Comparison.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "from CustomXGBoost import (XGBRegressor,\n",
    "                           XGBRegressorAdam,\n",
    "                           XGBRegressorRMS,\n",
    "                           XGBClassifier\n",
    "                          )\n",
    "import xgboost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.datasets import load_diabetes\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import mean_squared_error, accuracy_score, log_loss, mean_absolute_error\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import seaborn as sns;sns.set(style=\"whitegrid\")\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We'll use sklearn's diabetes dataset for evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = load_diabetes()\n",
    "dataset = pd.DataFrame(data.data, columns=data.feature_names)\n",
    "# split data into X and y\n",
    "X = dataset\n",
    "Y = data[\"target\"]\n",
    "\n",
    "test_size = 0.20\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, Y, test_size=test_size, random_state=7)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = dict() # to store results of each model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We'll use default values for parameters as given in XGBoost official Documentation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## XGBoost with Adam Optimizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0] validation_set-rmse : 70.7528930856603\t[0] train_set-rmse : 77.67252928913183\n",
      "[1] validation_set-rmse : 70.12902840400017\t[1] train_set-rmse : 76.65038564857271\n",
      "[2] validation_set-rmse : 69.35332686078266\t[2] train_set-rmse : 75.46255816470534\n",
      "[3] validation_set-rmse : 68.5897392807819\t[3] train_set-rmse : 74.1687074245387\n",
      "[4] validation_set-rmse : 67.82162488893127\t[4] train_set-rmse : 72.80644924629905\n",
      "[5] validation_set-rmse : 67.05220904617751\t[5] train_set-rmse : 71.39102371966652\n",
      "[6] validation_set-rmse : 66.25400762606795\t[6] train_set-rmse : 69.9462153179027\n",
      "[7] validation_set-rmse : 65.44717983935875\t[7] train_set-rmse : 68.48607144878196\n",
      "[8] validation_set-rmse : 64.64484663595937\t[8] train_set-rmse : 67.02473391975211\n",
      "[9] validation_set-rmse : 63.88218987773402\t[9] train_set-rmse : 65.56772580477039\n",
      "[10] validation_set-rmse : 63.17389444708086\t[10] train_set-rmse : 64.1207000837676\n",
      "[11] validation_set-rmse : 62.549635331263985\t[11] train_set-rmse : 62.6864757723445\n",
      "[12] validation_set-rmse : 61.99737730405226\t[12] train_set-rmse : 61.27127128161397\n",
      "[13] validation_set-rmse : 61.509435694032305\t[13] train_set-rmse : 59.88069226703248\n",
      "[14] validation_set-rmse : 61.08645882102075\t[14] train_set-rmse : 58.51787342426692\n",
      "[15] validation_set-rmse : 60.7102317190869\t[15] train_set-rmse : 57.18657771442372\n",
      "[16] validation_set-rmse : 60.3793603178282\t[16] train_set-rmse : 55.886328438971766\n",
      "[17] validation_set-rmse : 60.05227967883027\t[17] train_set-rmse : 54.6193384080856\n",
      "[18] validation_set-rmse : 59.74211586545251\t[18] train_set-rmse : 53.38677502760091\n",
      "[19] validation_set-rmse : 59.452795443288466\t[19] train_set-rmse : 52.18962212616742\n",
      "[20] validation_set-rmse : 59.18409694062034\t[20] train_set-rmse : 51.02909123484883\n",
      "[21] validation_set-rmse : 58.93605666510047\t[21] train_set-rmse : 49.90519409370983\n",
      "[22] validation_set-rmse : 58.714545990443476\t[22] train_set-rmse : 48.81636074339002\n",
      "[23] validation_set-rmse : 58.511804910576394\t[23] train_set-rmse : 47.76178321514528\n",
      "[24] validation_set-rmse : 58.333069585837485\t[24] train_set-rmse : 46.73825211141856\n",
      "[25] validation_set-rmse : 58.17958827227114\t[25] train_set-rmse : 45.74643088895146\n",
      "[26] validation_set-rmse : 58.03684938906901\t[26] train_set-rmse : 44.78679079108253\n",
      "[27] validation_set-rmse : 57.904095193043254\t[27] train_set-rmse : 43.858127345433864\n",
      "[28] validation_set-rmse : 57.78031670443018\t[28] train_set-rmse : 42.95659977633642\n",
      "[29] validation_set-rmse : 57.66646216183535\t[29] train_set-rmse : 42.082217151394744\n",
      "[30] validation_set-rmse : 57.555496790804376\t[30] train_set-rmse : 41.23967699812704\n",
      "[31] validation_set-rmse : 57.44859694153517\t[31] train_set-rmse : 40.42830615908872\n",
      "[32] validation_set-rmse : 57.344479584608735\t[32] train_set-rmse : 39.650328799304845\n",
      "[33] validation_set-rmse : 57.23976424342061\t[33] train_set-rmse : 38.90563268600624\n",
