homer_eda.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Exploratory Data Analysis of Homer Segementation Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.cm as cm\n",
    "# from kmodes import kmodes, kprototypes\n",
    "import scipy.stats as scs\n",
    "import seaborn as sns\n",
    "from homer_clean_data import remove_outliers\n",
    "sns.set(style=\"whitegrid\", color_codes=True)\n",
    "%matplotlib inline\n",
    "\n",
    "# df_users = pd.read_pickle('data/df_users.pkl')\n",
    "df = pd.read_pickle('data/df.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(794253, 16)"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# df_users.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# df_users['NumSims'].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# df_users.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# df_users.UserRole.value_counts(dropna=False)\n",
    "grps = df_users.UserRole.value_counts(dropna=False)\n",
    "pct = [x / float(sum(grps)) for x in grps]\n",
    "\n",
    "s1 = pd.Series(grps.values, index=grps.index, name='count')\n",
    "s2 = pd.Series(pct, index=grps.index, name='%total')\n",
    "s2 = s2.apply(lambda x: round(x*100, 2))\n",
    "df_grp = pd.concat([s1, s2], axis=1)\n",
    "df_grp"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# df_users.OrganizationType.value_counts(dropna=False)\n",
    "grps = df_users.OrganizationType.value_counts(dropna=False)\n",
    "pct = [x / float(sum(grps)) for x in grps]\n",
    "\n",
    "s1 = pd.Series(grps.values, index=grps.index, name='count')\n",
    "s2 = pd.Series(pct, index=grps.index, name='%total')\n",
    "s2 = s2.apply(lambda x: round(x*100, 2))\n",
    "df_grp = pd.concat([s1, s2], axis=1)\n",
    "df_grp"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# df_users.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "sims = df_users['NumSims'].values\n",
    "mu = sims.mean()\n",
    "std = sims.std()\n",
    "sims = [s for s in sims if s <= mu + (std*3) and s >= mu - (std*3)]\n",
    "\n",
    "fig = plt.figure(figsize=(8,6))\n",
    "ax = fig.add_subplot(111)\n",
    "\n",
    "ax.hist(sims, bins=50)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### EDA with clustered data (by user and simulation)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# df_users_clustered = pd.read_pickle('data/df_users_clustered.pkl')\n",
    "df_clustered = pd.read_pickle('data/df_clustered.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# df_users_clustered.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(794253, 17)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_clustered.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# df_users_clustered.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "bool_cols = ('ImportedWind', 'ImportedSolar', 'Sample')\n",
    "\n",
    "for col in bool_cols:\n",
    "    df_users_clustered[col] = df_users_clustered[col].astype(bool)\n",
    "    df_clustered[col] = df_clustered[col].astype(bool)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 794253 entries, 0 to 794252\n",
      "Data columns (total 17 columns):\n",
      "UserRole            794253 non-null object\n",
      "OrganizationType    794253 non-null object\n",
      "Latitude            794253 non-null float64\n",
      "Longitude           794253 non-null float64\n",
      "User                794253 non-null object\n",
      "Created             794253 non-null datetime64[ns]\n",
      "Sample              794253 non-null int64\n",
      "ImportedWind        794253 non-null int64\n",
      "ImportedSolar       794253 non-null int64\n",
      "MultiGenSearch      794253 non-null object\n",
      "MultiWindSearch     794253 non-null object\n",
      "MultiBatSearch      794253 non-null object\n",
      "MultiPvSearch       794253 non-null object\n",
      "MultiConSearch      794253 non-null object\n",
      "DefaultGenerator    794253 non-null object\n",
      "Country             794253 non-null object\n",
      "Cluster             794253 non-null int64\n",
      "dtypes: datetime64[ns](1), float64(2), int64(4), object(10)\n",
      "memory usage: 103.0+ MB\n"
     ]
    }
   ],
   "source": [
    "df_clustered.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# C1_u = df_users_clustered[df_users_clustered['Cluster'] == 1]\n",
    "# C2_u = df_users_clustered[df_users_clustered['Cluster'] == 2]\n",
    "# C3_u = df_users_clustered[df_users_clustered['Cluster'] == 3]\n",
    "# C4_u = df_users_clustered[df_users_clustered['Cluster'] == 4]\n",
    "\n",
    "C1_s = df_clustered[df_clustered['Cluster'] == 1]\n",
    "C2_s = df_clustered[df_clustered['Cluster'] == 2]\n",
    "C3_s = df_clustered[df_clustered['Cluster'] == 3]\n",
