plt.subplots(figsize=(10, 5))\n",
    "\n",
    "colors = ['#3182bd', '#9ecae1']\n",
    "edge_colors=['#08519c', '#4292c6']\n",
    "\n",
    "bar_width = 0.2\n",
    "opacity = 0.8\n",
    "linewidth = 2\n",
    "\n",
    "\n",
    "# Plot each group of bars\n",
    "for i, (col_name, values) in enumerate(grouped.items()):\n",
    "    x = np.arange(len(values))\n",
    "    label = 'contrastive' if col_name else 'no contrastive'\n",
    "    ax.bar(x + i * bar_width, values, bar_width, alpha=opacity, color=colors[i], linewidth=linewidth, label=label,\n",
    "                           edgecolor=edge_colors[i],)\n",
    "\n",
    "\n",
    "ax.set_xlabel('Dataset', fontsize=14)\n",
    "ax.set_ylabel('No. components', fontsize=14)\n",
    "ax.set_title('Number of learned components',fontsize=16, fontweight='bold')\n",
    "\n",
    "ax.set_xticks(x)\n",
    "ax.set_xticklabels(datasets)\n",
    "\n",
    "ax.legend(frameon=True, loc='lower right')\n",
    "plt.xticks(rotation=0);\n",
    "\n",
    "plt.savefig(f\"Figures/vanilla/no_components.pdf\", bbox_inches=\"tight\");\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "shell",
   "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": "3.10.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}