      "[34] validation_set-rmse : 57.14837025567875\t[34] train_set-rmse : 38.181150733371126\n",
      "[35] validation_set-rmse : 57.07067167232481\t[35] train_set-rmse : 37.47605669802872\n",
      "[36] validation_set-rmse : 57.00799993364779\t[36] train_set-rmse : 36.789970475936045\n",
      "[37] validation_set-rmse : 56.956102291866316\t[37] train_set-rmse : 36.127984072271566\n",
      "[38] validation_set-rmse : 56.914177901069124\t[38] train_set-rmse : 35.48909649662155\n",
      "[39] validation_set-rmse : 56.87808565580911\t[39] train_set-rmse : 34.87204914232287\n",
      "[40] validation_set-rmse : 56.85016627655774\t[40] train_set-rmse : 34.27688428465364\n",
      "[41] validation_set-rmse : 56.82668348390301\t[41] train_set-rmse : 33.70219228483329\n",
      "[42] validation_set-rmse : 56.815058710809566\t[42] train_set-rmse : 33.15044680319828\n",
      "[43] validation_set-rmse : 56.81270090309552\t[43] train_set-rmse : 32.616622675306616\n",
      "[44] validation_set-rmse : 56.81847480852833\t[44] train_set-rmse : 32.09971570612231\n",
      "[45] validation_set-rmse : 56.83203298360341\t[45] train_set-rmse : 31.59743140881979\n",
      "[46] validation_set-rmse : 56.85152006523447\t[46] train_set-rmse : 31.109363992700732\n",
      "[47] validation_set-rmse : 56.867137030231575\t[47] train_set-rmse : 30.628805294878127\n",
      "[48] validation_set-rmse : 56.882750445120664\t[48] train_set-rmse : 30.15635361964289\n",
      "[49] validation_set-rmse : 56.896765029448915\t[49] train_set-rmse : 29.691402997835123\n",
      "[50] validation_set-rmse : 56.91513146240945\t[50] train_set-rmse : 29.236848902183162\n",
      "[51] validation_set-rmse : 56.93477950542577\t[51] train_set-rmse : 28.79325624740551\n",
      "[52] validation_set-rmse : 56.95370016894687\t[52] train_set-rmse : 28.364064683412106\n",
      "[53] validation_set-rmse : 56.97184095185725\t[53] train_set-rmse : 27.94894965085329\n",
      "[54] validation_set-rmse : 56.98710657794945\t[54] train_set-rmse : 27.541777824962328\n",
      "[55] validation_set-rmse : 57.00343156227027\t[55] train_set-rmse : 27.142383066073492\n",
      "[56] validation_set-rmse : 57.02408664968978\t[56] train_set-rmse : 26.748675070791872\n",
      "[57] validation_set-rmse : 57.053124320118044\t[57] train_set-rmse : 26.366369467550058\n",
      "[58] validation_set-rmse : 57.088781157625746\t[58] train_set-rmse : 25.994387373089165\n",
      "[59] validation_set-rmse : 57.12036307879717\t[59] train_set-rmse : 25.633357016925917\n",
      "[60] validation_set-rmse : 57.14496007081077\t[60] train_set-rmse : 25.282249261022674\n",
      "[61] validation_set-rmse : 57.17942831820685\t[61] train_set-rmse : 24.939000510254953\n",
      "[62] validation_set-rmse : 57.216960208568594\t[62] train_set-rmse : 24.60350433977829\n",
      "[63] validation_set-rmse : 57.25449744659868\t[63] train_set-rmse : 24.27606837863743\n",
      "[64] validation_set-rmse : 57.28995470290382\t[64] train_set-rmse : 23.956469972336038\n",
      "[65] validation_set-rmse : 57.324873428622126\t[65] train_set-rmse : 23.6447479932092\n",
      "[66] validation_set-rmse : 57.35911446361627\t[66] train_set-rmse : 23.34044267564728\n",
      "[67] validation_set-rmse : 57.39100094326913\t[67] train_set-rmse : 23.04342088966213\n",
      "[68] validation_set-rmse : 57.41897208846221\t[68] train_set-rmse : 22.752869937835893\n",
      "[69] validation_set-rmse : 57.44371940403925\t[69] train_set-rmse : 22.468871572717553\n",
      "[70] validation_set-rmse : 57.46768446650976\t[70] train_set-rmse : 22.191655365980427\n",
      "[71] validation_set-rmse : 57.490900331507\t[71] train_set-rmse : 21.92160675747624\n",
      "[72] validation_set-rmse : 57.51194068067505\t[72] train_set-rmse : 21.658807019102618\n",
      "[73] validation_set-rmse : 57.53040653183714\t[73] train_set-rmse : 21.401835306448323\n",
      "[74] validation_set-rmse : 57.54681689044976\t[74] train_set-rmse : 21.150702169159043\n",
      "[75] validation_set-rmse : 57.560924789194985\t[75] train_set-rmse : 20.905702165770418\n",
      "[76] validation_set-rmse : 57.57308193599349\t[76] train_set-rmse : 20.666749910001776\n",
      "[77] validation_set-rmse : 57.58538503697808\t[77] train_set-rmse : 20.433672330319673\n",
      "[78] validation_set-rmse : 57.60215846558471\t[78] train_set-rmse : 20.204691176995745\n",
      "[79] validation_set-rmse : 57.62286790893358\t[79] train_set-rmse : 19.979810672441577\n",
      "[80] validation_set-rmse : 57.64164641197097\t[80] train_set-rmse : 19.75999844786836\n",
      "[81] validation_set-rmse : 57.662982598521516\t[81] train_set-rmse : 19.546655382641475\n",
