05-data-analysis.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Import necessary libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Importing plotly failed. Interactive plots will not work.\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from fbprophet import Prophet\n",
    "import matplotlib.pyplot as plt\n",
    " \n",
    "%matplotlib inline\n",
    " \n",
    "plt.rcParams['figure.figsize']=(20,10)\n",
    "plt.style.use('ggplot')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Read in the data\n",
    "\n",
    "Read the data in from the retail sales CSV file in the examples folder then set the index to the 'date' column. We are also parsing dates in the data file.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "sales_df = pd.read_csv('./data/retail_sales.csv', index_col='date', parse_dates=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "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>sales</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>date</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2009-10-01</th>\n",
       "      <td>338630</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2009-11-01</th>\n",
       "      <td>339386</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2009-12-01</th>\n",
       "      <td>400264</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2010-01-01</th>\n",
       "      <td>314640</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2010-02-01</th>\n",
       "      <td>311022</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             sales\n",
       "date              \n",
       "2009-10-01  338630\n",
       "2009-11-01  339386\n",
       "2009-12-01  400264\n",
       "2010-01-01  314640\n",
       "2010-02-01  311022"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sales_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Prepare for Prophet\n",
    "\n",
    "For prophet to work, we need to change the names of these columns to 'ds' and 'y', so lets just create a new dataframe and keep our old one handy (you'll see why later). The new dataframe will initially be created with an integer index so we can rename the columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "df = sales_df.reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "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>date</th>\n",
       "      <th>sales</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2009-10-01</td>\n",
       "      <td>338630</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2009-11-01</td>\n",
       "      <td>339386</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2009-12-01</td>\n",
       "      <td>400264</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2010-01-01</td>\n",
       "      <td>314640</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2010-02-01</td>\n",
       "      <td>311022</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        date   sales\n",
       "0 2009-10-01  338630\n",
       "1 2009-11-01  339386\n",
       "2 2009-12-01  400264\n",
       "3 2010-01-01  314640\n",
       "4 2010-02-01  311022"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "df=df.rename(columns={'date':'ds', 'sales':'y'})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "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>ds</th>\n",
       "      <th>y</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2009-10-01</td>\n",
       "      <td>338630</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2009-11-01</td>\n",
       "      <td>339386</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2009-12-01</td>\n",
       "      <td>400264</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2010-01-01</td>\n",
       "      <td>314640</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2010-02-01</td>\n",
       "      <td>311022</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          ds       y\n",
       "0 2009-10-01  338630\n",
       "1 2009-11-01  339386\n",
       "2 2009-12-01  400264\n",
       "3 2010-01-01  314640\n",
       "4 2010-02-01  311022"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now's a good time to take a look at your data.  Plot the data using pandas' ```plot``` function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2291a4bdd88>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df.set_index('ds').y.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "df['y'] = np.log(df['y'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "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>ds</th>\n",
       "      <th>y</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>67</th>\n",
       "      <td>2015-05-01</td>\n",
       "      <td>13.044650</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>68</th>\n",
       "      <td>2015-06-01</td>\n",
       "      <td>13.013060</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>69</th>\n",
       "      <td>2015-07-01</td>\n",
       "      <td>13.033991</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>70</th>\n",
       "      <td>2015-08-01</td>\n",
       "      <td>13.030993</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>71</th>\n",
       "      <td>2015-09-01</td>\n",
       "      <td>12.973671</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           ds          y\n",
       "67 2015-05-01  13.044650\n",
       "68 2015-06-01  13.013060\n",
       "69 2015-07-01  13.033991\n",
       "70 2015-08-01  13.030993\n",
       "71 2015-09-01  12.973671"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x2291c6cfc08>"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df.set_index('ds').y.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As you can see in the above chart, the plot looks the same as the first one but just at a different scale."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Running Prophet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'Prophet' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-23-ed5605b74878>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[1;32m----> 1\u001b[1;33m \u001b[0mmodel\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mProphet\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m      2\u001b[0m \u001b[0mmodel\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mfit\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mdf\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m;\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mNameError\u001b[0m: name 'Prophet' is not defined"
