Importance_Sampling.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Importing libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 195,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import gzip\n",
    "import csv\n",
    "import json\n",
    "import datetime\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import scipy \n",
    "import dateutil\n",
    "from IPython.core.display import display, HTML\n",
    "import os\n",
    "\n",
    "import pylab\n",
    "\n",
    "from IPython.display import display\n",
    "from scipy import optimize\n",
    "\n",
    "import time\n",
    "#from colorama import init \n",
    "#from termcolor import colored"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Read CSV file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 284,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = pd.read_csv('example_dataset.csv',sep=',\\s+', delimiter=',', encoding=\"utf-8\", skipinitialspace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 285,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['nm_bedrooms', 'ceiling_fan', 'clothes_dryer', 'clothes_washer',\n",
       "       'cooking_range', 'cool_stpnt', 'cool_stpnt_ofst_mag',\n",
       "       'cool_stpnt_ofst_prd', 'dish_washer', 'ducts', 'nm_floors',\n",
       "       'foundation', 'garage', 'house_size', 'heating_fuel', 'heat_stpnt',\n",
       "       'heat_stpnt_ofst_mag', 'heat_stpnt_ofst_prd', 'hot_water_distribution',\n",
       "       'cool_efficiency', 'cool_type', 'heat_pump', 'heat_electricity',\n",
       "       'heat_natural_gas', 'heat_propane', 'is_heat_pump', 'infiltration',\n",
       "       'insulation_crawlspace', 'insulation_finished_basement',\n",
       "       'insulation_finished_roof', 'insulation_interzonal_floor',\n",
       "       'insulation_slab', 'insulation_unfinished_attic',\n",
       "       'insulation_unfinished_basement', 'insulation_wall', 'lighting',\n",
       "       'location', 'extra_refrigerator', 'freezer', 'gas_fireplace', 'hot_tub',\n",
       "       'pool', 'well_pump', 'neighbors', 'orientation', 'plug_loads',\n",
       "       'pool_heater', 'pool_pump', 'refrigerator', 'roof_material', 'vintage',\n",
       "       'water_heater', 'windows', 'total_site_energy'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 285,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# show all headers\n",
    "df.columns"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Select specific variables from data set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 286,
   "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>nm_bedrooms</th>\n",
       "      <th>ceiling_fan</th>\n",
       "      <th>clothes_dryer</th>\n",
       "      <th>cool_stpnt</th>\n",
       "      <th>cool_stpnt_ofst_mag</th>\n",
       "      <th>ducts</th>\n",
       "      <th>nm_floors</th>\n",
       "      <th>total_site_energy</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2.0</td>\n",
       "      <td>1 Fan, 120% Usage</td>\n",
       "      <td>Gas, 120% Usage</td>\n",
       "      <td>85F Variable Schedule</td>\n",
       "      <td>0F</td>\n",
       "      <td>10% Leakage, R-4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>48.623018</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3.0</td>\n",
       "      <td>1 Fan, 100% Usage</td>\n",
       "      <td>Gas, 100% Usage</td>\n",
       "      <td>85F Variable Schedule</td>\n",
       "      <td>0F</td>\n",
       "      <td>20% Leakage, R-4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>38.585635</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.0</td>\n",
       "      <td>2 Fans, 120% Usage</td>\n",
       "      <td>Gas, 120% Usage</td>\n",
       "      <td>72F Variable Schedule</td>\n",
       "      <td>13F</td>\n",
       "      <td>20% Leakage, R-4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>29.628763</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3.0</td>\n",
       "      <td>1 Fan, 80% Usage</td>\n",
       "      <td>None</td>\n",
       "      <td>72F Variable Schedule</td>\n",
       "      <td>13F</td>\n",
       "      <td>30% Leakage, Uninsulated</td>\n",
       "      <td>1.0</td>\n",
       "      <td>35.808531</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3.0</td>\n",
       "      <td>3 Fans, 120% Usage</td>\n",
       "      <td>Gas, 120% Usage</td>\n",
       "      <td>72F Variable Schedule</td>\n",
       "      <td>13F</td>\n",
       "      <td>20% Leakage, Uninsulated</td>\n",
       "      <td>2.0</td>\n",
       "      <td>70.214292</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   nm_bedrooms         ceiling_fan    clothes_dryer             cool_stpnt  \\\n",
       "0          2.0   1 Fan, 120% Usage  Gas, 120% Usage  85F Variable Schedule   \n",
       "1          3.0   1 Fan, 100% Usage  Gas, 100% Usage  85F Variable Schedule   \n",
       "2          2.0  2 Fans, 120% Usage  Gas, 120% Usage  72F Variable Schedule   \n",
       "3          3.0    1 Fan, 80% Usage             None  72F Variable Schedule   \n",
       "4          3.0  3 Fans, 120% Usage  Gas, 120% Usage  72F Variable Schedule   \n",
       "\n",
       "  cool_stpnt_ofst_mag                     ducts  nm_floors  total_site_energy  \n",
       "0                  0F          10% Leakage, R-4        1.0          48.623018  \n",
       "1                  0F          20% Leakage, R-4        1.0          38.585635  \n",
       "2                 13F          20% Leakage, R-4        1.0          29.628763  \n",
       "3                 13F  30% Leakage, Uninsulated        1.0          35.808531  \n",
       "4                 13F  20% Leakage, Uninsulated        2.0          70.214292  "
      ]
     },
     "execution_count": 286,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#clothes_washer\n",
    "\n",
    "df = df[[ 'nm_bedrooms', 'ceiling_fan', 'clothes_dryer', #'clothes_washer',\n",
    "         'cool_stpnt', 'cool_stpnt_ofst_mag', #'cooking_range', \n",
    "         'ducts','nm_floors',#'cool_stpnt_ofst_prd','dish_washer',\n",
    "         ##'garage', 'heat_stpnt',#'foundation','house_size','heating_fuel',\n",
    "         #'heat_stpnt_ofst_mag', 'heat_stpnt_ofst_prd', 'hot_water_distribution',\n",
    "         #'cool_efficiency', 'cool_type', 'heat_pump', 'heat_electricity',\n",
    "         #'heat_natural_gas', 'heat_propane', 'is_heat_pump', 'infiltration', \n",
    "         #'lighting','location', 'extra_refrigerator', 'freezer', 'gas_fireplace', 'hot_tub',\n",
    "         #'pool','well_pump', 'neighbors',\n",
    "         #'pool_heater', 'pool_pump', 'refrigerator', 'roof_material', 'vintage',\n",
    "         #'water_heater', 'windows',\n",
    "         'total_site_energy' ]]\n",
    "\n",
    "\n",
    "df.head(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Method to update distriubtion based on new mean and variance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 287,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "def newWeightOptimization(oldWeights,vals,newMean,numberOfMoments):\n",
    "    #recommended Number of moments = len(oldWeights)\n",
    "    numberOfMoments = len(oldWeights)\n",
    "    oldMean = 0\n",
    "    for i in range(len(oldWeights)):\n",
    "        oldMean += oldWeights[i]*vals[i]\n",
    "    \n",
    "    #calculate moments:\n",
    "    moments = np.zeros(numberOfMoments)\n",
    "    for i in range(numberOfMoments):\n",
    "        for j in range(len(oldWeights)):\n",
    "            moments[i]+=oldWeights[j]*((vals[j]-oldMean)**(i+2))\n",
    "    \n",
    "    bounds = []\n",
    "    for i in range(len(oldWeights)):\n",
    "        bounds.append((0,1))\n",
    "    bounds = tuple(bounds)\n",
    "    \n",
    "    constraints = [{'type':'eq', 'fun':lambda x:np.sum(x)-1},{'type':'eq','fun':lambda x: np.dot(x,vals)-newMean}]\n",
    "    \n",
    "    def _func(x):\n",
    "        \n",
    "        Xmatrix = np.zeros((numberOfMoments,len(oldWeights)))\n",
    "\n",