      "[82] validation_set-rmse : 57.682338250950785\t[82] train_set-rmse : 19.33937228354989\n",
      "[83] validation_set-rmse : 57.710850551313214\t[83] train_set-rmse : 19.135119099362054\n",
      "[84] validation_set-rmse : 57.74280965849508\t[84] train_set-rmse : 18.934061297511015\n",
      "[85] validation_set-rmse : 57.78140647350806\t[85] train_set-rmse : 18.735774654288637\n",
      "[86] validation_set-rmse : 57.81730368468233\t[86] train_set-rmse : 18.541247254414458\n",
      "[87] validation_set-rmse : 57.849321324251505\t[87] train_set-rmse : 18.350797811886252\n",
      "[88] validation_set-rmse : 57.88220205870552\t[88] train_set-rmse : 18.16366262531718\n",
      "[89] validation_set-rmse : 57.91334324613627\t[89] train_set-rmse : 17.98026150184139\n",
      "[90] validation_set-rmse : 57.94131243489567\t[90] train_set-rmse : 17.80103408024871\n",
      "[91] validation_set-rmse : 57.965018615515945\t[91] train_set-rmse : 17.62625711841139\n",
      "[92] validation_set-rmse : 57.99016373470632\t[92] train_set-rmse : 17.45624530289148\n",
      "[93] validation_set-rmse : 58.01741857193171\t[93] train_set-rmse : 17.290938905411128\n",
      "[94] validation_set-rmse : 58.0473217571533\t[94] train_set-rmse : 17.12994959320826\n",
      "[95] validation_set-rmse : 58.07996710424634\t[95] train_set-rmse : 16.969431535433\n",
      "[96] validation_set-rmse : 58.110989641585384\t[96] train_set-rmse : 16.80068817610329\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[97] validation_set-rmse : 58.135152130251114\t[97] train_set-rmse : 16.626336558191394\n",
      "[98] validation_set-rmse : 58.15194523919698\t[98] train_set-rmse : 16.44596868289292\n",
      "[99] validation_set-rmse : 58.17059612151663\t[99] train_set-rmse : 16.26553783508625\n"
     ]
    }
   ],
   "source": [
    "custom = XGBRegressorAdam() ## Adam Optimizer is used in this implementation while doing gradient boosting\n",
    "custom.fit(X_train.values, y_train, eval_set = (X_test.values, y_test)) ## Check CustomXGBoost.py for implementation\n",
    "\n",
    "y_pred = custom.predict(X_test.values)\n",
    "results[\"Adam\"] = custom.history"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## XGBoost with RMSProp Optimizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0] validation_set-rmse : 66.97774624677706\t[0] train_set-rmse : 71.11615600163847\n",
      "[1] validation_set-rmse : 63.78491447411068\t[1] train_set-rmse : 66.06505405336955\n",
      "[2] validation_set-rmse : 61.94198001744035\t[2] train_set-rmse : 62.26835498176733\n",
      "[3] validation_set-rmse : 60.69542852579092\t[3] train_set-rmse : 59.1306916054687\n",
      "[4] validation_set-rmse : 60.35100355445515\t[4] train_set-rmse : 56.440578219003015\n",
      "[5] validation_set-rmse : 60.075541235250974\t[5] train_set-rmse : 54.113561692183715\n",
      "[6] validation_set-rmse : 59.38647609729904\t[6] train_set-rmse : 52.13431593205579\n",
      "[7] validation_set-rmse : 59.01586881922398\t[7] train_set-rmse : 50.271507378474084\n",
      "[8] validation_set-rmse : 58.66371134422615\t[8] train_set-rmse : 48.63312950537583\n",
      "[9] validation_set-rmse : 58.18842025791387\t[9] train_set-rmse : 47.153108176229765\n",
      "[10] validation_set-rmse : 58.05527704454894\t[10] train_set-rmse : 45.756142360692024\n",
      "[11] validation_set-rmse : 57.8256687368538\t[11] train_set-rmse : 44.34220872907417\n",
      "[12] validation_set-rmse : 57.578646719498565\t[12] train_set-rmse : 43.09685332597541\n",
      "[13] validation_set-rmse : 57.30765751478974\t[13] train_set-rmse : 41.923200168879774\n",
      "[14] validation_set-rmse : 57.381620249061115\t[14] train_set-rmse : 40.73099075628054\n",
      "[15] validation_set-rmse : 57.12716928111938\t[15] train_set-rmse : 39.63594229856563\n",
      "[16] validation_set-rmse : 56.762126224736\t[16] train_set-rmse : 38.528783826827826\n",
      "[17] validation_set-rmse : 56.97829869839805\t[17] train_set-rmse : 37.64866664257431\n",
      "[18] validation_set-rmse : 56.777442241555924\t[18] train_set-rmse : 36.79597104412123\n",
      "[19] validation_set-rmse : 56.67096388287034\t[19] train_set-rmse : 35.839412716201586\n",
      "[20] validation_set-rmse : 56.461696806704786\t[20] train_set-rmse : 35.10649344544482\n",
      "[21] validation_set-rmse : 56.466670503909704\t[21] train_set-rmse : 34.30960005500385\n",
      "[22] validation_set-rmse : 56.31345865503251\t[22] train_set-rmse : 33.41887753926461\n",
      "[23] validation_set-rmse : 56.40415565038645\t[23] train_set-rmse : 32.713550580881126\n",
      "[24] validation_set-rmse : 56.26288333191959\t[24] train_set-rmse : 31.996440222674227\n",
      "[25] validation_set-rmse : 56.358830912703084\t[25] train_set-rmse : 31.240911855733113\n",