    "C4_s = df_clustered[df_clustered['Cluster'] == 4]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# User role distributions per cluster\n",
    "print('Cluster 1')\n",
    "print(C1_u.UserRole.value_counts(dropna=False))\n",
    "print('\\n')\n",
    "print('Cluster 2')\n",
    "print(C2_u.UserRole.value_counts(dropna=False))\n",
    "print('\\n')\n",
    "print('Cluster 3')\n",
    "print(C3_u.UserRole.value_counts(dropna=False))\n",
    "print('\\n')\n",
    "print('Cluster 4')\n",
    "print(C4_u.UserRole.value_counts(dropna=False))\n",
    "\n",
    "grp = df_users_clustered.groupby('Cluster')['UserRole'].count()\n",
    "grp.plot(kind='bar').set_ylabel('Number of Users');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "print(df_users_clustered.KM_Cluster.value_counts(dropna=False))\n",
    "grp = df_users_clustered.groupby('KM_Cluster')['NumSims'].mean()\n",
    "grp.plot(kind='bar').set_ylabel('Number of Simulations');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfgAAAFkCAYAAADMjClHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nCIIoQkjBEwRBEEQRQgqeIAiCIIqQ/wcH9EQOClhlhgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x149a79748>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(8,6))\n",
    "ax = fig.add_subplot(111)\n",
    "\n",
    "date_count = df_clustered.groupby(df_clustered['Created'].dt.date)['User'].count()\n",
    "ax.plot(date_count)\n",
    "ax.set_ylabel('Number of Simulations')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "794253"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sum(date_count)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# categorical vs categorical vs numeric\n",
    "agg = df_users_clustered.groupby(['KM_Cluster', 'UserRole'])['NumSims'].mean()\n",
    "print(agg)\n",
    "agg = agg.unstack(level='UserRole')\n",
    "print(agg)\n",
    "fig, ax = plt.subplots(1, 1, figsize=(10, 4))\n",
    "agg.plot(kind='bar', ax=ax).set_ylabel('Average # Simulations')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Scatter plot by category\n",
    "clusters = df_users_clustered['KM_Cluster'].unique()\n",
    "for clust in clusters:\n",
    "    plt.plot(df_users_clustered.loc[df_users_clustered.KM_Cluster==clust, 'NumChangedInputs'], \n",
    "             df_users_clustered.loc[df_users_clustered.KM_Cluster==clust, 'NumSims'], \n",
    "             linestyle='',\n",
    "             marker='o',\n",
    "             alpha=.7, label='Cluster={}'.format(clust))\n",
    "plt.xlabel('Number of User Changed Inputs')\n",
    "plt.ylabel('Number of Simulations')\n",
    "plt.legend(numpoints=1);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Heat-map\n",
    "agg = df_users_clustered.groupby(['KM_Cluster', 'UserRole'])['NumSims'].mean()\n",
    "ax = sns.heatmap(agg.unstack(level='UserRole'), annot=True)\n",
    "ax.set_title('Average number of simulations by cluster and user role')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Heat-map\n",
    "agg = df_clustered.groupby(['KM_Cluster', 'UserRole'])['UserRole'].count()\n",
    "ax = sns.heatmap(agg.unstack(level='UserRole'), annot=True)\n",
    "ax.set_title('Average number of simulations by cluster and user role')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "df_usa = df_users_clustered[df_users_clustered['Country'] == 'US']\n",
    "df_usa.head()\n",
    "\n",
    "grp = df_usa.groupby('KM_Cluster')['NumSims'].mean()\n",
    "grp.plot(kind='bar').set_ylabel('Number of Simulations');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# categorical vs categorical vs numeric\n",
    "agg = df_users_clustered.groupby(['KM_Cluster', 'OrganizationType'])['OrganizationType'].count()\n",
    "print(agg)\n",
    "agg = agg.unstack(level='OrganizationType')\n",
    "print(agg)\n",
    "fig, ax = plt.subplots(1, 1, figsize=(10, 4))\n",
    "agg.plot(kind='bar', ax=ax).set_ylabel('Number of Organizations (by type)')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "grp = df_users_clustered.groupby('KM_Cluster')['ImportedSolar'].mean()\n",
    "grp.plot(kind='bar').set_ylabel('Average # of users importing solar');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "grp = df_users_clustered.groupby('KM_Cluster')['ImportedWind'].mean()\n",
    "grp.plot(kind='bar').set_ylabel('Average # of users importing wind');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "df_users_clustered.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "grp = df_users_clustered.groupby('KM_Cluster')['GenCostMultiLines'].mean()\n",
    "grp.plot(kind='bar').set_ylabel('Average # of users with multiple lines in generator cost table');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "grp = df_users_clustered.groupby('KM_Cluster')['WindCostMultiLines'].mean()\n",