    "        for i in range(numberOfMoments):\n",
    "            for j in range(len(oldWeights)):\n",
    "                Xmatrix[i,j] = (vals[j]-newMean)**(i+2)\n",
    "                \n",
    "        Xmatrix /= np.linalg.norm(Xmatrix)\n",
    "        #Xmatrix*=100\n",
    "        objective = 0\n",
    "        for i in range(numberOfMoments):\n",
    "            objective += ((.5)*i)*(Xmatrix[i,:].dot(x))**2\n",
    "        \n",
    "        #optional regularization:\n",
    "        \n",
    "#         for i in range(len(x)):\n",
    "#             objective += newMean*x[i]*np.log(x[i])\n",
    "            \n",
    "        return objective\n",
    "    x0 = np.random.rand(len(oldWeights))\n",
    "    x0/=sum(x0)\n",
    "    res = optimize.minimize(_func,x0,bounds=bounds,constraints=constraints)\n",
    "    \n",
    "    return res.x\n",
    "                \n",
    "\n",
    "\n",
    "def weighted_avg_and_std(values, weights):\n",
    "    \"\"\"\n",
    "    Return the weighted average and standard deviation.\n",
    "\n",
    "    values, weights -- Numpy ndarrays with the same shape.\n",
    "    \"\"\"\n",
    "    average = np.average(values, weights=weights)\n",
    "    # Fast and numerically precise:\n",
    "    variance = np.average((values-average)**2, weights=weights)\n",
    "    return (average, np.sqrt(variance))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1- Adding sample distribution columns to dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 288,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Processing nm_bedrooms\n",
      "Processing ceiling_fan\n",
      "Processing clothes_dryer\n",
      "Processing cool_stpnt\n",
      "Processing cool_stpnt_ofst_mag\n",
      "Processing ducts\n",
      "Processing nm_floors\n",
      "Wall time: 3.52 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "# get lenght of dataframe\n",
    "N = float(len(df))\n",
    "\n",
    "# get distributions \n",
    "for c in df.columns:  \n",
    "        if c == 'total_site_energy' or c == 'weight':\n",
    "            continue\n",
    "        print(f\"Processing {c}\")\n",
    "        counts = df[c].value_counts()\n",
    "        \n",
    "        # find distributions and add distribution columns to dataframe\n",
    "        df[c+'_dist'] = df[c].map(lambda x: counts.loc[x]) / N\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 289,
   "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>nm_bedrooms</th>\n",
       "      <th>ceiling_fan</th>\n",
       "      <th>clothes_dryer</th>\n",
       "      <th>cool_stpnt</th>\n",
       "      <th>cool_stpnt_ofst_mag</th>\n",
       "      <th>ducts</th>\n",
       "      <th>nm_floors</th>\n",
       "      <th>total_site_energy</th>\n",
       "      <th>nm_bedrooms_dist</th>\n",
       "      <th>ceiling_fan_dist</th>\n",
       "      <th>clothes_dryer_dist</th>\n",
       "      <th>cool_stpnt_dist</th>\n",
       "      <th>cool_stpnt_ofst_mag_dist</th>\n",
       "      <th>ducts_dist</th>\n",
       "      <th>nm_floors_dist</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2.0</td>\n",
       "      <td>1 Fan, 120% Usage</td>\n",
       "      <td>Gas, 120% Usage</td>\n",
       "      <td>85F Variable Schedule</td>\n",
       "      <td>0F</td>\n",
       "      <td>10% Leakage, R-4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>48.623018</td>\n",
       "      <td>0.618445</td>\n",
       "      <td>0.050718</td>\n",
       "      <td>0.184454</td>\n",
       "      <td>0.091214</td>\n",
       "      <td>0.278091</td>\n",
       "      <td>0.171499</td>\n",
       "      <td>0.696351</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3.0</td>\n",
       "      <td>1 Fan, 100% Usage</td>\n",
       "      <td>Gas, 100% Usage</td>\n",
       "      <td>85F Variable Schedule</td>\n",
       "      <td>0F</td>\n",
       "      <td>20% Leakage, R-4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>38.585635</td>\n",
       "      <td>0.255266</td>\n",
       "      <td>0.101216</td>\n",
       "      <td>0.368159</td>\n",
       "      <td>0.091214</td>\n",
       "      <td>0.278091</td>\n",
       "      <td>0.309950</td>\n",
       "      <td>0.696351</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.0</td>\n",
       "      <td>2 Fans, 120% Usage</td>\n",
       "      <td>Gas, 120% Usage</td>\n",
       "      <td>72F Variable Schedule</td>\n",
       "      <td>13F</td>\n",
       "      <td>20% Leakage, R-4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>29.628763</td>\n",
       "      <td>0.618445</td>\n",
       "      <td>0.031815</td>\n",
       "      <td>0.184454</td>\n",
       "      <td>0.565260</td>\n",
       "      <td>0.424077</td>\n",
       "      <td>0.309950</td>\n",
       "      <td>0.696351</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3.0</td>\n",
       "      <td>1 Fan, 80% Usage</td>\n",
       "      <td>None</td>\n",
       "      <td>72F Variable Schedule</td>\n",
       "      <td>13F</td>\n",
       "      <td>30% Leakage, Uninsulated</td>\n",
       "      <td>1.0</td>\n",
       "      <td>35.808531</td>\n",
       "      <td>0.255266</td>\n",
       "      <td>0.051071</td>\n",
       "      <td>0.081960</td>\n",
       "      <td>0.565260</td>\n",
       "      <td>0.424077</td>\n",
       "      <td>0.076055</td>\n",
       "      <td>0.696351</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3.0</td>\n",
       "      <td>3 Fans, 120% Usage</td>\n",
       "      <td>Gas, 120% Usage</td>\n",
       "      <td>72F Variable Schedule</td>\n",
       "      <td>13F</td>\n",
       "      <td>20% Leakage, Uninsulated</td>\n",
       "      <td>2.0</td>\n",
       "      <td>70.214292</td>\n",
       "      <td>0.255266</td>\n",
       "      <td>0.044461</td>\n",
       "      <td>0.184454</td>\n",
       "      <td>0.565260</td>\n",
       "      <td>0.424077</td>\n",
       "      <td>0.132238</td>\n",
       "      <td>0.272627</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   nm_bedrooms         ceiling_fan    clothes_dryer             cool_stpnt  \\\n",
       "0          2.0   1 Fan, 120% Usage  Gas, 120% Usage  85F Variable Schedule   \n",
       "1          3.0   1 Fan, 100% Usage  Gas, 100% Usage  85F Variable Schedule   \n",
       "2          2.0  2 Fans, 120% Usage  Gas, 120% Usage  72F Variable Schedule   \n",
       "3          3.0    1 Fan, 80% Usage             None  72F Variable Schedule   \n",
       "4          3.0  3 Fans, 120% Usage  Gas, 120% Usage  72F Variable Schedule   \n",
       "\n",
       "  cool_stpnt_ofst_mag                     ducts  nm_floors  total_site_energy  \\\n",
       "0                  0F          10% Leakage, R-4        1.0          48.623018   \n",
       "1                  0F          20% Leakage, R-4        1.0          38.585635   \n",
       "2                 13F          20% Leakage, R-4        1.0          29.628763   \n",
       "3                 13F  30% Leakage, Uninsulated        1.0          35.808531   \n",
       "4                 13F  20% Leakage, Uninsulated        2.0          70.214292   \n",
       "\n",
       "   nm_bedrooms_dist  ceiling_fan_dist  clothes_dryer_dist  cool_stpnt_dist  \\\n",
       "0          0.618445          0.050718            0.184454         0.091214   \n",
       "1          0.255266          0.101216            0.368159         0.091214   \n",
       "2          0.618445          0.031815            0.184454         0.565260   \n",
       "3          0.255266          0.051071            0.081960         0.565260   \n",
       "4          0.255266          0.044461            0.184454         0.565260   \n",
       "\n",
       "   cool_stpnt_ofst_mag_dist  ducts_dist  nm_floors_dist  \n",
       "0                  0.278091    0.171499        0.696351  \n",
       "1                  0.278091    0.309950        0.696351  \n",
       "2                  0.424077    0.309950        0.696351  \n",