      "[26] validation_set-rmse : 56.460846343096804\t[26] train_set-rmse : 30.623288539606147\n",
      "[27] validation_set-rmse : 56.263851008554084\t[27] train_set-rmse : 29.90721523634041\n",
      "[28] validation_set-rmse : 56.39623572576827\t[28] train_set-rmse : 29.310974553262366\n",
      "[29] validation_set-rmse : 56.31828718985569\t[29] train_set-rmse : 28.654824753277467\n",
      "[30] validation_set-rmse : 56.26895630202881\t[30] train_set-rmse : 28.041073331173617\n",
      "[31] validation_set-rmse : 56.2568908822404\t[31] train_set-rmse : 27.42200528042143\n",
      "[32] validation_set-rmse : 56.26237971380896\t[32] train_set-rmse : 26.835019787474724\n",
      "[33] validation_set-rmse : 56.25232325703132\t[33] train_set-rmse : 26.33079616047782\n",
      "[34] validation_set-rmse : 56.37085973967953\t[34] train_set-rmse : 25.791827745509792\n",
      "[35] validation_set-rmse : 56.49820058352158\t[35] train_set-rmse : 25.266063578377004\n",
      "[36] validation_set-rmse : 56.52787064762644\t[36] train_set-rmse : 24.79561990760905\n",
      "[37] validation_set-rmse : 56.465470426398596\t[37] train_set-rmse : 24.356533716808155\n",
      "[38] validation_set-rmse : 56.46824267449622\t[38] train_set-rmse : 23.909141555932113\n",
      "[39] validation_set-rmse : 56.4451511158998\t[39] train_set-rmse : 23.531940251254824\n",
      "[40] validation_set-rmse : 56.41646687713754\t[40] train_set-rmse : 23.15307920275868\n",
      "[41] validation_set-rmse : 56.45650011177105\t[41] train_set-rmse : 22.7161868065229\n",
      "[42] validation_set-rmse : 56.56982713456016\t[42] train_set-rmse : 22.329563139251476\n",
      "[43] validation_set-rmse : 56.60846827635513\t[43] train_set-rmse : 21.989074563010348\n",
      "[44] validation_set-rmse : 56.611583756725416\t[44] train_set-rmse : 21.616643696183026\n",
      "[45] validation_set-rmse : 56.53650298330179\t[45] train_set-rmse : 21.261404376929306\n",
      "[46] validation_set-rmse : 56.639364837662065\t[46] train_set-rmse : 20.896920228205015\n",
      "[47] validation_set-rmse : 56.65316808205064\t[47] train_set-rmse : 20.58012515210105\n",
      "[48] validation_set-rmse : 56.71097959289365\t[48] train_set-rmse : 20.232782896678028\n",
      "[49] validation_set-rmse : 56.743738798757754\t[49] train_set-rmse : 19.88186378703543\n",
      "[50] validation_set-rmse : 56.758137688930624\t[50] train_set-rmse : 19.589544748371647\n",
      "[51] validation_set-rmse : 56.74416697493383\t[51] train_set-rmse : 19.31033851211697\n",
      "[52] validation_set-rmse : 56.81781466419785\t[52] train_set-rmse : 18.980004700262487\n",
      "[53] validation_set-rmse : 56.870937380390494\t[53] train_set-rmse : 18.639814990255424\n",
      "[54] validation_set-rmse : 56.87289840749166\t[54] train_set-rmse : 18.378511151676236\n",
      "[55] validation_set-rmse : 56.891809825612114\t[55] train_set-rmse : 18.044148547643626\n",
      "[56] validation_set-rmse : 56.953312397257186\t[56] train_set-rmse : 17.749051764939967\n",
      "[57] validation_set-rmse : 56.995596826504894\t[57] train_set-rmse : 17.474514469887925\n",
      "[58] validation_set-rmse : 57.011422497176774\t[58] train_set-rmse : 17.227313391350044\n",
      "[59] validation_set-rmse : 56.99784790790164\t[59] train_set-rmse : 17.012135503017525\n",
      "[60] validation_set-rmse : 57.00998140839649\t[60] train_set-rmse : 16.78167999223981\n",
      "[61] validation_set-rmse : 56.97976367784766\t[61] train_set-rmse : 16.547333604571953\n",
      "[62] validation_set-rmse : 56.929984496713516\t[62] train_set-rmse : 16.32027736358286\n",
      "[63] validation_set-rmse : 56.96074031256534\t[63] train_set-rmse : 16.106560231254324\n",
      "[64] validation_set-rmse : 56.86551566902964\t[64] train_set-rmse : 15.866076904198605\n",
      "[65] validation_set-rmse : 56.86017878926962\t[65] train_set-rmse : 15.64452150535231\n",
      "[66] validation_set-rmse : 56.90795114058006\t[66] train_set-rmse : 15.441292385820395\n",
      "[67] validation_set-rmse : 56.88009232128105\t[67] train_set-rmse : 15.242869229131935\n",
      "[68] validation_set-rmse : 56.89092284717545\t[68] train_set-rmse : 15.008323956883975\n",
      "[69] validation_set-rmse : 56.919057899008976\t[69] train_set-rmse : 14.814473685208934\n",
      "[70] validation_set-rmse : 56.89652935665877\t[70] train_set-rmse : 14.593559914635897\n",
      "[71] validation_set-rmse : 56.92099419781018\t[71] train_set-rmse : 14.384483999750312\n",
      "[72] validation_set-rmse : 56.9483934927795\t[72] train_set-rmse : 14.21550185308213\n",
      "[73] validation_set-rmse : 56.991252215026044\t[73] train_set-rmse : 14.041757813104418\n",
