DataGenerator.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.colors as colors\n",
    "#import mayavi\n",
    "from itertools import product\n",
    "from random import shuffle\n",
    "from copy import deepcopy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "num_users = 1000 # number of users\n",
    "num_campaigns = 100 # number of campaigns with different options that all samples are equally distributed among\n",
    "samples_per_campaign = 100000 # number of samples per campaign\n",
    "num_actions = 10 # number of actions per campaign\n",
    "\n",
    "num_samples = num_campaigns * samples_per_campaign # total number of samples\n",
    "\n",
    "# Determine number of times that each user is seen\n",
    "\n",
    "# Step 1: draw relative frequencies for seeing different users from a uniform distribution\n",
    "user_nums = np.asarray([np.random.random() for i in range(num_users)])\n",
    "\n",
    "# Step 2: normalize frequencies to get probabilities, multiply by total number of samples,\n",
    "#   and convert to integers to get total numbers of times that each user is seen;\n",
    "#   total number of samples will be < num_samples due to rounding down\n",
    "user_nums = (num_samples*user_nums/user_nums.sum()).astype(int)\n",
    "\n",
    "# Step 3: for any users who are never seen, add 1 time\n",
    "i = 0\n",
    "j = user_nums.argsort()[0]\n",
    "while user_nums[j] == 0:\n",
    "    user_nums[j] += 1\n",
    "    i += 1\n",
    "    j = user_nums.argsort()[i]\n",
    "\n",
    "# Step 4: add 1 to users who are seen the most times to get the total number of samples = num_samples\n",
    "to_add = num_samples - user_nums.sum()\n",
    "for i in range(to_add):\n",
    "    j = user_nums.argsort()[num_users-i-1]\n",
    "    user_nums[j] += 1\n",
    "\n",
    "# Get correct total number of samples with correct numbers of samples for each user in random order\n",
    "user_id = []\n",
    "for user in range(num_users):\n",
    "    for num in range(user_nums[user]):\n",
    "        user_id.append(user+1)\n",
    "shuffle(user_id)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "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": [
    "sorted_user_nums = deepcopy(user_nums)\n",
    "sorted_user_nums.sort()\n",
    "plt.xlabel(\"Users sorted by times seen\")\n",
    "plt.ylabel(\"Times seen\")\n",
    "plt.plot(sorted_user_nums)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# For every user and each of the two user features, randomly pick the mean from a uniform distribution\n",
    "# over the interval (-0.5, 0.5) and the standard deviation from a uniform distribution over the interval (0, std)\n",
    "\n",
    "user_std1 = 0.2 # maximum standard deviation for user feature 1\n",
    "user_std2 = 0.3 # maximum standard deviation for user feature 2\n",
    "user_meanh = 0.2 # maximum absolute value of the mean for hidden user feature\n",
    "user_stdh = 0.1 # maximum standard deviation for hidden user feature\n",
    "\n",
    "user_feat1_means = {}\n",
    "user_feat1_stds = {}\n",
    "user_feat2_means = {}\n",
    "user_feat2_stds = {}\n",
    "user_hidden_means = {}\n",
    "user_hidden_stds = {}\n",
    "\n",
    "for n in range(num_users):\n",
    "    user_feat1_means[n+1] = np.random.random() - 0.5\n",
    "    user_feat1_stds[n+1] = user_std1 * np.random.random()\n",
    "    user_feat2_means[n+1] = np.random.random() - 0.5\n",
    "    user_feat2_stds[n+1] = user_std2 * np.random.random()\n",
    "    user_hidden_means[n+1] = 2 * user_meanh * (np.random.random() - 0.5)\n",
    "    user_hidden_stds[n+1] = user_stdh * np.random.random()\n",
    "\n",
    "# For every campaign and each of the two campaign features, randomly pick the mean from a uniform distribution\n",
    "# over the interval (-0.5, 0.5) and the standard deviation from a uniform distribution over the interval (0, std)\n",
    "\n",
    "campaign_std1 = 0.1 # maximum standard deviation for campaign feature 1\n",
    "campaign_std2 = 0.05 # maximum standard deviation for campaign feature 2\n",
    "campaign_meanh = 0.15 # maximum absolute value of the mean for hidden campaign feature\n",
    "campaign_stdh = 0.05 # maximum standard deviation for hidden campaign feature\n",
    "\n",
    "campaign_feat1_means = {}\n",
    "campaign_feat1_stds = {}\n",