       "3                  0.424077    0.076055        0.696351  \n",
       "4                  0.424077    0.132238        0.272627  "
      ]
     },
     "execution_count": 289,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# resulting dataframe\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 2- Adding target distributions to data set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 290,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2.3933198202167976"
      ]
     },
     "execution_count": 290,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# decide what variable you want to change ( ex. set a new target distribution for number of bedrooms)\n",
    "# find what is the mean of this variable and what would be your target mean\n",
    "df.nm_bedrooms.mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 291,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*************Processing nm_bedrooms**************\n",
      "old_dist = {2.0: 0.6184454040715608, 3.0: 0.255265708997973, 1.0: 0.04498986516259804, 5.0: 0.012646514497223937, 4.0: 0.0651714109456244, 6.0: 0.0026438706265973387, 7.0: 0.0008372256984224906}\n",
      "new_dist = {2.0: 0.17107555347828546, 3.0: 0.23580367466058635, 1.0: 0.2021807155026118, 5.0: 0.18449692812499216, 4.0: 0.20644312823352426, 6.0: 0.0, 7.0: 2.7755575615628914e-16}\n",
      "*************Processing ceiling_fan**************\n",
      "old_dist = {'1 Fan, 120% Usage': 0.05071825152022561, '1 Fan, 100% Usage': 0.10121618048823477, '2 Fans, 120% Usage': 0.03181457654005464, '1 Fan, 80% Usage': 0.051070767603771924, '3 Fans, 120% Usage': 0.04446109103727858, 'None': 0.49167180752621836, '3 Fans, 80% Usage': 0.04472547809993831, '2 Fans, 80% Usage': 0.032078963602714373, '2 Fans, 100% Usage': 0.06358508856966599, '3 Fans, 100% Usage': 0.08865779501189741}\n",
      "new_dist = {'1 Fan, 120% Usage': 0.05071825152022561, '1 Fan, 100% Usage': 0.10121618048823477, '2 Fans, 120% Usage': 0.03181457654005464, '1 Fan, 80% Usage': 0.051070767603771924, '3 Fans, 120% Usage': 0.04446109103727858, 'None': 0.49167180752621836, '3 Fans, 80% Usage': 0.04472547809993831, '2 Fans, 80% Usage': 0.032078963602714373, '2 Fans, 100% Usage': 0.06358508856966599, '3 Fans, 100% Usage': 0.08865779501189741}\n",
      "*************Processing clothes_dryer**************\n",
      "old_dist = {'Gas, 120% Usage': 0.18445404071560764, 'Gas, 100% Usage': 0.36815898475367936, 'None': 0.08195998942451749, 'Electric, 80% Usage': 0.045166123204371196, 'Gas, 80% Usage': 0.18573191151846302, 'Electric, 100% Usage': 0.08989160130430951, 'Electric, 120% Usage': 0.04463734907905173}\n",
      "new_dist = {'Gas, 120% Usage': 0.18445404071560764, 'Gas, 100% Usage': 0.36815898475367936, 'None': 0.08195998942451749, 'Electric, 80% Usage': 0.045166123204371196, 'Gas, 80% Usage': 0.18573191151846302, 'Electric, 100% Usage': 0.08989160130430951, 'Electric, 120% Usage': 0.04463734907905173}\n",
      "*************Processing cool_stpnt**************\n",
      "old_dist = {'85F Variable Schedule': 0.09121353661760818, '72F Variable Schedule': 0.565259539966511, '78F Variable Schedule': 0.1878029435092976, '75F Variable Schedule': 0.14739578743280163, '82F Variable Schedule': 0.008328192473781616}\n",
      "new_dist = {'85F Variable Schedule': 0.09121353661760818, '72F Variable Schedule': 0.565259539966511, '78F Variable Schedule': 0.1878029435092976, '75F Variable Schedule': 0.14739578743280163, '82F Variable Schedule': 0.008328192473781616}\n",
      "*************Processing cool_stpnt_ofst_mag**************\n",
      "old_dist = {'0F': 0.27809112540759673, '13F': 0.4240768485062131, '7F': 0.13118004758967128, '10F': 0.07565876443112717, '3F': 0.06728650744690226, '4F': 0.019917158720366618, '6F': 0.003789547898122852}\n",
      "new_dist = {'0F': 0.27809112540759673, '13F': 0.4240768485062131, '7F': 0.13118004758967128, '10F': 0.07565876443112717, '3F': 0.06728650744690226, '4F': 0.019917158720366618, '6F': 0.003789547898122852}\n",
      "*************Processing ducts**************\n",
      "old_dist = {'10% Leakage, R-4': 0.1714990746452807, '20% Leakage, R-4': 0.30994976645809463, '30% Leakage, Uninsulated': 0.07605534502511677, '20% Leakage, Uninsulated': 0.13223759584031022, '30% Leakage, R-4': 0.17810875121177402, '20% Leakage, R-8': 0.009650127787080286, '30% Leakage, R-6': 0.0022472900326077376, '10% Leakage, Uninsulated': 0.07332334537763285, '10% Leakage, R-8': 0.005331805763637966, 'In Finished Space': 0.03000793161187979, '30% Leakage, R-8': 0.0055961928262977, '20% Leakage, R-6': 0.00387767691900943, '10% Leakage, R-6': 0.0021150965012778707}\n",
      "new_dist = {'10% Leakage, R-4': 0.1714990746452807, '20% Leakage, R-4': 0.30994976645809463, '30% Leakage, Uninsulated': 0.07605534502511677, '20% Leakage, Uninsulated': 0.13223759584031022, '30% Leakage, R-4': 0.17810875121177402, '20% Leakage, R-8': 0.009650127787080286, '30% Leakage, R-6': 0.0022472900326077376, '10% Leakage, Uninsulated': 0.07332334537763285, '10% Leakage, R-8': 0.005331805763637966, 'In Finished Space': 0.03000793161187979, '30% Leakage, R-8': 0.0055961928262977, '20% Leakage, R-6': 0.00387767691900943, '10% Leakage, R-6': 0.0021150965012778707}\n",
      "*************Processing nm_floors**************\n",
      "old_dist = {1.0: 0.6963514585352957, 2.0: 0.2726271261126289, 3.0: 0.03102141535207544}\n",
      "new_dist = {1.0: 0.6963514585352957, 2.0: 0.2726271261126289, 3.0: 0.03102141535207544}\n",
      "Wall time: 58.9 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# generate target distributions\n",
    "for c in df.columns:  \n",
    "    \n",
    "        if '_dist' not in c and '_target' not in c and 'site_energy' not in c and 'weight' not in c:\n",
    "            \n",
    "            print(f\"*************Processing {c}**************\")\n",
    "            # zip variables and distributions\n",
    "            z = dict(zip(df[c],df[c+'_dist']))\n",
    "            print (f\"old_dist = {z}\")\n",
    "\n",
    "            ### set inputs to weightConverion method\n",
    "            #set vals\n",
    "            vals = np.array(list(z.keys()))\n",
    "            #set oldDist\n",
    "            oldDist = np.array(list(z.values()))\n",
    "            #set targeted means of each variable distribution that is going to change\n",
    "            if c == 'nm_bedrooms':\n",
    "                newMean = 3\n",
    "            #for non changing variables the target distribution will be same as the original distribution\n",
    "            else: \n",
    "                print (f\"new_dist = {z}\")\n",
    "                df[c+'_target'] = df[c+'_dist']\n",
    "                continue\n",
    "                \n",
    "            numberOfMoments = 3\n",
    "\n",
    "            # find target distributions(weights) for each variable that have a new target mean\n",
    "            w = newWeightOptimization(oldDist,vals,newMean,numberOfMoments)\n",
    "            # output new distributions as a list\n",
    "            w = w.tolist()\n",
    "            \n",
    "            #zip weight and values\n",
    "            new_dist = dict(zip(vals,w))\n",
    "            print(f\"new_dist = {new_dist}\")\n",
    "\n",
    "            # layout target distribution in data frame\n",
    "            y=[]\n",
    "            for x in df[c]:\n",
    "                y.append(new_dist[x])\n",
    "\n",
    "            df[c+'_target'] = y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 292,
   "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>nm_bedrooms</th>\n",
       "      <th>ceiling_fan</th>\n",
       "      <th>clothes_dryer</th>\n",
       "      <th>cool_stpnt</th>\n",
       "      <th>cool_stpnt_ofst_mag</th>\n",
       "      <th>ducts</th>\n",
       "      <th>nm_floors</th>\n",
       "      <th>total_site_energy</th>\n",
       "      <th>nm_bedrooms_dist</th>\n",
       "      <th>ceiling_fan_dist</th>\n",
       "      <th>clothes_dryer_dist</th>\n",