      "[74] validation_set-rmse : 57.06279672951853\t[74] train_set-rmse : 13.879909981827035\n",
      "[75] validation_set-rmse : 57.12050014667471\t[75] train_set-rmse : 13.696376200061847\n",
      "[76] validation_set-rmse : 57.112891096648816\t[76] train_set-rmse : 13.548364944147073\n",
      "[77] validation_set-rmse : 57.112707642256716\t[77] train_set-rmse : 13.389664877883543\n",
      "[78] validation_set-rmse : 57.12777110182281\t[78] train_set-rmse : 13.199024723629218\n",
      "[79] validation_set-rmse : 57.142314321929234\t[79] train_set-rmse : 13.015901171632853\n",
      "[80] validation_set-rmse : 57.138152690260966\t[80] train_set-rmse : 12.84662255383639\n",
      "[81] validation_set-rmse : 57.158987216003375\t[81] train_set-rmse : 12.676573709492141\n",
      "[82] validation_set-rmse : 57.16271463359731\t[82] train_set-rmse : 12.514476996010847\n",
      "[83] validation_set-rmse : 57.16690039798333\t[83] train_set-rmse : 12.369098617848875\n",
      "[84] validation_set-rmse : 57.167644675122794\t[84] train_set-rmse : 12.215847200626673\n",
      "[85] validation_set-rmse : 57.22253187572458\t[85] train_set-rmse : 12.059544097492664\n",
      "[86] validation_set-rmse : 57.24099319793091\t[86] train_set-rmse : 11.8948995253456\n",
      "[87] validation_set-rmse : 57.23678836787264\t[87] train_set-rmse : 11.736683102458665\n",
      "[88] validation_set-rmse : 57.32522817173627\t[88] train_set-rmse : 11.600363530292954\n",
      "[89] validation_set-rmse : 57.35333678802823\t[89] train_set-rmse : 11.445742614328424\n",
      "[90] validation_set-rmse : 57.340193434941334\t[90] train_set-rmse : 11.300256156063991\n",
      "[91] validation_set-rmse : 57.37547700994674\t[91] train_set-rmse : 11.161764657672025\n",
      "[92] validation_set-rmse : 57.37691692691597\t[92] train_set-rmse : 11.029287501859388\n",
      "[93] validation_set-rmse : 57.38809646009161\t[93] train_set-rmse : 10.887174789509048\n",
      "[94] validation_set-rmse : 57.45402744993705\t[94] train_set-rmse : 10.725184478984108\n",
      "[95] validation_set-rmse : 57.503125863959674\t[95] train_set-rmse : 10.586629262534808\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[96] validation_set-rmse : 57.546796236593764\t[96] train_set-rmse : 10.447898685569573\n",
      "[97] validation_set-rmse : 57.52068850223238\t[97] train_set-rmse : 10.314120705650389\n",
      "[98] validation_set-rmse : 57.55860773770601\t[98] train_set-rmse : 10.174528534511182\n",
      "[99] validation_set-rmse : 57.60825860205378\t[99] train_set-rmse : 10.065238137402385\n"
     ]
    }
   ],
   "source": [
    "custom = XGBRegressorRMS()\n",
    "custom.fit(X_train.values, y_train, eval_set = (X_test.values, y_test))\n",
    "\n",
    "y_pred = custom.predict(X_test.values)\n",
    "results[\"RMSprop\"] = custom.history"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Original XGBoost "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0]\tvalidation_0-rmse:119.32756\n",
      "[1]\tvalidation_0-rmse:93.22636\n",
      "[2]\tvalidation_0-rmse:77.79602\n",
      "[3]\tvalidation_0-rmse:68.24764\n",
      "[4]\tvalidation_0-rmse:63.38055\n",
      "[5]\tvalidation_0-rmse:60.97445\n",
      "[6]\tvalidation_0-rmse:60.24472\n",
      "[7]\tvalidation_0-rmse:60.80915\n",
      "[8]\tvalidation_0-rmse:60.59282\n",
      "[9]\tvalidation_0-rmse:61.14302\n",
      "[10]\tvalidation_0-rmse:61.39846\n",
      "[11]\tvalidation_0-rmse:61.69669\n",
      "[12]\tvalidation_0-rmse:61.95347\n",
      "[13]\tvalidation_0-rmse:62.01958\n",
      "[14]\tvalidation_0-rmse:62.18780\n",
      "[15]\tvalidation_0-rmse:62.28210\n",
      "[16]\tvalidation_0-rmse:62.38843\n",
      "[17]\tvalidation_0-rmse:62.43783\n",
      "[18]\tvalidation_0-rmse:62.50838\n",
      "[19]\tvalidation_0-rmse:62.64378\n",
      "[20]\tvalidation_0-rmse:62.86266\n",
      "[21]\tvalidation_0-rmse:62.67214\n",
      "[22]\tvalidation_0-rmse:62.75052\n",
      "[23]\tvalidation_0-rmse:62.80328\n",
      "[24]\tvalidation_0-rmse:62.86892\n",
      "[25]\tvalidation_0-rmse:63.12924\n",
      "[26]\tvalidation_0-rmse:63.06793\n",
      "[27]\tvalidation_0-rmse:63.04744\n",
      "[28]\tvalidation_0-rmse:63.02758\n",
      "[29]\tvalidation_0-rmse:62.89526\n",
      "[30]\tvalidation_0-rmse:62.87672\n",
      "[31]\tvalidation_0-rmse:62.82021\n",
      "[32]\tvalidation_0-rmse:62.84444\n",
      "[33]\tvalidation_0-rmse:62.90360\n",
      "[34]\tvalidation_0-rmse:62.95918\n",
      "[35]\tvalidation_0-rmse:62.97465\n",
      "[36]\tvalidation_0-rmse:62.94654\n",
      "[37]\tvalidation_0-rmse:62.91347\n",
      "[38]\tvalidation_0-rmse:62.89030\n",
      "[39]\tvalidation_0-rmse:62.90730\n",