     ]
    }
   ],
   "source": [
    "\n",
    "\n",
    "model = Prophet()\n",
    "model.fit(df);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'model' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-21-542366c09568>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[1;32m----> 1\u001b[1;33m \u001b[0mfuture\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mmake_future_dataframe\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mperiods\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;36m24\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mfreq\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;34m'm'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m      2\u001b[0m \u001b[0mfuture\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtail\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mNameError\u001b[0m: name 'model' is not defined"
     ]
    }
   ],
   "source": [
    "future = model.make_future_dataframe(periods=24, freq = 'm')\n",
    "future.tail()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To forecast this future data, we need to run it through Prophet's model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'model' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-22-1785450a81fc>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[1;32m----> 1\u001b[1;33m \u001b[0mforecast\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpredict\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mfuture\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m: name 'model' is not defined"
     ]
    }
   ],
   "source": [
    "forecast = model.predict(future)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The resulting forecast dataframe contains quite a bit of data, but we really only care about a few columns.  First, let's look at the full dataframe:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'forecast' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-19-f17a858a2daa>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[1;32m----> 1\u001b[1;33m \u001b[0mforecast\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtail\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m: name 'forecast' is not defined"
     ]
    }
   ],
   "source": [
    "forecast.tail()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We really only want to look at yhat, yhat_lower and yhat_upper, so we can do that with:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "forecast[['ds', 'yhat', 'yhat_lower', 'yhat_upper']].tail()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Plotting Prophet results\n",
    "\n",
    "Prophet has a plotting mechanism called ```plot```.  This plot functionality draws the original data (black dots), the model (blue line) and the error of the forecast (shaded blue area)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "model.plot(forecast);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Personally, I'm not a fan of this visualization so I like to break the data up and build a chart myself.  The next section describes how I build my own visualization for Prophet modeling "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Visualizing Prophet models\n",
    "\n",
    "In order to build a useful dataframe to visualize our model versus our original data, we need to combine the output of the Prophet model with our original data set, then we'll build a new chart manually using pandas and matplotlib.\n",
    "\n",
    "First, let's set our dataframes to have the same index of ```ds```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "df.set_index('ds', inplace=True)\n",
    "forecast.set_index('ds', inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, we'll combine the original data and our forecast model data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "viz_df = sales_df.join(forecast[['yhat', 'yhat_lower','yhat_upper']], how = 'outer')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "If we look at the ```head()```, we see the data has been joined correctly but the scales of our original data (sales) and our model (yhat) are different. We need to rescale the yhat colums(s) to get the same scale, so we'll use numpy's ```exp``` function to do that."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "viz_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "viz_df['yhat_rescaled'] = np.exp(viz_df['yhat'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "viz_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's take a look at the ```sales``` and ```yhat_rescaled``` data together in a chart."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "viz_df[['sales', 'yhat_rescaled']].plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "sales_df.index = pd.to_datetime(sales_df.index) #make sure our index as a datetime object\n",
    "connect_date = sales_df.index[-2] #select the 2nd to last date"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Using the ```connect_date``` we can now grab only the model data that after that date (you'll see why in a minute). To do this, we'll mask the forecast data."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "mask = (forecast.index > connect_date)\n",
    "predict_df = forecast.loc[mask]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "predict_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, let's build a dataframe to use in our new visualization. We'll follow the same steps we did before."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "viz_df = sales_df.join(predict_df[['yhat', 'yhat_lower','yhat_upper']], how = 'outer')\n",