       "      <th>cool_stpnt_dist</th>\n",
       "      <th>cool_stpnt_ofst_mag_dist</th>\n",
       "      <th>ducts_dist</th>\n",
       "      <th>nm_floors_dist</th>\n",
       "      <th>nm_bedrooms_target</th>\n",
       "      <th>ceiling_fan_target</th>\n",
       "      <th>clothes_dryer_target</th>\n",
       "      <th>cool_stpnt_target</th>\n",
       "      <th>cool_stpnt_ofst_mag_target</th>\n",
       "      <th>ducts_target</th>\n",
       "      <th>nm_floors_target</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2.0</td>\n",
       "      <td>1 Fan, 120% Usage</td>\n",
       "      <td>Gas, 120% Usage</td>\n",
       "      <td>85F Variable Schedule</td>\n",
       "      <td>0F</td>\n",
       "      <td>10% Leakage, R-4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>48.623018</td>\n",
       "      <td>0.618445</td>\n",
       "      <td>0.050718</td>\n",
       "      <td>0.184454</td>\n",
       "      <td>0.091214</td>\n",
       "      <td>0.278091</td>\n",
       "      <td>0.171499</td>\n",
       "      <td>0.696351</td>\n",
       "      <td>0.171076</td>\n",
       "      <td>0.050718</td>\n",
       "      <td>0.184454</td>\n",
       "      <td>0.091214</td>\n",
       "      <td>0.278091</td>\n",
       "      <td>0.171499</td>\n",
       "      <td>0.696351</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3.0</td>\n",
       "      <td>1 Fan, 100% Usage</td>\n",
       "      <td>Gas, 100% Usage</td>\n",
       "      <td>85F Variable Schedule</td>\n",
       "      <td>0F</td>\n",
       "      <td>20% Leakage, R-4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>38.585635</td>\n",
       "      <td>0.255266</td>\n",
       "      <td>0.101216</td>\n",
       "      <td>0.368159</td>\n",
       "      <td>0.091214</td>\n",
       "      <td>0.278091</td>\n",
       "      <td>0.309950</td>\n",
       "      <td>0.696351</td>\n",
       "      <td>0.235804</td>\n",
       "      <td>0.101216</td>\n",
       "      <td>0.368159</td>\n",
       "      <td>0.091214</td>\n",
       "      <td>0.278091</td>\n",
       "      <td>0.309950</td>\n",
       "      <td>0.696351</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.0</td>\n",
       "      <td>2 Fans, 120% Usage</td>\n",
       "      <td>Gas, 120% Usage</td>\n",
       "      <td>72F Variable Schedule</td>\n",
       "      <td>13F</td>\n",
       "      <td>20% Leakage, R-4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>29.628763</td>\n",
       "      <td>0.618445</td>\n",
       "      <td>0.031815</td>\n",
       "      <td>0.184454</td>\n",
       "      <td>0.565260</td>\n",
       "      <td>0.424077</td>\n",
       "      <td>0.309950</td>\n",
       "      <td>0.696351</td>\n",
       "      <td>0.171076</td>\n",
       "      <td>0.031815</td>\n",
       "      <td>0.184454</td>\n",
       "      <td>0.565260</td>\n",
       "      <td>0.424077</td>\n",
       "      <td>0.309950</td>\n",
       "      <td>0.696351</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3.0</td>\n",
       "      <td>1 Fan, 80% Usage</td>\n",
       "      <td>None</td>\n",
       "      <td>72F Variable Schedule</td>\n",
       "      <td>13F</td>\n",
       "      <td>30% Leakage, Uninsulated</td>\n",
       "      <td>1.0</td>\n",
       "      <td>35.808531</td>\n",
       "      <td>0.255266</td>\n",
       "      <td>0.051071</td>\n",
       "      <td>0.081960</td>\n",
       "      <td>0.565260</td>\n",
       "      <td>0.424077</td>\n",
       "      <td>0.076055</td>\n",
       "      <td>0.696351</td>\n",
       "      <td>0.235804</td>\n",
       "      <td>0.051071</td>\n",
       "      <td>0.081960</td>\n",
       "      <td>0.565260</td>\n",
       "      <td>0.424077</td>\n",
       "      <td>0.076055</td>\n",
       "      <td>0.696351</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3.0</td>\n",
       "      <td>3 Fans, 120% Usage</td>\n",
       "      <td>Gas, 120% Usage</td>\n",
       "      <td>72F Variable Schedule</td>\n",
       "      <td>13F</td>\n",
       "      <td>20% Leakage, Uninsulated</td>\n",
       "      <td>2.0</td>\n",
       "      <td>70.214292</td>\n",
       "      <td>0.255266</td>\n",
       "      <td>0.044461</td>\n",
       "      <td>0.184454</td>\n",
       "      <td>0.565260</td>\n",
       "      <td>0.424077</td>\n",
       "      <td>0.132238</td>\n",
       "      <td>0.272627</td>\n",
       "      <td>0.235804</td>\n",
       "      <td>0.044461</td>\n",
       "      <td>0.184454</td>\n",
       "      <td>0.565260</td>\n",
       "      <td>0.424077</td>\n",
       "      <td>0.132238</td>\n",
       "      <td>0.272627</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   nm_bedrooms         ceiling_fan    clothes_dryer             cool_stpnt  \\\n",
       "0          2.0   1 Fan, 120% Usage  Gas, 120% Usage  85F Variable Schedule   \n",
       "1          3.0   1 Fan, 100% Usage  Gas, 100% Usage  85F Variable Schedule   \n",
       "2          2.0  2 Fans, 120% Usage  Gas, 120% Usage  72F Variable Schedule   \n",
       "3          3.0    1 Fan, 80% Usage             None  72F Variable Schedule   \n",
       "4          3.0  3 Fans, 120% Usage  Gas, 120% Usage  72F Variable Schedule   \n",
       "\n",
       "  cool_stpnt_ofst_mag                     ducts  nm_floors  total_site_energy  \\\n",
       "0                  0F          10% Leakage, R-4        1.0          48.623018   \n",
       "1                  0F          20% Leakage, R-4        1.0          38.585635   \n",
       "2                 13F          20% Leakage, R-4        1.0          29.628763   \n",
       "3                 13F  30% Leakage, Uninsulated        1.0          35.808531   \n",
       "4                 13F  20% Leakage, Uninsulated        2.0          70.214292   \n",
       "\n",
       "   nm_bedrooms_dist  ceiling_fan_dist  clothes_dryer_dist  cool_stpnt_dist  \\\n",
       "0          0.618445          0.050718            0.184454         0.091214   \n",
       "1          0.255266          0.101216            0.368159         0.091214   \n",
       "2          0.618445          0.031815            0.184454         0.565260   \n",
       "3          0.255266          0.051071            0.081960         0.565260   \n",
       "4          0.255266          0.044461            0.184454         0.565260   \n",
       "\n",
       "   cool_stpnt_ofst_mag_dist  ducts_dist  nm_floors_dist  nm_bedrooms_target  \\\n",
       "0                  0.278091    0.171499        0.696351            0.171076   \n",
       "1                  0.278091    0.309950        0.696351            0.235804   \n",
       "2                  0.424077    0.309950        0.696351            0.171076   \n",
       "3                  0.424077    0.076055        0.696351            0.235804   \n",
       "4                  0.424077    0.132238        0.272627            0.235804   \n",
       "\n",
       "   ceiling_fan_target  clothes_dryer_target  cool_stpnt_target  \\\n",
       "0            0.050718              0.184454           0.091214   \n",
       "1            0.101216              0.368159           0.091214   \n",
       "2            0.031815              0.184454           0.565260   \n",
       "3            0.051071              0.081960           0.565260   \n",
       "4            0.044461              0.184454           0.565260   \n",
       "\n",
       "   cool_stpnt_ofst_mag_target  ducts_target  nm_floors_target  \n",
       "0                    0.278091      0.171499          0.696351  \n",
       "1                    0.278091      0.309950          0.696351  \n",
       "2                    0.424077      0.309950          0.696351  \n",
       "3                    0.424077      0.076055          0.696351  \n",
       "4                    0.424077      0.132238          0.272627  "
      ]
     },
     "execution_count": 292,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#resulting dataframe\n",