      "[40]\tvalidation_0-rmse:62.90269\n",
      "[41]\tvalidation_0-rmse:62.87709\n",
      "[42]\tvalidation_0-rmse:62.81004\n",
      "[43]\tvalidation_0-rmse:62.80897\n",
      "[44]\tvalidation_0-rmse:62.82931\n",
      "[45]\tvalidation_0-rmse:62.81844\n",
      "[46]\tvalidation_0-rmse:62.81035\n",
      "[47]\tvalidation_0-rmse:62.77665\n",
      "[48]\tvalidation_0-rmse:62.76542\n",
      "[49]\tvalidation_0-rmse:62.75137\n",
      "[50]\tvalidation_0-rmse:62.70380\n",
      "[51]\tvalidation_0-rmse:62.69337\n",
      "[52]\tvalidation_0-rmse:62.73114\n",
      "[53]\tvalidation_0-rmse:62.73658\n",
      "[54]\tvalidation_0-rmse:62.74466\n",
      "[55]\tvalidation_0-rmse:62.73385\n",
      "[56]\tvalidation_0-rmse:62.74187\n",
      "[57]\tvalidation_0-rmse:62.74771\n",
      "[58]\tvalidation_0-rmse:62.75997\n",
      "[59]\tvalidation_0-rmse:62.77365\n",
      "[60]\tvalidation_0-rmse:62.78655\n",
      "[61]\tvalidation_0-rmse:62.78333\n",
      "[62]\tvalidation_0-rmse:62.78059\n",
      "[63]\tvalidation_0-rmse:62.78622\n",
      "[64]\tvalidation_0-rmse:62.78748\n",
      "[65]\tvalidation_0-rmse:62.79037\n",
      "[66]\tvalidation_0-rmse:62.78830\n",
      "[67]\tvalidation_0-rmse:62.80218\n",
      "[68]\tvalidation_0-rmse:62.79469\n",
      "[69]\tvalidation_0-rmse:62.79469\n",
      "[70]\tvalidation_0-rmse:62.80472\n",
      "[71]\tvalidation_0-rmse:62.82616\n",
      "[72]\tvalidation_0-rmse:62.81472\n",
      "[73]\tvalidation_0-rmse:62.82496\n",
      "[74]\tvalidation_0-rmse:62.83067\n",
      "[75]\tvalidation_0-rmse:62.81585\n",
      "[76]\tvalidation_0-rmse:62.81643\n",
      "[77]\tvalidation_0-rmse:62.82286\n",
      "[78]\tvalidation_0-rmse:62.81864\n",
      "[79]\tvalidation_0-rmse:62.81200\n",
      "[80]\tvalidation_0-rmse:62.81341\n",
      "[81]\tvalidation_0-rmse:62.81264\n",
      "[82]\tvalidation_0-rmse:62.82072\n",
      "[83]\tvalidation_0-rmse:62.82040\n",
      "[84]\tvalidation_0-rmse:62.81860\n",
      "[85]\tvalidation_0-rmse:62.81997\n",
      "[86]\tvalidation_0-rmse:62.82180\n",
      "[87]\tvalidation_0-rmse:62.82156\n",
      "[88]\tvalidation_0-rmse:62.81746\n",
      "[89]\tvalidation_0-rmse:62.81950\n",
      "[90]\tvalidation_0-rmse:62.81885\n",
      "[91]\tvalidation_0-rmse:62.81327\n",
      "[92]\tvalidation_0-rmse:62.81378\n",
      "[93]\tvalidation_0-rmse:62.81720\n",
      "[94]\tvalidation_0-rmse:62.81294\n",
      "[95]\tvalidation_0-rmse:62.81543\n",
      "[96]\tvalidation_0-rmse:62.81676\n",
      "[97]\tvalidation_0-rmse:62.81976\n",
      "[98]\tvalidation_0-rmse:62.81667\n",
      "[99]\tvalidation_0-rmse:62.81472\n"
     ]
    }
   ],
   "source": [
    "import xgboost\n",
    "model = xgboost.XGBRegressor()\n",
    "model.fit(X_train, y_train, verbose=True, eval_metric=\"rmse\", eval_set=[(X_test, y_test)])\n",
    "\n",
    "y_pred = model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "eval_res = model.evals_result()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Comparing Performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2f63f64af48>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10, 8))\n",
    "sns.lineplot(y=eval_res[\"validation_0\"][\"rmse\"], x=range(len(eval_res[\"validation_0\"][\"rmse\"])), label=\"simple xgboost\")\n",
    "sns.lineplot(y = results['Adam'][\"test\"], x = range(len(results['Adam'][\"test\"])), label=\"xgboost Adam\")\n",
    "sns.lineplot(y = results['RMSprop'][\"test\"], x = range(len(results['RMSprop'][\"test\"])), label=\"xgboost RMSProp\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Comparing CustomXGBoost with Original XGBoost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0] validation_set-rmse : 62.20998923459462\t[0] train_set-rmse : 61.6388100479816\n",
      "[1] validation_set-rmse : 59.96235273267726\t[1] train_set-rmse : 49.56706463046613\n",
      "[2] validation_set-rmse : 59.13410037709241\t[2] train_set-rmse : 39.58608060958659\n",
      "[3] validation_set-rmse : 60.240603998508064\t[3] train_set-rmse : 32.39192242015271\n",
      "[4] validation_set-rmse : 60.46175609043559\t[4] train_set-rmse : 27.06104138044848\n",
      "[5] validation_set-rmse : 59.81222723613043\t[5] train_set-rmse : 23.495106513351974\n",
      "[6] validation_set-rmse : 60.79925413047281\t[6] train_set-rmse : 19.734204137001445\n",
      "[7] validation_set-rmse : 60.87938481389498\t[7] train_set-rmse : 17.269524083362114\n",
      "[8] validation_set-rmse : 61.410392172449775\t[8] train_set-rmse : 15.50432080738733\n",
      "[9] validation_set-rmse : 61.51134653998612\t[9] train_set-rmse : 13.851231511096923\n",
      "[10] validation_set-rmse : 61.61949656552175\t[10] train_set-rmse : 12.354215286802965\n",
