plots-latex.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import csv\n",
    "import os\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "from pathlib import Path\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "from matplotlib.ticker import MaxNLocator\n",
    "from matplotlib.ticker import MultipleLocator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def sort_df_by_custom_order(df, column, custom_order):\n",
    "    df[column] = pd.Categorical(df[column], categories=custom_order, ordered=True)\n",
    "    df = df.sort_values(by=column)\n",
    "    return df\n",
    "\n",
    "def get_experiment_dataframe(experiment_csv_folder_path):\n",
    "    path = experiment_csv_folder_path\n",
    "    # or unix / linux / mac path\n",
    "\n",
    "    # Get the files from the path provided in the OP\n",
    "    files = Path(path).glob('*.csv')  # .rglob to get subdirectories\n",
    "\n",
    "    # for f in files:\n",
    "    #     print(f.stem)\n",
    "\n",
    "\n",
    "    dfs = list()\n",
    "    for f in files:\n",
    "        data = pd.read_csv(f, index_col=\"id\", header=0)\n",
    "        # .stem is method for pathlib objects to get the filename w/o the extension\n",
    "        data['file'] = f.stem\n",
    "        dfs.append(data)\n",
    "\n",
    "    exp_df = pd.concat(dfs, ignore_index=False)\n",
    "    return exp_df\n",
    "\n",
    "def output_latex_table_content(experiment_dataframe, output_file_name, configs):\n",
    "\n",
    "    #standard deviation is rounded to 3 decimal places\n",
    "    means = experiment_dataframe.groupby(['weight initialization policy'])[\n",
    "        'best_accuracy'].mean().round(3)\n",
    "    stds = experiment_dataframe.groupby(['weight initialization policy'])[\n",
    "        'best_accuracy'].std().round(3)\n",
    "\n",
    "    table_values = means.astype(str) + \"$\\pm$\" + stds.astype(str)\n",
    "\n",
    "    configs = ['hp-final', 'random', '2', '5', '10']\n",
    "    table_latex_string = \"\"\n",
    "    for config in configs:\n",
    "        table_latex_string += \"& \" + f\"{table_values['config']}\"\n",
    "        if config == 'random':\n",
    "            table_latex_string += \" \\\\\\\\ \\n\"\n",
    "\n",
    "    with open(output_file_name, \"w\") as text_file:\n",
    "        text_file.write(table_latex_string)\n",
    "\n",
    "\n",
    "def sort_mixed_array(array):\n",
    "    # Convert all elements to strings\n",
    "    str_array = [str(x) for x in array]\n",
    "\n",
    "    # Define a key function that sorts integers before strings and sorts by value within each type\n",
    "    def sort_key(s):\n",
    "        return (not s.isdigit(), s)\n",
    "\n",
    "    # Sort the array using the key function\n",
    "    return sorted(str_array, key=sort_key)\n",
    "\n",
    "def add_category_string_column(df):\n",
    "    df['weight initialization policy'] = df.apply(lambda row: str(row['hp_epoch']) if row['weight_initialization'] == 'hp-epoch' else row['weight_initialization'], axis=1)\n",
    "    return df\n",
    "\n",
    "def export_latex_table(df, grouping_attribute):\n",
    "\n",
    "    means = df.groupby(['weight initialization policy'])[\n",
    "        'best_accuracy'].mean().round(3)\n",
    "    stds = df.groupby(['weight initialization policy'])[\n",
    "        'best_accuracy'].std().round(3)\n",
    "    ttas = df.groupby(['weight initialization policy'])[\n",
    "        'target_accuracy_epoch'].mean()\n",
    "\n",
    "    table_values = means.astype(str) + \"$\\pm$\" + stds.astype(str) + \" & \" + ttas.astype(str)\n",
    "\n",
    "    table_latex_string = \"\"\n",
    "\n",
    "    for config in df[grouping_attribute].unique():\n",
    "        table_latex_string += config + \" & \" + table_values[config] + \"\\\\\\\\ \\n\"\n",
    "    print(table_latex_string)\n",
    "    pass\n",