    "viz_df['yhat_scaled']=np.exp(viz_df['yhat'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, if we take a look at the ```head()``` of ```viz_df``` we'll see 'NaN's everywhere except for our original data rows.  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "viz_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "If we take a look at the ```tail()``` of the ```viz_df``` you'll see we have data for the forecasted data and NaN's for the original data series."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "viz_df.tail()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## time to plot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "fig, ax1 = plt.subplots()\n",
    "ax1.plot(viz_df.sales)\n",
    "ax1.plot(viz_df.yhat_scaled, color='black', linestyle=':')\n",
    "ax1.fill_between(viz_df.index, np.exp(viz_df['yhat_upper']), np.exp(viz_df['yhat_lower']), alpha=0.5, color='darkgray')\n",
    "ax1.set_title('Sales (Orange) vs Sales Forecast (Black)')\n",
    "ax1.set_ylabel('Dollar Sales')\n",
    "ax1.set_xlabel('Date')\n",
    "\n",
    "L=ax1.legend() #get the legend\n",
    "L.get_texts()[0].set_text('Actual Sales') #change the legend text for 1st plot\n",
    "L.get_texts()[1].set_text('Forecasted Sales') #change the legend text for 2nd plot"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This visualization is much better (in my opinion) than the default fbprophet plot. It is much easier to quickly understand and describe what's happening. The orange line is actual sales data and the black dotted line is the forecast. The gray shaded area is the uncertaintity estimation of the forecast."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import pyflux as pf\n",
    "import matplotlib.pyplot as plt\n",
    "from fbprophet import Prophet\n",
    "    \n",
    "%matplotlib inline\n",
    "plt.rcParams['figure.figsize']=(20,10)\n",
    "plt.style.use('ggplot')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load the data\n",
    "\n",
    "For this work, we're going to use the same retail sales data that we've used before. It can be found in the examples directory of this repository."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "sales_df = pd.read_csv('../data/sales.csv', index_col='date', parse_dates=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "sales_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Like all good modeling projects, we need to take a look at the data to get an idea of what it looks like."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "sales_df.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# Prep data for prophet and run prophet\n",
    "df = sales_df.reset_index()\n",
    "df=df.rename(columns={'date':'ds', 'sales':'y'})\n",
    "\n",
    "model = Prophet(weekly_seasonality=True)\n",
    "model.fit(df);\n",
    "future = model.make_future_dataframe(periods=24, freq = 'm')\n",
    "\n",
    "forecast = model.predict(future)\n",
    "model.plot(forecast);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "With our prophet model ready for comparison, let's build a model with pyflux's dynamic linear regresion model.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### More Data Viz "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "diff_log = pd.DataFrame(np.diff(np.log(sales_df['sales'].values)))\n",
    "diff_log.index = sales_df.index.values[1:sales_df.index.values.shape[0]]\n",
    "diff_log.columns = [\"Sales DiffLog\"]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "sales_df['logged']=np.log(sales_df['sales'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "sales_df.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "sales_df.plot(subplots=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "pf.acf_plot(diff_log.values.T[0])\n",
    "pf.acf_plot(np.square(diff_log.values.T[0]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see that at a lag of 1 and 2 months, there are positive correlations for sales but as time goes on, that correlation drops quickly to a negative correlation that stays in place over time, which hints at the fact that there are some autoregressive effects within this data.  \n",
    "\n",
    "Because of this fact, we can start our modeling by using an ARMA model of some sort."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "Logged = pd.DataFrame(np.log(sales_df['sales']))\n",
    "Logged.index = pd.to_datetime(sales_df.index)\n",
    "Logged.columns = ['Sales - Logged']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "Logged.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "modelLLT = pf.LLT(data=Logged)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "x = modelLLT.fit()\n",
    "x.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "model.plot_fit(figsize=(20,10))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "modelLLT.plot_predict_is(h=len(Logged)-1, figsize=(20,10))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "predicted = modelLLT.predict_is(h=len(Logged)-1)\n",
    "predicted.columns = ['Predicted']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "predicted.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "np.exp(predicted).plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "sales_df_future=sales_df\n",
    "sales_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "final_sales=sales_df.merge(np.exp(predicted),right_on=predicted.index)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "final_sales = sales_df.merge()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "final_sales.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "final_sales.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "is_executing": false,
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.3"
  },
  "pycharm": {
   "stem_cell": {
    "cell_type": "raw",
    "metadata": {
     "collapsed": false
    },
    "source": []
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}