    "campaign_feat2_means = {}\n",
    "campaign_feat2_stds = {}\n",
    "campaign_hidden_means = {}\n",
    "campaign_hidden_stds = {}\n",
    "\n",
    "for n in range(num_campaigns):\n",
    "    campaign_feat1_means[n+1] = np.random.random() - 0.5\n",
    "    campaign_feat1_stds[n+1] = campaign_std1 * np.random.random()\n",
    "    campaign_feat2_means[n+1] = np.random.random() - 0.5\n",
    "    campaign_feat2_stds[n+1] = campaign_std2 * np.random.random()\n",
    "    campaign_hidden_means[n+1] = 2 * campaign_meanh * (np.random.random() - 0.5)\n",
    "    campaign_hidden_stds[n+1] = campaign_stdh * np.random.random()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1000000\n",
      "2000000\n",
      "3000000\n",
      "4000000\n",
      "5000000\n",
      "6000000\n",
      "7000000\n",
      "8000000\n",
      "9000000\n",
      "10000000\n"
     ]
    }
   ],
   "source": [
    "campaign_id = []\n",
    "user_feat1 = []\n",
    "user_feat2 = []\n",
    "user_hidden = []\n",
    "campaign_feat1 = []\n",
    "campaign_feat2 = []\n",
    "campaign_hidden = []\n",
    "reward = []\n",
    "optimal = []\n",
    "\n",
    "for n in range(num_samples):\n",
    "    if (n+1) % 1000000 == 0:\n",
    "        print (n+1)\n",
    "    \n",
    "# assign user id for each sample so that each user appears the correct number of times\n",
    "    j = user_id[n]\n",
    "    # sample the values for the user features from normal distributions with pre-defined means and stds\n",
    "    user_feat1.append(np.random.normal(user_feat1_means[j], user_feat1_stds[j]))\n",
    "    user_feat2.append(np.random.normal(user_feat2_means[j], user_feat2_stds[j]))\n",
    "    user_hidden.append(np.random.normal(user_hidden_means[j], user_hidden_stds[j]))\n",
    "    k = n // samples_per_campaign + 1 # assign a campaign id (based on the sample number)\n",
    "    campaign_id.append(k)\n",
    "    campaign_feat1.append(np.random.normal(campaign_feat1_means[k], campaign_feat1_stds[k]))\n",
    "    campaign_feat2.append(np.random.normal(campaign_feat2_means[k], campaign_feat2_stds[k]))\n",
    "    campaign_hidden.append(np.random.normal(campaign_hidden_means[k], campaign_hidden_stds[k]))\n",
    "    # set up 3D user and campaign vectors: (feature 1, feature 2, hidden feature)        \n",
    "    user_vec = np.asarray([user_feat1[-1], user_feat2[-1], user_hidden[-1]])\n",
    "    campaign_vec = np.asarray([campaign_feat1[-1], campaign_feat2[-1], campaign_hidden[-1]])\n",
    "    # find angle between user and campaign vectors\n",
    "    cosphi = np.dot(user_vec, campaign_vec.T) / (np.linalg.norm(user_vec) * np.linalg.norm(campaign_vec))\n",
    "    # define reward as the arccos of this angle, shifted and scaled to be between -0.5 and +0.5\n",
    "    reward.append(np.arccos(cosphi)/np.pi - 0.5)\n",
    "    # select optimal action based on reward value:\n",
    "    #   -0.5 < reward < -0.4 --> action 1, -0.4 < reward < -0.3 --> action 2, etc.\n",
    "    action = 1\n",
    "    while reward[-1] > -0.5 + action/num_actions:\n",
    "        action += 1\n",
    "    optimal.append(action)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "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": [
    "bins = [i*0.01 - 0.5 for i in range(101)]\n",
    "plt.xlim(-0.5, 0.5)\n",
    "plt.xlabel(\"Reward\")\n",
    "plt.ylabel(\"Counts\")\n",
    "plt.hist(reward, bins=bins)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1000000\n",
      "2000000\n",
      "3000000\n",
      "4000000\n",
      "5000000\n",
      "6000000\n",
      "7000000\n",
      "8000000\n",
      "9000000\n",
      "10000000\n"
     ]
    }
   ],
   "source": [
    "# Bin user features\n",
    "bins = [round(i/100 - 1, 2) for i in range(201)] # with a step of 0.01\n",
    "grid_pts = list(product(bins, bins))\n",
    "grid = {}\n",
    "for pt in grid_pts:\n",
    "    grid[pt] = 0\n",
    "\n",
    "for n in range(num_samples):\n",
    "    if (n + 1) % 1000000 == 0:\n",
    "        print (n + 1)\n",
    "    pt = [round(np.ceil(user_feat1[n]*100)/100, 2), round(np.ceil(user_feat2[n]*100)/100, 2)]\n",
    "    # if user features are < -1 or > 1, bin them with -1 or 1 (a few outliers, not visible in graph)\n",
    "    if pt[0] < -1:\n",
    "        pt[0] = -1\n",
    "    if pt[1] < -1:\n",
    "        pt[1] = -1\n",