    "pd.set_option('display.max_columns', 100)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3- Applying weighting method"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 293,
   "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>nm_bedrooms</th>\n",
       "      <th>ceiling_fan</th>\n",
       "      <th>clothes_dryer</th>\n",
       "      <th>cool_stpnt</th>\n",
       "      <th>cool_stpnt_ofst_mag</th>\n",
       "      <th>ducts</th>\n",
       "      <th>nm_floors</th>\n",
       "      <th>total_site_energy</th>\n",
       "      <th>nm_bedrooms_dist</th>\n",
       "      <th>ceiling_fan_dist</th>\n",
       "      <th>clothes_dryer_dist</th>\n",
       "      <th>cool_stpnt_dist</th>\n",
       "      <th>cool_stpnt_ofst_mag_dist</th>\n",
       "      <th>ducts_dist</th>\n",
       "      <th>nm_floors_dist</th>\n",
       "      <th>nm_bedrooms_target</th>\n",
       "      <th>ceiling_fan_target</th>\n",
       "      <th>clothes_dryer_target</th>\n",
       "      <th>cool_stpnt_target</th>\n",
       "      <th>cool_stpnt_ofst_mag_target</th>\n",
       "      <th>ducts_target</th>\n",
       "      <th>nm_floors_target</th>\n",
       "      <th>weight</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2.0</td>\n",
       "      <td>1 Fan, 120% Usage</td>\n",
       "      <td>Gas, 120% Usage</td>\n",
       "      <td>85F Variable Schedule</td>\n",
       "      <td>0F</td>\n",
       "      <td>10% Leakage, R-4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>48.623018</td>\n",
       "      <td>0.618445</td>\n",
       "      <td>0.050718</td>\n",
       "      <td>0.184454</td>\n",
       "      <td>0.091214</td>\n",
       "      <td>0.278091</td>\n",
       "      <td>0.171499</td>\n",
       "      <td>0.696351</td>\n",
       "      <td>0.171076</td>\n",
       "      <td>0.050718</td>\n",
       "      <td>0.184454</td>\n",
       "      <td>0.091214</td>\n",
       "      <td>0.278091</td>\n",
       "      <td>0.171499</td>\n",
       "      <td>0.696351</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3.0</td>\n",
       "      <td>1 Fan, 100% Usage</td>\n",
       "      <td>Gas, 100% Usage</td>\n",
       "      <td>85F Variable Schedule</td>\n",
       "      <td>0F</td>\n",
       "      <td>20% Leakage, R-4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>38.585635</td>\n",
       "      <td>0.255266</td>\n",
       "      <td>0.101216</td>\n",
       "      <td>0.368159</td>\n",
       "      <td>0.091214</td>\n",
       "      <td>0.278091</td>\n",
       "      <td>0.309950</td>\n",
       "      <td>0.696351</td>\n",
       "      <td>0.235804</td>\n",
       "      <td>0.101216</td>\n",
       "      <td>0.368159</td>\n",
       "      <td>0.091214</td>\n",
       "      <td>0.278091</td>\n",
       "      <td>0.309950</td>\n",
       "      <td>0.696351</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.0</td>\n",
       "      <td>2 Fans, 120% Usage</td>\n",
       "      <td>Gas, 120% Usage</td>\n",
       "      <td>72F Variable Schedule</td>\n",
       "      <td>13F</td>\n",
       "      <td>20% Leakage, R-4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>29.628763</td>\n",
       "      <td>0.618445</td>\n",
       "      <td>0.031815</td>\n",
       "      <td>0.184454</td>\n",
       "      <td>0.565260</td>\n",
       "      <td>0.424077</td>\n",
       "      <td>0.309950</td>\n",
       "      <td>0.696351</td>\n",
       "      <td>0.171076</td>\n",
       "      <td>0.031815</td>\n",
       "      <td>0.184454</td>\n",
       "      <td>0.565260</td>\n",
       "      <td>0.424077</td>\n",
       "      <td>0.309950</td>\n",
       "      <td>0.696351</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3.0</td>\n",
       "      <td>1 Fan, 80% Usage</td>\n",
       "      <td>None</td>\n",
       "      <td>72F Variable Schedule</td>\n",
       "      <td>13F</td>\n",
       "      <td>30% Leakage, Uninsulated</td>\n",
       "      <td>1.0</td>\n",
       "      <td>35.808531</td>\n",
       "      <td>0.255266</td>\n",
       "      <td>0.051071</td>\n",
       "      <td>0.081960</td>\n",
       "      <td>0.565260</td>\n",
       "      <td>0.424077</td>\n",
       "      <td>0.076055</td>\n",
       "      <td>0.696351</td>\n",
       "      <td>0.235804</td>\n",
       "      <td>0.051071</td>\n",
       "      <td>0.081960</td>\n",
       "      <td>0.565260</td>\n",
       "      <td>0.424077</td>\n",
       "      <td>0.076055</td>\n",
       "      <td>0.696351</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3.0</td>\n",
       "      <td>3 Fans, 120% Usage</td>\n",
       "      <td>Gas, 120% Usage</td>\n",
       "      <td>72F Variable Schedule</td>\n",
       "      <td>13F</td>\n",
       "      <td>20% Leakage, Uninsulated</td>\n",
       "      <td>2.0</td>\n",
       "      <td>70.214292</td>\n",
       "      <td>0.255266</td>\n",
       "      <td>0.044461</td>\n",
       "      <td>0.184454</td>\n",
       "      <td>0.565260</td>\n",
       "      <td>0.424077</td>\n",
       "      <td>0.132238</td>\n",
       "      <td>0.272627</td>\n",
       "      <td>0.235804</td>\n",
       "      <td>0.044461</td>\n",
       "      <td>0.184454</td>\n",
       "      <td>0.565260</td>\n",
       "      <td>0.424077</td>\n",
       "      <td>0.132238</td>\n",
       "      <td>0.272627</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   nm_bedrooms         ceiling_fan    clothes_dryer             cool_stpnt  \\\n",
       "0          2.0   1 Fan, 120% Usage  Gas, 120% Usage  85F Variable Schedule   \n",
       "1          3.0   1 Fan, 100% Usage  Gas, 100% Usage  85F Variable Schedule   \n",
       "2          2.0  2 Fans, 120% Usage  Gas, 120% Usage  72F Variable Schedule   \n",
       "3          3.0    1 Fan, 80% Usage             None  72F Variable Schedule   \n",
       "4          3.0  3 Fans, 120% Usage  Gas, 120% Usage  72F Variable Schedule   \n",
       "\n",
       "  cool_stpnt_ofst_mag                     ducts  nm_floors  total_site_energy  \\\n",
       "0                  0F          10% Leakage, R-4        1.0          48.623018   \n",
       "1                  0F          20% Leakage, R-4        1.0          38.585635   \n",
       "2                 13F          20% Leakage, R-4        1.0          29.628763   \n",
       "3                 13F  30% Leakage, Uninsulated        1.0          35.808531   \n",
       "4                 13F  20% Leakage, Uninsulated        2.0          70.214292   \n",
       "\n",
       "   nm_bedrooms_dist  ceiling_fan_dist  clothes_dryer_dist  cool_stpnt_dist  \\\n",
       "0          0.618445          0.050718            0.184454         0.091214   \n",
       "1          0.255266          0.101216            0.368159         0.091214   \n",
       "2          0.618445          0.031815            0.184454         0.565260   \n",
       "3          0.255266          0.051071            0.081960         0.565260   \n",
       "4          0.255266          0.044461            0.184454         0.565260   \n",
       "\n",
       "   cool_stpnt_ofst_mag_dist  ducts_dist  nm_floors_dist  nm_bedrooms_target  \\\n",
       "0                  0.278091    0.171499        0.696351            0.171076   \n",
       "1                  0.278091    0.309950        0.696351            0.235804   \n",
       "2                  0.424077    0.309950        0.696351            0.171076   \n",
       "3                  0.424077    0.076055        0.696351            0.235804   \n",
       "4                  0.424077    0.132238        0.272627            0.235804   \n",
       "\n",
       "   ceiling_fan_target  clothes_dryer_target  cool_stpnt_target  \\\n",
       "0            0.050718              0.184454           0.091214   \n",
       "1            0.101216              0.368159           0.091214   \n",
       "2            0.031815              0.184454           0.565260   \n",
       "3            0.051071              0.081960           0.565260   \n",