    "\n",
    "def draw_plot_from_df(experiment_df, x, y, y_step=None, plot_title=\"\", file_name=\"\"):\n",
    "    sns.set_style(\"whitegrid\")\n",
    "\n",
    "    df_with_category_string_column = add_category_string_column(experiment_df)\n",
    "    custom_order = df_with_category_string_column['weight initialization policy'].unique()\n",
    "\n",
    "    # Separate the custom order into numeric strings and non-numeric strings\n",
    "    numeric_strings = [s for s in custom_order if s.isdigit()]\n",
    "    non_numeric_strings = [s for s in custom_order if not s.isdigit()]\n",
    "\n",
    "    # Sort the numeric strings numerically and the non-numeric strings lexicographically\n",
    "    numeric_strings = [str(i) for i in sorted(int(x) for x in numeric_strings)]\n",
    "    non_numeric_strings.sort()\n",
    "\n",
    "    # Concatenate the two sorted lists\n",
    "    sorted_order = numeric_strings + non_numeric_strings\n",
    "    # custom_order = sort_mixed_array(unique_settings)\n",
    "\n",
    "    # sorted_df = sort_df_by_custom_order(df_with_category_string_column, 'category_string', custom_order)\n",
    "    \n",
    "    plt.title(plot_title)\n",
    "\n",
    "    box_plot = sns.boxplot(data=df_with_category_string_column, x=x,\n",
    "                           y=y, \n",
    "                           order=sorted_order,\n",
    "                           hue=x,\n",
    "                           legend=False,\n",
    "                           showmeans=True,\n",
    "                           meanprops={\n",
    "                               \"marker\": \"s\", \"markerfacecolor\": \"white\", \"markeredgecolor\": \"blue\"}\n",
    "                           )\n",
    "\n",
    "    # box_plot.yaxis.set_major_locator(MaxNLocator(nbins='auto', steps=[1, 2, 5, 10]))\n",
    "    if y_step is not None:\n",
    "        box_plot.yaxis.set_major_locator(MultipleLocator(y_step))\n",
    " \n",
    "     # Set the size of the plot\n",
    "    sns.set_theme(rc={\"figure.figsize\":(10, 5)})\n",
    "    # Set global font parameters to match LaTeX\n",
    "    plt.rcParams['font.family'] = 'serif'\n",
    "    plt.rcParams['font.serif'] = ['Computer Modern Roman']\n",
    "    plt.rcParams['text.usetex'] = True\n",
    "    plt.rcParams['font.size'] = 12\n",
    "\n",
    "\n",
    "    plots_dir = os.getcwd() + '/plots'\n",
    "    if not os.path.exists(plots_dir):\n",
    "        os.makedirs(plots_dir)\n",
    "    \n",
    "    if file_name == \"\":\n",
    "        file_name = plot_title\n",
    "    box_plot.get_figure().savefig(os.path.join(plots_dir, file_name))\n",
    "    del box_plot\n",
    "    # return plot_df"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# EXP1: ResNet-18 on CIFAR-10, Target Test Accuracy: 90.00%\n",
    "These correspond to Figures 2-3 and Table 1 in the paper."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "experiment_df = get_experiment_dataframe(\"./PaperCSVs/res18-c10/\")\n",
    "\n",
    "draw_plot_from_df(experiment_df=experiment_df, x='weight initialization policy', y='best_accuracy', plot_title=\"\", file_name=\"fig2-res18-c10-accuracy.pdf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "draw_plot_from_df(experiment_df=experiment_df, x='weight initialization policy', y='target_accuracy_epoch', plot_title=\"\", file_name=\"fig3-res18-c10-tta.pdf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Random & 94.231$\\pm$0.098 & 29.125\\\\ \n",
      "5 & 94.289$\\pm$0.107 & 21.25\\\\ \n",
      "2 & 94.198$\\pm$0.108 & 24.0\\\\ \n",
      "HP Final & 94.316$\\pm$0.061 & 12.75\\\\ \n",
      "25 & 94.391$\\pm$0.061 & 14.375\\\\ \n",
      "20 & 94.351$\\pm$0.103 & 14.625\\\\ \n",
      "15 & 94.339$\\pm$0.166 & 18.5\\\\ \n",
      "10 & 94.319$\\pm$0.085 & 20.75\\\\ \n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Table 1 contents\n",