    "grp.plot(kind='bar').set_ylabel('Average # of users with multiple lines in wind cost table');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "grp = df_users_clustered.groupby('KM_Cluster')['BatCostMultiLines'].mean()\n",
    "grp.plot(kind='bar').set_ylabel('Average # of users with multiple lines in battery cost table');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "grp = df_users_clustered.groupby('KM_Cluster')['PvCostMultiLines'].mean()\n",
    "grp.plot(kind='bar').set_ylabel('Average # of users with multiple lines in PV cost table');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "grp = df_users_clustered.groupby('KM_Cluster')['ConCostMultiLines'].mean()\n",
    "grp.plot(kind='bar').set_ylabel('Average # of users with multiple lines in converter cost table');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "grp = df_users_clustered.groupby('KM_Cluster')['ElectricNotDefault'].mean()\n",
    "grp.plot(kind='bar').set_ylabel('Average # of users not using default electric peak');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "grp = df_users_clustered.groupby('KM_Cluster')['GeneratorNotDefault'].mean()\n",
    "grp.plot(kind='bar').set_ylabel('Average # of users not using default generator');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "grp = df_users_clustered.groupby('KM_Cluster')['GenCapCost'].mean()\n",
    "grp.plot(kind='bar').set_ylabel('Average # of users providing generator capital cost');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "grp = df_users_clustered.groupby('KM_Cluster')['BatCapCost'].mean()\n",
    "grp.plot(kind='bar').set_ylabel('Average # of users providing battery capital cost');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "grp = df_users_clustered.groupby('KM_Cluster')['WindCapCost'].mean()\n",
    "grp.plot(kind='bar').set_ylabel('Average # of users providing turbine capital cost');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "grp = df_users_clustered.groupby('KM_Cluster')['PvCapCost'].mean()\n",
    "grp.plot(kind='bar').set_ylabel('Average # of users providing PV capital cost');"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# np.round(cluster3[['NumSims', 'DaysSinceFirst']].describe(), 2).T[['count','mean', 'std', 'min', 'max']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "sns.stripplot(x='KM_Cluster', y='NumSims', data=df_users_clustered, jitter=True);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "sns.barplot(x='UserRole', y='NumChangedInputs', hue='KM_Cluster', data=df_users_clustered);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "sns.swarmplot(x='KM_Cluster', y='NumSims', data=df_users_clustered);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "sns.boxplot(x='KM_Cluster', y='NumSims', hue='UserRole', data=df_users_clustered);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "sns.countplot(x='KM_Cluster', hue='UserRole', data=df_users_clustered, palette=\"Greens_d\");"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "df_users_clustered['NumSims'].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "sns.pointplot(x='KM_Cluster', y='NumSims', hue='UserRole', data=df_users_clustered_clean);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "f, ax = plt.subplots(figsize=(7, 3))\n",
    "sns.countplot(y='KM_Cluster', data=df_users_clustered, color=\"c\");"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "print('Cluster 1')\n",
    "print(C1_s.MultiGenSearch.value_counts(dropna=False))\n",
    "print('\\n')\n",
    "print('Cluster 2')\n",
    "print(C2_s.MultiGenSearch.value_counts(dropna=False))\n",
    "print('\\n')\n",
    "print('Cluster 3')\n",
    "print(C3_s.MultiGenSearch.value_counts(dropna=False))\n",
    "print('\\n')\n",
    "print('Cluster 4')\n",
    "print(C4_s.MultiGenSearch.value_counts(dropna=False))\n",
    "print('\\n')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "grps = C1_s.MultiGenSearch.value_counts(dropna=False)\n",
    "pct = [x / float(sum(grps)) for x in grps]\n",
    "\n",
    "s1 = pd.Series(grps.values, index=grps.index, name='count')\n",
    "s2 = pd.Series(pct, index=grps.index, name='%total')\n",
    "s2 = s2.apply(lambda x: round(x*100, 2))\n",
    "df_grp = pd.concat([s1, s2], axis=1)\n",
    "df_grp"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def by_cluster(col):\n",
    "    frames = []\n",
    "    clusters = [C1_s, C2_s, C3_s, C4_s]\n",
    "    for idx, cluster in enumerate(clusters):\n",
    "        grps = cluster[col].value_counts(dropna=False)\n",
    "        pct = [x / float(sum(grps)) for x in grps]\n",
    "\n",
    "        s1 = pd.Series(grps.values, index=grps.index, name='Count')\n",
    "        s2 = pd.Series(pct, index=grps.index, name='Total%')\n",
    "        s2 = s2.apply(lambda x: round(x*100, 2))\n",
    "        \n",
    "        df_grp = pd.concat([s1, s2], axis=1)\n",
    "        df_grp =df_grp.reindex(['True', 'False', 'NA'])\n",