       "4            0.044461              0.184454           0.565260   \n",
       "\n",
       "   cool_stpnt_ofst_mag_target  ducts_target  nm_floors_target  weight  \n",
       "0                    0.278091      0.171499          0.696351       1  \n",
       "1                    0.278091      0.309950          0.696351       1  \n",
       "2                    0.424077      0.309950          0.696351       1  \n",
       "3                    0.424077      0.076055          0.696351       1  \n",
       "4                    0.424077      0.132238          0.272627       1  "
      ]
     },
     "execution_count": 293,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[\"weight\"] = 1\n",
    "#initial dataframe\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## running weighting method"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 294,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "NORM DELTA_W = \n",
      "0.012452893011634002,0.004402001950130982,0.003352550332236883,0.0026073989620784325,0.0025110469880077328,0.0023224016056889716,0.0022532826682335905,0.002121687234863484,0.002057167574733328,0.0019447748613340041,0.0018839300658468725,0.0017840996346169397,0.0017265325450942846,0.0016368763558670047,0.0015824911926602774,0.0015016294052776515,0.001450402096287438,0.0013773114737645142,0.001329204951512615,0.0012630438798081041,0.0012179923161936682,0.001158039287130929,0.0011159512275980639,0.0010615761465537147,0.0010223408881139334,0.0009729880850004578,0.0009364815515671188,0.0008916580204807941,0.0008577475432694128,0.0008170140434059581,0.0007855621408693516,0.0007485261134528608,0.0007193934563351881,0.0006857032100153766,0.0006587509315189797,0.0006280907693623815,0.0006031822452972923,0.0005752683198374518,0.0005522705168060316,0.0005268472648462776,0.0005056317354932792,0.00048246878506364203,0.00046291237550290265,0.0004418018442831011,0.00042378716804583204,0.0004045412910600929,0.0003879570151567855,0.0003704060527057158,0.00035514703402970125,0.00033913742040253715,0.00032510472444065975,0.0003104974251103907,0.0002975982535262121,0.0002842673039814572,0.0002724148530123426,0.000260246056626044,0.00024935932429231825,0.00023824909000038939,0.00022825264676191399,0.00021810695003716585,0.00020893068470310912,0.00019966413755645514,0.00019124298788694938,0.00018277800547290954,0.00017505168092757785,0.0001673177339162075,0.00016023043637590053,0.00015316337957711288,0.00014666352652226172,0.00014020499539965178,0.0001342449489300554,0.00012834181663998686,0.00012287762083515623,0.00011748150927544569,0.0001124726376743561,0.00010753947679160959,0.00010294859120267704,9.843822146223078e-05,9.423094285857583e-05,9.010675635131045e-05,8.625144824894478e-05,8.248006442714494e-05,7.894762886697637e-05,7.549860134554396e-05,7.22622873927541e-05,6.910783863914505e-05,6.614306314532612e-05,6.325784423934683e-05,6.05420245035158e-05,5.7902897446483836e-05,5.5415295321330735e-05,5.3001135652675246e-05,5.072271259114135e-05,4.8514230303880835e-05,4.6427512797807155e-05,4.440708976076224e-05,4.2496044613238526e-05,4.06475868957658e-05,3.889750575330336e-05,3.720630941537397e-05,3.5603702000260186e-05,3.4056331059052256e-05,3.258882653802212e-05,3.1173001949339224e-05,2.9829257910236325e-05,2.853375653257784e-05,2.7303375039768964e-05,2.6117937637826564e-05,2.4991387880418863e-05,2.3906635331905945e-05,2.2875182386740737e-05,2.1882539328388503e-05,2.0938178593623832e-05,2.002980382613423e-05,1.9165200707981495e-05,1.833392373814017e-05,1.7542358190576106e-05,1.678162135254965e-05,1.60569369054469e-05,1.536074255886086e-05,1.4697299487750573e-05,1.4060161831221793e-05,1.3452794144833193e-05,1.2869695232854171e-05,1.2313671181188031e-05,1.1780020768979133e-05,1.1271006595379119e-05,1.078260547001532e-05,1.031663214424385e-05,9.869638634246989e-06,"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "maximum of delta sample_target distributions = 3.125758135646972e-05\n",
      "mean of delta sample_target distributions = 2.8856634250998773e-05\n",
      "Wall time: 18.9 s\n"
     ]
    }
   ],
   "source": [
    "\n",
    "%%time\n",
    "\n",
    "# initialize number of itterations\n",
    "count = 0\n",
    "itterations = []\n",
    "# initialize delta_w list \n",
    "delta_w_list = []\n",
    "\n",
    "# initialize delta_dist_vs_target\n",
    "delta_dist_target_list = []\n",
    "\n",
    "#initialize delta_output_list\n",
    "delta_output_list = []\n",
    "\n",
    "# check how the algorithm is progressing by traching delta weight\n",
    "print(\"NORM DELTA_W = \")\n",
    "\n",
    "# initialize original output\n",
    "output_old = df['total_site_energy'].copy()\n",
    "\n",
    "# Set Solving tolerance(a lower tolerance will take more run time but disttribution will get closer to the target)\n",
    "tol = 1e-5\n",
    "\n",
    "# Initialize delta weight\n",
    "delta_w_norm = 1.0/tol\n",
    "\n",
    "# add Weight column to data frame (note that this weight column is for the whole dataframe and not for each column)\n",
    "# all the Weights are intialized to be 1 as a start\n",
    "df[\"weight\"] = 1 \n",
    "\n",
    "\n",
    "# while the delta weigth norm is less than the tolerance\n",
    "while delta_w_norm > tol:\n",
    "\n",
    "    #get total weight\n",
    "    tW = df.weight.sum()\n",
    "    \n",
    "    # get w_old\n",
    "    w_old = df['weight'].copy()\n",
    "    \n",
    "\n",
    "    \n",
    "    # add number of itterations to a list\n",
    "    count += 1\n",
    "    itterations.append(count)  \n",
    "    \n",
    "    # change original distributions\n",
    "    # distribution = weight for a corresponing categroy of a variable / tW \n",
    "    for c in df.columns:\n",
    "\n",
    "        if 'total_site_energy' not in c and 'weight' not in c and '_dist' not in c and '_target' not in c and '_weight' not in c:\n",
    "\n",
    "            #print (f\"********processing {c}***********\")   \n",
    "            a = df[[c , 'weight']].groupby(c, as_index = False ).sum()\n",
    "            dict_a = dict(zip(a[c], a['weight']))          \n",
    "            df[c+'_dist'] = df[c].map(lambda x: dict_a[x]/tW)\n",
    "            \n",
    "            #get delta delta_dist_target\n",
    "            # add number of itterations to a list\n",
    "            delta_d_t = np.abs(df[c+'_target'] - df[c+'_dist'])\n",
    "            \n",
    "    \n",
    "    # apply reweighting\n",
    "    for c in df.columns:\n",
    "        \n",
    "        if 'total_site_energy' not in c and 'weight' not in c and '_dist' not in c and '_target' not in c and '_weight' not in c:\n",
    "            # find weights for each variable\n",
    "            # new weight = target distribution / sample distribution\n",
    "            # update the main weight column by multiplying it with the new weights\n",
    "            df['weight'] = (df[c+'_target'] / (df[c+'_dist']+1e-14)) * df['weight']\n",
    "    \n",
    "    \n",
    "    \n",
    "    # calculate the change total site energy output\n",
    "    df['total_site_energy_n'] = df['weight'] * df['total_site_energy']\n",
    "    \n",
    "    # initialize output_new\n",
    "    output_new = df['total_site_energy_n'].copy()\n",
    " \n",
    "\n",
    "    # find new weight of output; w_new\n",
    "    w_new = df['weight'].copy()\n",
    "    \n",
    "    ### find mean delta of output change\n",
    "    delta_output = np.abs(output_new - output_old)\n",
    "    delta_output_list.append(delta_output.mean())\n",
    "    \n",
    "    ### find delta weight\n",
    "    delta_w = np.abs(w_new - w_old)\n",
    "    # get delta weight norm\n",
    "    delta_w_norm = np.sqrt(np.sum(np.power(delta_w,2.0)))/float(len(delta_w))\n",
    "    # put delta_w in a list \n",
    "    delta_w_list.append(delta_w_norm)  \n",
    "    #print delta w norm\n",
    "    print(f\"{delta_w_norm},\", end='')\n",
    "    time.sleep(0.01)\n",
    "    \n",
    "    ###get maximum delta_dist_target\n",