    "\n",
    "df_with_category_string_column = add_category_string_column(experiment_df)\n",
    "\n",
    "means = df_with_category_string_column.groupby(['weight initialization policy'])[\n",
    "    'best_accuracy'].mean().round(3)\n",
    "stds = df_with_category_string_column.groupby(['weight initialization policy'])[\n",
    "    'best_accuracy'].std().round(3)\n",
    "\n",
    "# output_latex_table_content(df_with_category_string_column, \"test.txt\", ['hp-final', 'random', '2', '5', '10'])\n",
    "\n",
    "\n",
    "export_latex_table(df_with_category_string_column, 'weight initialization policy')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# EXP2: ResNet-18 on CIFAR-100, Target Test Accuracy 75.00%\n",
    "These correspond to Figure 4 and Table 2. The TTA Figure is not included in the paper."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "experiment_df = get_experiment_dataframe(\"./PaperCSVs/res18-c100/\")\n",
    "\n",
    "draw_plot_from_df(experiment_df=experiment_df, x='weight initialization policy', y='best_accuracy', plot_title=\"\", file_name=\"fig4-res18-c100-accuracy.pdf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "draw_plot_from_df(experiment_df=experiment_df, x='weight initialization policy', y='target_accuracy_epoch', plot_title=\"\", file_name=\"fig4tta-res18-c100-tta.pdf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Random & 77.08$\\pm$0.299 & 122.375\\\\ \n",
      "15 & 77.112$\\pm$0.138 & 123.125\\\\ \n",
      "30 & 77.292$\\pm$0.181 & 124.5\\\\ \n",
      "HP Final & 76.947$\\pm$0.193 & 124.25\\\\ \n",
      "10 & 77.152$\\pm$0.219 & 123.25\\\\ \n",
      "5 & 77.044$\\pm$0.191 & 122.75\\\\ \n",
      "35 & 77.124$\\pm$0.287 & 124.25\\\\ \n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Table 2 contents\n",
    "\n",
    "df_with_category_string_column = add_category_string_column(experiment_df)\n",
    "\n",
    "means = df_with_category_string_column.groupby(['weight initialization policy'])[\n",
    "    'best_accuracy'].mean().round(3)\n",
    "stds = df_with_category_string_column.groupby(['weight initialization policy'])[\n",
    "    'best_accuracy'].std().round(3)\n",
    "\n",
    "# output_latex_table_content(df_with_category_string_column, \"test.txt\", ['hp-final', 'random', '2', '5', '10'])\n",
    "\n",
    "\n",
    "export_latex_table(df_with_category_string_column, 'weight initialization policy')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# EXP3: ResNet-18 on Tiny ImageNet, Target Test Accuracy 40.00%\n",
    "These correspond to Figure 5 and Table 3. The TTA Figure is not included in the paper."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "experiment_df = get_experiment_dataframe(\"./PaperCSVs/res18-tinyimagenet/\")\n",
    "\n",
    "draw_plot_from_df(experiment_df=experiment_df, x='weight initialization policy', y='best_accuracy', plot_title=\"\", file_name=\"fig5-res18-tiny-imagenet-accuracy.pdf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "draw_plot_from_df(experiment_df=experiment_df, x='weight initialization policy', y='target_accuracy_epoch', plot_title=\"\", file_name=\"fig5tta-res18-tiny-imagenet-tta.pdf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "35 & 44.362$\\pm$0.415 & 154.125\\\\ \n",
      "30 & 44.328$\\pm$0.479 & 153.875\\\\ \n",
      "10 & 44.085$\\pm$0.403 & 153.25\\\\ \n",
      "HP Final & 44.223$\\pm$0.275 & 154.25\\\\ \n",
      "5 & 44.062$\\pm$0.536 & 153.75\\\\ \n",
      "15 & 43.885$\\pm$0.245 & 153.5\\\\ \n",
      "Random & 44.428$\\pm$0.485 & 154.5\\\\ \n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Table 3 contents\n",
    "\n",
    "df_with_category_string_column = add_category_string_column(experiment_df)\n",
    "\n",
    "means = df_with_category_string_column.groupby(['weight initialization policy'])[\n",
    "    'best_accuracy'].mean().round(3)\n",
