analysis/notebook_tob.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "DATA_DIR_PATH = \"../.data/measurements-public/\"\n",
    "LABEL = \"f18999999t20371443_blocks_tob10_aotonly\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Paired t-test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 90 pairs with min-block-num: 19001369 and max-block-num: 20354220\n"
     ]
    }
   ],
   "source": [
    "path = DATA_DIR_PATH + LABEL + \".csv\"\n",
    "df = pd.read_csv(path).drop_duplicates()\n",
    "df = df.groupby(['block_num', 'run_type']).mean().reset_index() # For any duplicated block_num, we take the mean of the values \n",
    "df_wide = df.pivot(index='block_num', columns='run_type', values='exe_time').dropna(axis=0).reset_index()\n",
    "print(f\"Found {len(df_wide)} pairs with min-block-num: {df_wide['block_num'].min()} and max-block-num: {df_wide['block_num'].max()}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "import scipy.stats as stats\n",
    "import numpy as np\n",
    "\n",
    "def perform_t_test(sample, alpha=0.05):\n",
    "    t_stat, p_value = stats.ttest_1samp(sample, 0)\n",
    "\n",
    "    print(f\"mean-ratio: {np.mean(sample):.2f}\")\n",
    "    print(f\"T-statistic: {t_stat:.2f}\")\n",
    "    print(f\"P-value: {p_value:.2f}\")\n",
    "\n",
    "    print(\"\\nVerdict:\")\n",
    "    if p_value < alpha:\n",
    "        print(f\"Reject the null hypothesis H0 at significance level {alpha}.\")\n",
    "        if t_stat < 0:\n",
    "            print(\"H1: The mean latency of the new method is significantly less than the old one.\")\n",
    "        else:\n",
    "            print(\"H2: The mean latency of the new method is significantly greater than the old one.\")\n",
    "    else:\n",
    "        print(\"Fail to reject the null hypothesis H0: There is no significant difference in latency.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### AOT vs Native"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mean-ratio: -0.07\n",
      "T-statistic: -11.62\n",
      "P-value: 0.00\n",
      "\n",
      "Verdict:\n",
      "Reject the null hypothesis H0 at significance level 0.05.\n",
      "H1: The mean latency of the new method is significantly less than the old one.\n"
     ]
    }
   ],
   "source": [
    "diff_ratios = (df_wide['aot'] - df_wide['native']) / df_wide['native']\n",
    "perform_t_test(diff_ratios)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### JIT vs Native"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# perform_t_test(df_wide['jit'], df_wide['native'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Histograms"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "\n",
    "# jit_diff = (df_wide['jit'] - df_wide['native']) \\\n",
    "#     .loc[lambda x: np.abs(stats.zscore(x)) < 2]\n",
    "aot_diff = diff_ratios\n",
    "\n",
    "# plt.hist(jit_diff, bins=100, alpha=0.5, label='jit-native')\n",
    "plt.hist(aot_diff, bins=20, alpha=0.5, label=r\"$\\frac{t_{aot}-t_{native}}{t_{native}}$\")\n",
    "\n",
    "plt.xlabel('Change in execution time relative to the native')\n",
    "plt.ylabel('Frequency')\n",
    "plt.title('Histogram of change in execution time')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  }
 ],
 "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": "3.11.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}