    "    delta_d_t_max = delta_d_t.max()\n",
    "    #put delta_dist_target in a list\n",
    "    delta_dist_target_list.append(delta_d_t_max)\n",
    "    \n",
    "\n",
    "\n",
    "#### visualize results\n",
    "fig,ax = plt.subplots()\n",
    "fig,ax1 = plt.subplots()\n",
    "fig,ax2 = plt.subplots()\n",
    "\n",
    "# delta weight plot\n",
    "ax.plot(itterations, delta_w_list)\n",
    "ax.set_title('Delta weight norm')\n",
    "ax.set_ylabel('delta wieght norm')\n",
    "ax.set_xlabel('itteration')\n",
    "ax.set_yscale('log')\n",
    "#ax.set_ylim(top=0, bottom=None)\n",
    "#ax.set_ylim(ymin = 0)\n",
    "\n",
    "# delta smaple and target distributions\n",
    "ax1.plot(itterations[1:], delta_dist_target_list[1:])\n",
    "ax1.set_title('Change in samples-targets distributions')\n",
    "ax1.set_ylabel('delta dist_targets (max)')\n",
    "ax1.set_xlabel('itteration')\n",
    "ax1.set_yscale('log')\n",
    "#ax1.set_ylim(ymin = 0)\n",
    "\n",
    "# delta output max\n",
    "ax2.plot(itterations, delta_output_list)\n",
    "ax2.set_title('change in output')\n",
    "ax2.set_ylabel('delta output (mean)')\n",
    "ax2.set_xlabel('itteration')\n",
    "ax2.set_yscale('log')\n",
    "\n",
    "plt.show()\n",
    "\n",
    "#print info\n",
    "print (f\"maximum of delta sample_target distributions = {delta_d_t.max()}\")\n",
    "print (f\"mean of delta sample_target distributions = {delta_d_t.mean()}\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 298,
   "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>total_site_energy</th>\n",
       "      <th>total_site_energy_n</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>48.623018</td>\n",
       "      <td>20.224981</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>38.585635</td>\n",
       "      <td>54.126566</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>29.628763</td>\n",
       "      <td>8.800471</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>35.808531</td>\n",
       "      <td>40.250120</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>70.214292</td>\n",
       "      <td>48.227583</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>34.832279</td>\n",
       "      <td>10.067544</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>55.200869</td>\n",
       "      <td>61.184766</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>45.011835</td>\n",
       "      <td>50.367965</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>44.727490</td>\n",
       "      <td>12.813411</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>19.155384</td>\n",
       "      <td>93.652234</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   total_site_energy  total_site_energy_n\n",
       "0          48.623018            20.224981\n",
       "1          38.585635            54.126566\n",
       "2          29.628763             8.800471\n",
       "3          35.808531            40.250120\n",
       "4          70.214292            48.227583\n",
       "5          34.832279            10.067544\n",
       "6          55.200869            61.184766\n",
       "7          45.011835            50.367965\n",
       "8          44.727490            12.813411\n",
       "9          19.155384            93.652234"
      ]
     },
     "execution_count": 298,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# resulting change in outputs\n",
    "df_output = df[['total_site_energy','total_site_energy_n']]\n",
    "df_output.head(10)\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# END"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 297,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "59.93070864285551"
      ]
     },
     "execution_count": 297,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.abs(df_output.total_site_energy - df_output.total_site_energy_n).mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 209,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0    0.250348\n",
      "1    0.188350\n",
      "2    0.250348\n",
      "3    0.188350\n",
      "4    0.188350\n",
      "Name: nm_bedrooms_dist, dtype: float64\n",
      "0    0.25035\n",
      "1    0.18835\n",
      "2    0.25035\n",
      "3    0.18835\n",
      "4    0.18835\n",
      "Name: nm_bedrooms_target, dtype: float64\n"
     ]
    }
   ],
   "source": [
    "print(df['nm_bedrooms_dist'].head(5))\n",
    "print(df['nm_bedrooms_target'].head(5))\n",
    "#print(w_old[:5])\n",
    "#print(w_new[:5])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "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>nm_bedrooms</th>\n",
       "      <th>ceiling_fan</th>\n",
       "      <th>clothes_dryer</th>\n",
       "      <th>cool_stpnt</th>\n",
       "      <th>cool_stpnt_ofst_mag</th>\n",
       "      <th>ducts</th>\n",
       "      <th>nm_floors</th>\n",
       "      <th>garage</th>\n",
       "      <th>total_site_energy</th>\n",
       "      <th>nm_bedrooms_dist</th>\n",
       "      <th>ceiling_fan_dist</th>\n",
       "      <th>clothes_dryer_dist</th>\n",
       "      <th>cool_stpnt_dist</th>\n",
       "      <th>cool_stpnt_ofst_mag_dist</th>\n",
       "      <th>ducts_dist</th>\n",
       "      <th>nm_floors_dist</th>\n",
       "      <th>garage_dist</th>\n",
       "      <th>nm_bedrooms_target</th>\n",
       "      <th>ceiling_fan_target</th>\n",
       "      <th>clothes_dryer_target</th>\n",
       "      <th>cool_stpnt_target</th>\n",
       "      <th>cool_stpnt_ofst_mag_target</th>\n",
       "      <th>ducts_target</th>\n",
       "      <th>nm_floors_target</th>\n",
       "      <th>garage_target</th>\n",
       "      <th>weight</th>\n",
       "      <th>total_site_energy_n</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2.0</td>\n",
       "      <td>1 Fan, 120% Usage</td>\n",
       "      <td>Gas, 120% Usage</td>\n",
       "      <td>85F Variable Schedule</td>\n",
       "      <td>0F</td>\n",
       "      <td>10% Leakage, R-4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2 Car</td>\n",
       "      <td>48.623018</td>\n",
       "      <td>0.259811</td>\n",
       "      <td>0.050719</td>\n",
       "      <td>0.184455</td>\n",
       "      <td>0.091213</td>\n",
       "      <td>0.278091</td>\n",
       "      <td>0.171498</td>\n",
       "      <td>0.696348</td>\n",
       "      <td>0.130477</td>\n",
       "      <td>0.259814</td>\n",
       "      <td>0.050718</td>\n",
       "      <td>0.184454</td>\n",
       "      <td>0.091214</td>\n",
       "      <td>0.278091</td>\n",
       "      <td>0.171499</td>\n",
       "      <td>0.696351</td>\n",
       "      <td>0.130475</td>\n",
       "      <td>0.516234</td>\n",
       "      <td>9.824153e-114</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3.0</td>\n",
       "      <td>1 Fan, 100% Usage</td>\n",
       "      <td>Gas, 100% Usage</td>\n",
       "      <td>85F Variable Schedule</td>\n",
       "      <td>0F</td>\n",
       "      <td>20% Leakage, R-4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>None</td>\n",
       "      <td>38.585635</td>\n",
       "      <td>0.232265</td>\n",
       "      <td>0.101216</td>\n",
       "      <td>0.368158</td>\n",
       "      <td>0.091213</td>\n",
       "      <td>0.278091</td>\n",
       "      <td>0.309950</td>\n",
       "      <td>0.696348</td>\n",
       "      <td>0.663256</td>\n",
       "      <td>0.232265</td>\n",
       "      <td>0.101216</td>\n",
       "      <td>0.368159</td>\n",
       "      <td>0.091214</td>\n",
       "      <td>0.278091</td>\n",
       "      <td>0.309950</td>\n",
       "      <td>0.696351</td>\n",
       "      <td>0.663259</td>\n",
       "      <td>1.819754</td>\n",
       "      <td>8.145127e+98</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2.0</td>\n",
       "      <td>2 Fans, 120% Usage</td>\n",
       "      <td>Gas, 120% Usage</td>\n",
       "      <td>72F Variable Schedule</td>\n",
       "      <td>13F</td>\n",
       "      <td>20% Leakage, R-4</td>\n",
