fb_model.ipynb

fb_model.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "fb_model.ipynb",
      "private_outputs": true,
      "provenance": [],
      "collapsed_sections": [],
      "authorship_tag": "ABX9TyNwxITfC7weYWxRkS7/4yOb",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/gist/Caellwyn/3517deb91ed2a8720266305aea4b6b79/fb_model.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NjGd8SYr9k1w"
      },
      "source": [
        "import pandas as pd\r\n",
        "#!pip install fbprophet\r\n",
        "from fbprophet import Prophet\r\n",
        "import matplotlib.pyplot as plt\r\n",
        "\r\n",
        "division = 'country'  #regional data is available for some countries\r\n",
        "region = 'United States'\r\n",
        "prediction = 'ConfirmedCases' #ConfirmedDeaths is also available for forecasting.\r\n",
        "\r\n",
        "#get the latest data from OxCGRT\r\n",
        "DATA_URL = 'https://raw.githubusercontent.com/OxCGRT/covid-policy-tracker/master/data/OxCGRT_latest.csv'\r\n",
        "full_df = pd.read_csv(DATA_URL,\r\n",
        "                usecols=['Date','CountryName','RegionName','Jurisdiction',\r\n",
        "                           'ConfirmedCases','ConfirmedDeaths'],\r\n",
        "                parse_dates=['Date'],\r\n",
        "                encoding=\"ISO-8859-1\",\r\n",
        "                dtype={\"RegionName\": str,\r\n",
        "                        \"CountryName\":str})\r\n",
        "\r\n",
        "#Filter the region we want to predict\r\n",
        "if division == 'country':\r\n",
        "    df = full_df[(full_df['Jurisdiction'] == 'NAT_TOTAL') \r\n",
        "    & (full_df['CountryName'] == region)][:-1]\r\n",
        "elif division == 'state':\r\n",
        "    df = full_df[(full_df['Jurisdiction'] == 'STATE_TOTAL') \r\n",
        "    & (full_df['RegionName'] == region)][:-1]\r\n",
        "\r\n",
        "#Since we are not using exogenous variables, we just keep the dates and endogenous data\r\n",
        "df = df[['Date',prediction]].rename(columns = {'Date':'ds', prediction:'y'})"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1s0h3Trm9xJD"
      },
      "source": [
        "# set how many days to forecast\r\n",
        "forecast_length = 30\r\n",
        "# instantiate and fit the model\r\n",
        "m = Prophet()\r\n",
        "m.fit(df)\r\n",
        "# create the prediction dataframe 'forecast_length' days past the fit data\r\n",
        "future = m.make_future_dataframe(periods=forecast_length)\r\n",
        "# make the forecast to the end of the 'future' dataframe\r\n",
        "forecast = m.predict(future)\r\n",
        "\r\n",
        "to_plot = forecast[forecast.ds > '2020-12-01'].merge(df, how='left')\r\n",
        "\r\n",
        "plt.figure(figsize = (10,7))\r\n",
        "plt.plot(to_plot['ds'], to_plot['yhat'], label='Forecasted Cases')\r\n",
        "plt.plot(to_plot['ds'], to_plot['y'], label='True Cases')\r\n",
        "plt.fill_between(to_plot['ds'], to_plot['yhat_upper'], to_plot['yhat_lower'],\r\n",
        "                 alpha=.2, label='Confidence')\r\n",
        "plt.title('Facebook Prophet Forecasted COVID-19 cases, 1-22-2021 to 2-20-2021')\r\n",
        "plt.legend()\r\n",
        "plt.savefig('prophet_forecast.png')\r\n",
        "plt.show()\r\n",
        "print('\\n The \"forecast\" DataFrame \\n')\r\n",
        "forecast"
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}