    "    if pt[0] > 1:\n",
    "        pt[0] = 1\n",
    "    if pt[1] > 1: \n",
    "        pt[1] = 1\n",
    "    pt = tuple(pt)\n",
    "    grid[pt] += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of samples with each combination of user features\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot distribution of user features\n",
    "matr_user = np.zeros((201, 201))\n",
    "for i in range(201):\n",
    "    for j in range(201):\n",
    "        matr_user[i][j] = grid[(bins[i], bins[j])]\n",
    "\n",
    "print('Number of samples with each combination of user features')\n",
    "plt.imshow(matr_user, cmap = 'hot', norm=colors.PowerNorm(gamma=0.8), extent=[-1,1,-1,1])\n",
    "plt.colorbar()\n",
    "plt.xlabel('User feature 1')\n",
    "plt.ylabel('User feature 2')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "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": [
    "# Plot histogram for hidden user feature\n",
    "plt.xlim(-0.5, 0.5)\n",
    "plt.xlabel(\"Hidden user feature\")\n",
    "plt.ylabel(\"Counts\")\n",
    "plt.hist(user_hidden, bins=bins)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1000000\n",
      "2000000\n",
      "3000000\n",
      "4000000\n",
      "5000000\n",
      "6000000\n",
      "7000000\n",
      "8000000\n",
      "9000000\n",
      "10000000\n"
     ]
    }
   ],
   "source": [
    "# Bin campaign features\n",
    "bins = [round(i/100 - 1, 2) for i in range(201)]\n",
    "grid_pts = list(product(bins, bins))\n",
    "grid_campaign = {}\n",
    "for pt in grid_pts:\n",
    "    grid_campaign[pt] = 0\n",
    "\n",
    "for n in range(num_samples):\n",
    "    if (n + 1) % 1000000 == 0:\n",
    "        print (n + 1)\n",
    "    pt = [round(np.ceil(campaign_feat1[n]*100)/100, 2), round(np.ceil(campaign_feat2[n]*100)/100, 2)]\n",
    "    # if campaign features are < -1 or > 1, bin them with -1 or 1 (a few outliers, not visible in graph)\n",
    "    if pt[0] < -1:\n",
    "        pt[0] = -1\n",
    "    if pt[1] < -1:\n",
    "        pt[1] = -1\n",
    "    if pt[0] > 1:\n",
    "        pt[0] = 1\n",
    "    if pt[1] > 1: \n",
    "        pt[1] = 1\n",
    "    pt = tuple(pt)\n",
    "    grid_campaign[pt] += 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of samples with each combination of campaign features\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Plot campaign features\n",
    "matr_campaign = np.zeros((201, 201))\n",
    "for i in range(201):\n",
    "    for j in range(201):\n",
    "        matr_campaign[i][j] = grid_campaign[(bins[i], bins[j])]\n",
    "\n",
    "print('Number of samples with each combination of campaign features')\n",
    "plt.imshow(matr_campaign, cmap = 'hot', norm=colors.PowerNorm(gamma=0.5), extent=[-1,1,-1,1])\n",
    "plt.colorbar()\n",
    "plt.xlabel('Campaign feature 1')\n",
    "plt.ylabel('Campaign feature 2')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "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": [
    "# Plot histogram for hidden campaign feature\n",
    "plt.xlim(-0.3, 0.3)\n",
    "plt.xlabel(\"Hidden campaign feature\")\n",
    "plt.ylabel(\"Counts\")\n",
    "plt.hist(campaign_hidden, bins=bins)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "scrolled": true
   },
   "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": [
    "# print numbers of samples for all optimal actions\n",
    "opts = np.zeros(10).astype(int)\n",
    "for n in range(num_samples):\n",
    "    opts[optimal[n]-1] += 1\n",
    "\n",
    "a = [i+1 for i in range(num_actions)]\n",
    "s = [opts[i] for i in range(num_actions)]\n",
    "#print (\"{:>12s}{:>12s}\".format(\"Action\", \"# Samples\"))\n",
    "#for n in range(num_actions):\n",
    "#    print (\"{:12d}{:12d}\".format(n+1, opts[n]))\n",
    "\n",
    "plt.xlabel(\"Action\")\n",
    "plt.ylabel(\"Times action is optimal\")\n",
    "plt.bar(a, s)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "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>user_id</th>\n",
       "      <th>user_feature_1</th>\n",
       "      <th>user_feature_2</th>\n",
       "      <th>user_feature_hidden</th>\n",
       "      <th>campaign_id</th>\n",
       "      <th>campaign_feature_1</th>\n",
       "      <th>campaign_feature_2</th>\n",
       "      <th>campaign_feature_hidden</th>\n",
       "      <th>reward</th>\n",
