WMT19 Analysis Score Variations.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load packages \n",
    "import json\n",
    "import pandas as pd \n",
    "import numpy as np \n",
    "from tqdm import tqdm\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt \n",
    "%matplotlib inline\n",
    "from collections import defaultdict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load dataset for analysis\n",
    "with open('./wmt.jsonl', 'r') as json_file:\n",
    "    json_list = list(json_file)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "##Create multiple dictionaries to store values corresponding to a source\n",
    "# candidate_dict = {}\n",
    "\n",
    "# for json_str in tqdm(json_list):\n",
    "#     result = json.loads(json_str)\n",
    "#     if result['source'] not in candidate_dict:\n",
    "#         candidate_dict[result['source']] = []\n",
    "#     else:\n",
    "#         candidate_dict[result['source']].append((result['candidate'].lower(),result['raw_rating'], result['score'], result['system'], result['lang'] ))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Map and Create data with \"Source\" as a key\n",
    "# from collections import defaultdict\n",
    "# candidate_dict = defaultdict(list)\n",
    "\n",
    "# for json_str in tqdm(json_list):\n",
    "#     result = json.loads(json_str)\n",
    "#     candidate_dict[result['source']].append((result['candidate'].lower(),result['raw_rating'], result['score'], result['system'], result['lang'] ))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### EXACT SAME ANALYSIS (CHECK WITH LOWERCASE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████████████████████████████████████| 114072/114072 [00:01<00:00, 100767.65it/s]\n"
     ]
    }
   ],
   "source": [
    "out_dict = defaultdict(list)\n",
    "\n",
    "for json_str in tqdm(json_list):\n",
    "    result = json.loads(json_str)\n",
    "    out_dict[result['candidate'].lower()].append(result['raw_rating'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "112908"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(out_dict.keys())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "red_out_dict_1 = {k: v for k, v in out_dict.items() if len(v) > 1}\n",
    "red_out_dict_2 = {k: v for k, v in out_dict.items() if len(v) > 2}\n",
    "red_out_dict_3 = {k: v for k, v in out_dict.items() if len(v) > 3}\n",
    "red_out_dict_4 = {k: v for k, v in out_dict.items() if len(v) > 4}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Absolute numbers: \n",
      "1073 77 10 4\n",
      "Percentage numbers\n",
      "0.9503312431360046 0.06819711623622772 0.008856768342367236 0.003542707336946895\n"
     ]
    }
   ],
   "source": [
    "print(\"Absolute numbers: \")\n",
    "print(len(red_out_dict_1),len(red_out_dict_2),len(red_out_dict_3),len(red_out_dict_4))\n",
    "print(\"Percentage numbers\")\n",
    "print(len(red_out_dict_1)*100/len(out_dict.keys()),len(red_out_dict_2)*100/len(out_dict.keys()),len(red_out_dict_3)*100/len(out_dict.keys()),len(red_out_dict_4)*100/len(out_dict.keys()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "score_diff_list = []\n",
    "for key in red_out_dict_1.keys():\n",
    "    score_diff_list.append(max(red_out_dict_1[key])-min(red_out_dict_1[key]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x215e3bdaf70>"
      ]
     },
     "execution_count": 47,
     "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": [
    "plt.title(\"Plot for difference between minimum and maximum scores for same candidate translations\")\n",
    "plt.xlabel(\"Score Difference\")\n",
    "plt.ylabel(\"# of Sentence Instances\")\n",
    "sns.distplot(score_diff_list,hist=True,kde=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(20.246187919463654, 13.0, 21.639940748262095)"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.mean(score_diff_list), np.median(score_diff_list), np.std(score_diff_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}