    "        frames.append(df_grp)\n",
    "        \n",
    "    df_combo = pd.concat(frames, axis=1)\n",
    "    vals = df_combo.values\n",
    "\n",
    "    l1 = ['Cluster 1', 'Cluster 1', 'Cluster 2', 'Cluster 2', 'Cluster 3', 'Cluster 3', 'Cluster 4', 'Cluster 4']\n",
    "    l2 = ['Count', 'Total%', 'Count', 'Total%', 'Count', 'Total%', 'Count', 'Total%']\n",
    "\n",
    "    tuples = list(zip(l1, l2))\n",
    "    index = pd.MultiIndex.from_tuples(tuples, names=[col, ''])\n",
    "\n",
    "    dff = pd.DataFrame(vals, index=['True', 'False', 'NA'], columns=index)\n",
    "    return dff\n",
    "\n",
    "dff = by_cluster('MultiBatSearch')\n",
    "dff"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "df_grp.sort_index(ascending=False)\n",
    "df_grp.reindex(['NA', 'False', 'True'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def user_counts(df):\n",
    "    grps = df.UserRole.value_counts(dropna=False)\n",
    "    pct = [x / float(sum(grps)) for x in grps]\n",
    "    s1 = pd.Series(grps.values, index=grps.index, name='Count')\n",
    "    s2 = pd.Series(pct, index=grps.index, name='Total%')\n",
    "    s2 = s2.apply(lambda x: round(x*100, 2))\n",
    "    df_grp = pd.concat([s1, s2], axis=1)\n",
    "    return df_grp\n",
    "\n",
    "print(user_counts(C2_s))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def org_counts(df):\n",
    "    grps = df.OrganizationType.value_counts(dropna=False)\n",
    "    pct = [x / float(sum(grps)) for x in grps]\n",
    "\n",
    "    s1 = pd.Series(grps.values, index=grps.index, name='Count')\n",
    "    s2 = pd.Series(pct, index=grps.index, name='Total%')\n",
    "    s2 = s2.apply(lambda x: round(x*100, 2))\n",
    "    df_grp = pd.concat([s1, s2], axis=1)\n",
    "    return df_grp\n",
    "\n",
    "print(org_counts(C2_s))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "df_clustered.describe().loc[['mean', 'std']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "df_clustered.Cluster.value_counts(dropna=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "C1 = df_clustered[df_clustered['Cluster'] == 1]\n",
    "C2 = df_clustered[df_clustered['Cluster'] == 2]\n",
    "C3 = df_clustered[df_clustered['Cluster'] == 3]\n",
    "C4 = df_clustered[df_clustered['Cluster'] == 4]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "pt = pd.pivot_table(df_clustered, values=['Created'], index=['User'], columns=['Cluster'], aggfunc='count', fill_value=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3831"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cluster_df = pd.DataFrame(pt.iloc[:, 0].index)\n",
    "cluster_df['Cluster 1'] = pt.iloc[:, 0].values\n",
    "cluster_df['Cluster 2'] = pt.iloc[:, 1].values\n",
    "cluster_df['Cluster 3'] = pt.iloc[:, 2].values\n",
    "cluster_df['Cluster 4'] = pt.iloc[:, 3].values\n",
    "cluster_df['Cluster 1'].max()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Cluster 1</th>\n",
       "      <th>Cluster 2</th>\n",
       "      <th>Cluster 3</th>\n",
       "      <th>Cluster 4</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>18769.00</td>\n",
       "      <td>18769.00</td>\n",
       "      <td>18769.00</td>\n",
       "      <td>18769.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>14.61</td>\n",
       "      <td>5.42</td>\n",
       "      <td>11.23</td>\n",
       "      <td>11.06</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>73.40</td>\n",
       "      <td>35.28</td>\n",
       "      <td>65.24</td>\n",
       "      <td>50.44</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>1.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>7.00</td>\n",
       "      <td>1.00</td>\n",
       "      <td>3.00</td>\n",
       "      <td>4.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>3831.00</td>\n",
       "      <td>1512.00</td>\n",
       "      <td>3621.00</td>\n",
       "      <td>2276.00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Cluster 1  Cluster 2  Cluster 3  Cluster 4\n",
       "count   18769.00   18769.00   18769.00   18769.00\n",
       "mean       14.61       5.42      11.23      11.06\n",
       "std        73.40      35.28      65.24      50.44\n",
       "min         0.00       0.00       0.00       0.00\n",
       "25%         0.00       0.00       0.00       0.00\n",
       "50%         1.00       0.00       0.00       0.00\n",
       "75%         7.00       1.00       3.00       4.00\n",
       "max      3831.00    1512.00    3621.00    2276.00"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "output = round(cluster_df.describe(), 2)\n",
    "output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "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": "3.6.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}