      "[11] validation_set-rmse : 61.48882869823704\t[11] train_set-rmse : 11.396688059156261\n",
      "[12] validation_set-rmse : 61.94024949121112\t[12] train_set-rmse : 9.940800510378937\n",
      "[13] validation_set-rmse : 62.070637514506295\t[13] train_set-rmse : 8.95786454136467\n",
      "[14] validation_set-rmse : 62.3605647096008\t[14] train_set-rmse : 8.008229106890985\n",
      "[15] validation_set-rmse : 62.62745830226496\t[15] train_set-rmse : 7.237394115084987\n",
      "[16] validation_set-rmse : 62.52053324661032\t[16] train_set-rmse : 6.492540796482953\n",
      "[17] validation_set-rmse : 62.55914596983534\t[17] train_set-rmse : 6.364162058850732\n",
      "[18] validation_set-rmse : 62.567556866859846\t[18] train_set-rmse : 5.936738579066816\n",
      "[19] validation_set-rmse : 62.693955342075824\t[19] train_set-rmse : 5.369091974381867\n",
      "[20] validation_set-rmse : 62.686254898199195\t[20] train_set-rmse : 5.21457346402503\n",
      "[21] validation_set-rmse : 62.73542535339529\t[21] train_set-rmse : 5.091413573620784\n",
      "[22] validation_set-rmse : 62.809397233656554\t[22] train_set-rmse : 4.747177162057868\n",
      "[23] validation_set-rmse : 62.82609399101161\t[23] train_set-rmse : 4.4168273102255355\n",
      "[24] validation_set-rmse : 62.84174544417315\t[24] train_set-rmse : 4.367689094031079\n",
      "[25] validation_set-rmse : 62.975791297037645\t[25] train_set-rmse : 3.854558841248456\n",
      "[26] validation_set-rmse : 62.974199449893845\t[26] train_set-rmse : 3.5777454735811975\n",
      "[27] validation_set-rmse : 62.96791337329087\t[27] train_set-rmse : 3.3830753096226114\n",
      "[28] validation_set-rmse : 62.99715061924426\t[28] train_set-rmse : 3.151947996753259\n",
      "[29] validation_set-rmse : 63.008508658652985\t[29] train_set-rmse : 3.014909826058204\n",
      "[30] validation_set-rmse : 63.01257662978671\t[30] train_set-rmse : 2.8381076267070044\n",
      "[31] validation_set-rmse : 63.01261940010042\t[31] train_set-rmse : 2.565493706897404\n",
      "[32] validation_set-rmse : 62.967175059450454\t[32] train_set-rmse : 2.3409027680557206\n",
      "[33] validation_set-rmse : 63.00907584438185\t[33] train_set-rmse : 2.155451839715626\n",
      "[34] validation_set-rmse : 63.07223325755398\t[34] train_set-rmse : 2.0389235801737153\n",
      "[35] validation_set-rmse : 63.11268116412119\t[35] train_set-rmse : 1.848755716138401\n",
      "[36] validation_set-rmse : 63.0899453508511\t[36] train_set-rmse : 1.7407190060762077\n",
      "[37] validation_set-rmse : 63.119391524059154\t[37] train_set-rmse : 1.5771002088684425\n",
      "[38] validation_set-rmse : 63.1353990608258\t[38] train_set-rmse : 1.4790868926972973\n",
      "[39] validation_set-rmse : 63.12319499946759\t[39] train_set-rmse : 1.3668884740703136\n",
      "[40] validation_set-rmse : 63.129468144842896\t[40] train_set-rmse : 1.3353724736374983\n",
      "[41] validation_set-rmse : 63.13090460823058\t[41] train_set-rmse : 1.2532909867537432\n",
      "[42] validation_set-rmse : 63.13302830391134\t[42] train_set-rmse : 1.139756424856222\n",
      "[43] validation_set-rmse : 63.176774266008024\t[43] train_set-rmse : 0.9811266950590822\n",
      "[44] validation_set-rmse : 63.15724924184356\t[44] train_set-rmse : 0.8923975670482543\n",
      "[45] validation_set-rmse : 63.171993220588305\t[45] train_set-rmse : 0.7774194266439798\n",
      "[46] validation_set-rmse : 63.164930535605116\t[46] train_set-rmse : 0.6934647747027443\n",
      "[47] validation_set-rmse : 63.164315108269534\t[47] train_set-rmse : 0.6758168736860853\n",
      "[48] validation_set-rmse : 63.1634536412901\t[48] train_set-rmse : 0.6287183058050291\n",
      "[49] validation_set-rmse : 63.17480923069229\t[49] train_set-rmse : 0.582054720617954\n",
      "[50] validation_set-rmse : 63.17001146080733\t[50] train_set-rmse : 0.5568667223110207\n",
      "[51] validation_set-rmse : 63.175914208519124\t[51] train_set-rmse : 0.5099202375466357\n",
      "[52] validation_set-rmse : 63.17147643744025\t[52] train_set-rmse : 0.4599699680555324\n",
      "[53] validation_set-rmse : 63.17723760406996\t[53] train_set-rmse : 0.43885200415586156\n",
      "[54] validation_set-rmse : 63.18257559523469\t[54] train_set-rmse : 0.42045412499779633\n",
      "[55] validation_set-rmse : 63.18821309705404\t[55] train_set-rmse : 0.39192442656760235\n",
      "[56] validation_set-rmse : 63.18964572431803\t[56] train_set-rmse : 0.37206292903979193\n",
      "[57] validation_set-rmse : 63.20385964095918\t[57] train_set-rmse : 0.336998364062672\n",
      "[58] validation_set-rmse : 63.20583368583949\t[58] train_set-rmse : 0.3138702824667073\n",