    "stds = df_with_category_string_column.groupby(['weight initialization policy'])[\n",
    "    'best_accuracy'].std().round(3)\n",
    "\n",
    "# output_latex_table_content(df_with_category_string_column, \"test.txt\", ['hp-final', 'random', '2', '5', '10'])\n",
    "\n",
    "\n",
    "export_latex_table(df_with_category_string_column, 'weight initialization policy')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# EXP4: ResNet-152 on CIFAR-100, Target Test Accuracy 75.00%\n",
    "These correspond to Figure 6 and Table 4. The TTA Figure is not included in the paper."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "experiment_df = get_experiment_dataframe(\"./PaperCSVs/res152-c100/\")\n",
    "\n",
    "draw_plot_from_df(experiment_df=experiment_df, x='weight initialization policy', y='best_accuracy', plot_title=\"\", file_name=\"fig6-res152-c100-accuracy.pdf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "draw_plot_from_df(experiment_df=experiment_df, x='weight initialization policy', y='target_accuracy_epoch', plot_title=\"\", file_name=\"fig6ttares152-c100-tta.pdf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Random & 80.372$\\pm$0.169 & 143.25\\\\ \n",
      "75 & 80.168$\\pm$0.107 & 143.25\\\\ \n",
      "HP Final & 80.072$\\pm$0.285 & 142.5\\\\ \n",
      "65 & 80.23$\\pm$0.189 & 144.0\\\\ \n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Table 4 contents\n",
    "\n",
    "df_with_category_string_column = add_category_string_column(experiment_df)\n",
    "\n",
    "means = df_with_category_string_column.groupby(['weight initialization policy'])[\n",
    "    'best_accuracy'].mean().round(3)\n",
    "stds = df_with_category_string_column.groupby(['weight initialization policy'])[\n",
    "    'best_accuracy'].std().round(3)\n",
    "\n",
    "# output_latex_table_content(df_with_category_string_column, \"test.txt\", ['hp-final', 'random', '2', '5', '10'])\n",
    "\n",
    "\n",
    "export_latex_table(df_with_category_string_column, 'weight initialization policy')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# EXP5: InceptionV3 on Food-101, Target Test Accuracy 70.00%\n",
    "These correspond to Figure 7-8 and Table 5."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "experiment_df = get_experiment_dataframe(\"./PaperCSVs/inceptionv3-food101/\")\n",
    "\n",
    "draw_plot_from_df(experiment_df=experiment_df, x='weight initialization policy', y='best_accuracy', plot_title=\"\", file_name=\"fig7-inception-v3-food101-accuracy.pdf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "draw_plot_from_df(experiment_df=experiment_df, x='weight initialization policy', y='target_accuracy_epoch', plot_title=\"\", file_name=\"fig8-inceptionv3-food101-tta.pdf\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "15 & 76.572$\\pm$0.155 & 20.5\\\\ \n",
      "10 & 75.935$\\pm$0.23 & 23.25\\\\ \n",
      "5 & 74.93$\\pm$0.115 & 28.75\\\\ \n",
      "HP Final & 76.894$\\pm$0.158 & 18.375\\\\ \n",
      "Random & 73.385$\\pm$1.073 & 33.875\\\\ \n",
      "\n"
     ]
    }
   ],
   "source": [
    "# Table 5 contents\n",
    "\n",
    "df_with_category_string_column = add_category_string_column(experiment_df)\n",
    "\n",
    "means = df_with_category_string_column.groupby(['weight initialization policy'])[\n",
    "    'best_accuracy'].mean().round(3)\n",
    "stds = df_with_category_string_column.groupby(['weight initialization policy'])[\n",
    "    'best_accuracy'].std().round(3)\n",
    "\n",
    "# output_latex_table_content(df_with_category_string_column, \"test.txt\", ['hp-final', 'random', '2', '5', '10'])\n",
    "\n",
    "\n",
    "export_latex_table(df_with_category_string_column, 'weight initialization policy')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "py39",
   "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": "3.9.18"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}