       "      <th>optimal_action</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>897</td>\n",
       "      <td>-0.274970</td>\n",
       "      <td>0.202852</td>\n",
       "      <td>-0.103931</td>\n",
       "      <td>1</td>\n",
       "      <td>-0.231624</td>\n",
       "      <td>0.462141</td>\n",
       "      <td>0.148505</td>\n",
       "      <td>-0.264811</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>35</td>\n",
       "      <td>0.489462</td>\n",
       "      <td>-0.040341</td>\n",
       "      <td>0.066766</td>\n",
       "      <td>1</td>\n",
       "      <td>-0.317610</td>\n",
       "      <td>0.440337</td>\n",
       "      <td>0.149169</td>\n",
       "      <td>0.198915</td>\n",
       "      <td>7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>689</td>\n",
       "      <td>-0.458089</td>\n",
       "      <td>0.120899</td>\n",
       "      <td>0.060884</td>\n",
       "      <td>1</td>\n",
       "      <td>-0.234123</td>\n",
       "      <td>0.429651</td>\n",
       "      <td>0.146294</td>\n",
       "      <td>-0.241988</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>350</td>\n",
       "      <td>-0.390775</td>\n",
       "      <td>0.041562</td>\n",
       "      <td>-0.096576</td>\n",
       "      <td>1</td>\n",
       "      <td>-0.233778</td>\n",
       "      <td>0.420403</td>\n",
       "      <td>0.149173</td>\n",
       "      <td>-0.153331</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>854</td>\n",
       "      <td>-0.000106</td>\n",
       "      <td>0.425490</td>\n",
       "      <td>0.181184</td>\n",
       "      <td>1</td>\n",
       "      <td>-0.205593</td>\n",
       "      <td>0.453234</td>\n",
       "      <td>0.149373</td>\n",
       "      <td>-0.367993</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   user_id  user_feature_1  user_feature_2  user_feature_hidden  campaign_id  \\\n",
       "0      897       -0.274970        0.202852            -0.103931            1   \n",
       "1       35        0.489462       -0.040341             0.066766            1   \n",
       "2      689       -0.458089        0.120899             0.060884            1   \n",
       "3      350       -0.390775        0.041562            -0.096576            1   \n",
       "4      854       -0.000106        0.425490             0.181184            1   \n",
       "\n",
       "   campaign_feature_1  campaign_feature_2  campaign_feature_hidden    reward  \\\n",
       "0           -0.231624            0.462141                 0.148505 -0.264811   \n",
       "1           -0.317610            0.440337                 0.149169  0.198915   \n",
       "2           -0.234123            0.429651                 0.146294 -0.241988   \n",
       "3           -0.233778            0.420403                 0.149173 -0.153331   \n",
       "4           -0.205593            0.453234                 0.149373 -0.367993   \n",
       "\n",
       "   optimal_action  \n",
       "0               3  \n",
       "1               7  \n",
       "2               3  \n",
       "3               4  \n",
       "4               2  "
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# write all features to dataframe\n",
    "df = pd.DataFrame()\n",
    "df['user_id'] = user_id\n",
    "df['user_feature_1'] = user_feat1\n",
    "df['user_feature_2'] = user_feat2\n",
    "df['user_feature_hidden'] = user_hidden\n",
    "df['campaign_id'] = campaign_id\n",
    "df['campaign_feature_1'] = campaign_feat1\n",
    "df['campaign_feature_2'] = campaign_feat2\n",
    "df['campaign_feature_hidden'] = campaign_hidden\n",
    "df['reward'] = reward\n",
    "df['optimal_action'] = optimal\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# write dataframe to file\n",
    "df.to_csv('/home/aleksey/Downloads/generated4.csv', index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Distribution of optimal actions among users\n",
      "      Action    Mean (% total)   Stdev (% total)\n",
      "           1            4.0567            1.7278\n",
      "           2            9.7429            2.3406\n",
      "           3           11.8132            1.7940\n",
      "           4           12.2696            1.9328\n",
      "           5           12.2847            1.8799\n",
      "           6           12.1714            1.8959\n",