      "[59] validation_set-rmse : 63.2123104172935\t[59] train_set-rmse : 0.2977648360524079\n",
      "[60] validation_set-rmse : 63.21611712329484\t[60] train_set-rmse : 0.28449523581486946\n",
      "[61] validation_set-rmse : 63.2168726836456\t[61] train_set-rmse : 0.2746227318357808\n",
      "[62] validation_set-rmse : 63.21865802211024\t[62] train_set-rmse : 0.26586525170241804\n",
      "[63] validation_set-rmse : 63.229522667852144\t[63] train_set-rmse : 0.2367285558593202\n",
      "[64] validation_set-rmse : 63.228453173305525\t[64] train_set-rmse : 0.21657667213116777\n",
      "[65] validation_set-rmse : 63.230817940436346\t[65] train_set-rmse : 0.1913073976095854\n",
      "[66] validation_set-rmse : 63.23140302402415\t[66] train_set-rmse : 0.17295825668648881\n",
      "[67] validation_set-rmse : 63.231804757966735\t[67] train_set-rmse : 0.16929973678000246\n",
      "[68] validation_set-rmse : 63.233484217560694\t[68] train_set-rmse : 0.15422475151102022\n",
      "[69] validation_set-rmse : 63.231491731099354\t[69] train_set-rmse : 0.14280068602772275\n",
      "[70] validation_set-rmse : 63.2305977773129\t[70] train_set-rmse : 0.13580439406302253\n",
      "[71] validation_set-rmse : 63.232263976122546\t[71] train_set-rmse : 0.13070794902477414\n",
      "[72] validation_set-rmse : 63.23199990670322\t[72] train_set-rmse : 0.1201185998632132\n",
      "[73] validation_set-rmse : 63.233329741006635\t[73] train_set-rmse : 0.11233331559897694\n",
      "[74] validation_set-rmse : 63.235667213620694\t[74] train_set-rmse : 0.09809661384661371\n",
      "[75] validation_set-rmse : 63.23776993802657\t[75] train_set-rmse : 0.08626185041453938\n",
      "[76] validation_set-rmse : 63.23686638295448\t[76] train_set-rmse : 0.0829784661751698\n",
      "[77] validation_set-rmse : 63.23709079893377\t[77] train_set-rmse : 0.07769923010992204\n",
      "[78] validation_set-rmse : 63.237296014349425\t[78] train_set-rmse : 0.07554742530795523\n",
      "[79] validation_set-rmse : 63.2364062931938\t[79] train_set-rmse : 0.07116527844087726\n",
      "[80] validation_set-rmse : 63.23457763463539\t[80] train_set-rmse : 0.0649810604169263\n",
      "[81] validation_set-rmse : 63.23513639300082\t[81] train_set-rmse : 0.058043862728061876\n",
      "[82] validation_set-rmse : 63.232933414892045\t[82] train_set-rmse : 0.05124747256435582\n",
      "[83] validation_set-rmse : 63.23224824640385\t[83] train_set-rmse : 0.04714820033155031\n",
      "[84] validation_set-rmse : 63.232457955098724\t[84] train_set-rmse : 0.04429664872632399\n",
      "[85] validation_set-rmse : 63.233068018171096\t[85] train_set-rmse : 0.040144207197842764\n",
      "[86] validation_set-rmse : 63.233462863607095\t[86] train_set-rmse : 0.03889415984444265\n",
      "[87] validation_set-rmse : 63.233080302363916\t[87] train_set-rmse : 0.03747981188946734\n",
      "[88] validation_set-rmse : 63.23361640831447\t[88] train_set-rmse : 0.03601558194433423\n",
      "[89] validation_set-rmse : 63.233308015407715\t[89] train_set-rmse : 0.031811026830430876\n",
      "[90] validation_set-rmse : 63.23282869530602\t[90] train_set-rmse : 0.028206505549672967\n",
      "[91] validation_set-rmse : 63.23352139846105\t[91] train_set-rmse : 0.02529247041752865\n",
      "[92] validation_set-rmse : 63.233195732722656\t[92] train_set-rmse : 0.024187327047256024\n",
      "[93] validation_set-rmse : 63.23358247020864\t[93] train_set-rmse : 0.021955011825585128\n",
      "[94] validation_set-rmse : 63.23401672901844\t[94] train_set-rmse : 0.02056349311905244\n",
      "[95] validation_set-rmse : 63.23426284247966\t[95] train_set-rmse : 0.019420301137460703\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[96] validation_set-rmse : 63.234406656400054\t[96] train_set-rmse : 0.019233010494799918\n",
      "[97] validation_set-rmse : 63.23431472447363\t[97] train_set-rmse : 0.018128788424545586\n",
      "[98] validation_set-rmse : 63.23453163139256\t[98] train_set-rmse : 0.01627900504896643\n",
      "[99] validation_set-rmse : 63.234574256479426\t[99] train_set-rmse : 0.015843570197845592\n"
     ]
    }
   ],
   "source": [
    "custom = XGBRegressor()\n",
    "custom.fit(X_train.values, y_train, eval_set = (X_test.values, y_test))\n",
    "\n",
    "y_pred = custom.predict(X_test.values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2f64a231ac8>"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10, 8))\n",
    "sns.lineplot(y=eval_res[\"validation_0\"][\"rmse\"], x=range(len(eval_res[\"validation_0\"][\"rmse\"])), label=\"Original XGBoost\")\n",
    "sns.lineplot(y=custom.history[\"test\"], x=range(len(custom.history[\"test\"])), label=\"Custom XGBoost\")"
   ]
  },
  {
   "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}