       "      <td>1.0</td>\n",
       "      <td>None</td>\n",
       "      <td>29.628763</td>\n",
       "      <td>0.259811</td>\n",
       "      <td>0.031815</td>\n",
       "      <td>0.184455</td>\n",
       "      <td>0.565268</td>\n",
       "      <td>0.424084</td>\n",
       "      <td>0.309950</td>\n",
       "      <td>0.696348</td>\n",
       "      <td>0.663256</td>\n",
       "      <td>0.259814</td>\n",
       "      <td>0.031815</td>\n",
       "      <td>0.184454</td>\n",
       "      <td>0.565260</td>\n",
       "      <td>0.424077</td>\n",
       "      <td>0.309950</td>\n",
       "      <td>0.696351</td>\n",
       "      <td>0.663259</td>\n",
       "      <td>0.634292</td>\n",
       "      <td>2.427118e-79</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3.0</td>\n",
       "      <td>1 Fan, 80% Usage</td>\n",
       "      <td>None</td>\n",
       "      <td>72F Variable Schedule</td>\n",
       "      <td>13F</td>\n",
       "      <td>30% Leakage, Uninsulated</td>\n",
       "      <td>1.0</td>\n",
       "      <td>None</td>\n",
       "      <td>35.808531</td>\n",
       "      <td>0.232265</td>\n",
       "      <td>0.051071</td>\n",
       "      <td>0.081960</td>\n",
       "      <td>0.565268</td>\n",
       "      <td>0.424084</td>\n",
       "      <td>0.076056</td>\n",
       "      <td>0.696348</td>\n",
       "      <td>0.663256</td>\n",
       "      <td>0.232265</td>\n",
       "      <td>0.051071</td>\n",
       "      <td>0.081960</td>\n",
       "      <td>0.565260</td>\n",
       "      <td>0.424077</td>\n",
       "      <td>0.076055</td>\n",
       "      <td>0.696351</td>\n",
       "      <td>0.663259</td>\n",
       "      <td>1.532705</td>\n",
       "      <td>1.031547e+70</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3.0</td>\n",
       "      <td>3 Fans, 120% Usage</td>\n",
       "      <td>Gas, 120% Usage</td>\n",
       "      <td>72F Variable Schedule</td>\n",
       "      <td>13F</td>\n",
       "      <td>20% Leakage, Uninsulated</td>\n",
       "      <td>2.0</td>\n",
       "      <td>None</td>\n",
       "      <td>70.214292</td>\n",
       "      <td>0.232265</td>\n",
       "      <td>0.044461</td>\n",
       "      <td>0.184455</td>\n",
       "      <td>0.565268</td>\n",
       "      <td>0.424084</td>\n",
       "      <td>0.132238</td>\n",
       "      <td>0.272630</td>\n",
       "      <td>0.663256</td>\n",
       "      <td>0.232265</td>\n",
       "      <td>0.044461</td>\n",
       "      <td>0.184454</td>\n",
       "      <td>0.565260</td>\n",
       "      <td>0.424077</td>\n",
       "      <td>0.132238</td>\n",
       "      <td>0.272627</td>\n",
       "      <td>0.663259</td>\n",
       "      <td>1.002560</td>\n",
       "      <td>1.036576e-01</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   nm_bedrooms         ceiling_fan    clothes_dryer             cool_stpnt  \\\n",
       "0          2.0   1 Fan, 120% Usage  Gas, 120% Usage  85F Variable Schedule   \n",
       "1          3.0   1 Fan, 100% Usage  Gas, 100% Usage  85F Variable Schedule   \n",
       "2          2.0  2 Fans, 120% Usage  Gas, 120% Usage  72F Variable Schedule   \n",
       "3          3.0    1 Fan, 80% Usage             None  72F Variable Schedule   \n",
       "4          3.0  3 Fans, 120% Usage  Gas, 120% Usage  72F Variable Schedule   \n",
       "\n",
       "  cool_stpnt_ofst_mag                     ducts  nm_floors garage  \\\n",
       "0                  0F          10% Leakage, R-4        1.0  2 Car   \n",
       "1                  0F          20% Leakage, R-4        1.0   None   \n",
       "2                 13F          20% Leakage, R-4        1.0   None   \n",
       "3                 13F  30% Leakage, Uninsulated        1.0   None   \n",
       "4                 13F  20% Leakage, Uninsulated        2.0   None   \n",
       "\n",
       "   total_site_energy  nm_bedrooms_dist  ceiling_fan_dist  clothes_dryer_dist  \\\n",
       "0          48.623018          0.259811          0.050719            0.184455   \n",
       "1          38.585635          0.232265          0.101216            0.368158   \n",
       "2          29.628763          0.259811          0.031815            0.184455   \n",
       "3          35.808531          0.232265          0.051071            0.081960   \n",
       "4          70.214292          0.232265          0.044461            0.184455   \n",
       "\n",
       "   cool_stpnt_dist  cool_stpnt_ofst_mag_dist  ducts_dist  nm_floors_dist  \\\n",
       "0         0.091213                  0.278091    0.171498        0.696348   \n",
       "1         0.091213                  0.278091    0.309950        0.696348   \n",
       "2         0.565268                  0.424084    0.309950        0.696348   \n",
       "3         0.565268                  0.424084    0.076056        0.696348   \n",
       "4         0.565268                  0.424084    0.132238        0.272630   \n",
       "\n",
       "   garage_dist  nm_bedrooms_target  ceiling_fan_target  clothes_dryer_target  \\\n",
       "0     0.130477            0.259814            0.050718              0.184454   \n",
       "1     0.663256            0.232265            0.101216              0.368159   \n",
       "2     0.663256            0.259814            0.031815              0.184454   \n",
       "3     0.663256            0.232265            0.051071              0.081960   \n",
       "4     0.663256            0.232265            0.044461              0.184454   \n",
       "\n",
       "   cool_stpnt_target  cool_stpnt_ofst_mag_target  ducts_target  \\\n",
       "0           0.091214                    0.278091      0.171499   \n",
       "1           0.091214                    0.278091      0.309950   \n",
       "2           0.565260                    0.424077      0.309950   \n",
       "3           0.565260                    0.424077      0.076055   \n",
       "4           0.565260                    0.424077      0.132238   \n",
       "\n",
       "   nm_floors_target  garage_target    weight  total_site_energy_n  \n",
       "0          0.696351       0.130475  0.516234        9.824153e-114  \n",
       "1          0.696351       0.663259  1.819754         8.145127e+98  \n",
       "2          0.696351       0.663259  0.634292         2.427118e-79  \n",
       "3          0.696351       0.663259  1.532705         1.031547e+70  \n",
       "4          0.272627       0.663259  1.002560         1.036576e-01  "
      ]
     },
     "execution_count": 191,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Visualize results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 192,
   "metadata": {},
   "outputs": [],
   "source": [
    "#len(itterations)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "#len(delta_w_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 539,
   "metadata": {},
   "outputs": [],
   "source": [
    "#del itterations[-1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 540,
   "metadata": {},
   "outputs": [],
   "source": [
    "#len(itterations)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 685,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "fig = plt.figure()\n",
    "ax = fig.add_subplot(2, 1, 1)\n",
    "\n",
    "ax.plot(itterations, delta_w_list, color='blue', lw=2)\n",
    "\n",
    "ax.set_yscale('log')\n",
    "\n",
    "pylab.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "fig = plt.figure()\n",
    "fig.suptitle('delta_dist_target')\n",
    "\n",
    "#fig, ax_lst = plt.subplots(2,2)\n",
    "ax = fig.add_subplot()\n",
    "\n",
    "ax.plot(itterations[1:], delta_dist_target[1:], color='blue', lw=2)\n",
    "\n",
    "ax.set_yscale('log')\n",
    "\n",
    "pylab.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 495,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[nan]\n",
      "[1]\n"
     ]
    }
   ],
   "source": [
    "print(delta_w_list)\n",
    "print(itterations)"
   ]
  },
  {
   "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}