      "           7           12.1474            1.9056\n",
      "           8           11.7570            1.7500\n",
      "           9            9.7208            2.4561\n",
      "          10            4.0361            1.8019\n"
     ]
    }
   ],
   "source": [
    "# Find the mean and standard deviation of the numbers of optimal actions over users\n",
    "all_actions = []\n",
    "for user in range(num_users):\n",
    "    df_subset = df.loc[df['user_id'] == user+1]\n",
    "    actions = df_subset.groupby('optimal_action').count()['user_id']\n",
    "    if len(actions) != num_actions:\n",
    "        actions2 = []\n",
    "        keys = list(actions.keys())\n",
    "        for action in range(num_actions):\n",
    "            if action in keys:\n",
    "                actions2.append(actions[action])\n",
    "            else:\n",
    "                actions2.append(0)\n",
    "        actions = np.asarray(actions2)\n",
    "    else:\n",
    "        actions = actions.to_numpy()\n",
    "    all_actions.append(100*actions/actions.sum())\n",
    "all_actions = np.stack(all_actions, axis=0)\n",
    "print (\"Distribution of optimal actions among users\")\n",
    "print (\"{:>12s}{:>18s}{:>18s}\".format(\"Action\", \"Mean (% total)\", \"Stdev (% total)\"))\n",
    "means = all_actions.mean(axis=0)\n",
    "stdevs = all_actions.std(axis=0)\n",
    "for n in range(means.shape[0]):\n",
    "    print(\"{:>12d}{:>18.4f}{:>18.4f}\".format(n+1, means[n], stdevs[n]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Distribution of optimal actions among campaigns\n",
      "      Action    Mean (% total)   Stdev (% total)\n",
      "           1            4.0912            1.4276\n",
      "           2            9.7894            1.8876\n",
      "           3           11.8266            1.4253\n",
      "           4           12.2005            1.4549\n",
      "           5           12.2614            2.2261\n",
      "           6           12.2014            2.1947\n",
      "           7           12.1514            1.5408\n",
      "           8           11.7604            1.5302\n",
      "           9            9.6913            1.8123\n",
      "          10            4.0265            1.3604\n"
     ]
    }
   ],
   "source": [
    "# Find the mean and standard deviation of the numbers of optimal actions over campaigns\n",
    "all_actions = []\n",
    "for campaign in range(num_campaigns):\n",
    "    df_subset = df.loc[df['campaign_id'] == campaign+1]\n",
    "    actions = df_subset.groupby('optimal_action').count()['campaign_id']\n",
    "    if len(actions) != num_actions:\n",
    "        actions2 = []\n",
    "        keys = list(actions.keys())\n",
    "        for action in range(num_actions):\n",
    "            if action in keys:\n",
    "                actions2.append(actions[action])\n",
    "            else:\n",
    "                actions2.append(0)\n",
    "        actions = np.asarray(actions2)\n",
    "    else:\n",
    "        actions = actions.to_numpy()\n",
    "    all_actions.append(100*actions/actions.sum())\n",
    "all_actions = np.stack(all_actions, axis=0)\n",
    "print (\"Distribution of optimal actions among campaigns\")\n",
    "print (\"{:>12s}{:>18s}{:>18s}\".format(\"Action\", \"Mean (% total)\", \"Stdev (% total)\"))\n",
    "means = all_actions.mean(axis=0)\n",
    "stdevs = all_actions.std(axis=0)\n",
    "for n in range(means.shape[0]):\n",
    "    print(\"{:>12d}{:>18.4f}{:>18.4f}\".format(n+1, means[n], stdevs[n]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "User seen maximum number of times: 605 , seen 20044 times\n",
      "User seen minimum number of times: 817 , seen 17 times\n"
     ]
    }
   ],
   "source": [
    "# Users seen maximum and minimum number of times\n",
    "num_times_seen = []\n",
    "for user in range(num_users):\n",
    "    df_subset = df.loc[df['user_id'] == user+1]\n",
    "    num_times_seen.append(df_subset.shape[0])\n",
    "num_times_seen = np.asarray(num_times_seen)\n",
    "print (\"User seen maximum number of times:\", num_times_seen.argmax(), \", seen\", num_times_seen.max(), \"times\")\n",
    "print (\"User seen minimum number of times:\", num_times_seen.argmin(), \", seen\", num_times_seen.min(), \"times\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}