get_clean_covid_data.py

def get_data(df,transform_cols):
    """ This is the main function that transforms the raw OWID data into something we can use in the app
    Args:
        Original DataFrame from csv
    Returns:
        Processed / cleaned DataFrame
    """
    # loop through and subset each country to a list
    country_dfs = []

    # loop through each country
    for country in df.location.unique():
        df_country = df[df.location == country]  # df masked on country
        df_country.date = pd.to_datetime(df_country.date)  # convert string date to datetime
        df_country = df_country.sort_values(by='date')  # sort by date

        # transform our continuous columns
        for col in transform_cols:
            df_country[col] = update_series(df_country[col]).astype(int)

        # we will group by week and use the max value in that week
        df_country = df_country.groupby(pd.Grouper(key='date', freq='W')).max()
        country_dfs.append(df_country)  # append unique country dataframe to list

    df_final = pd.concat(country_dfs)

    df_final = df_final.reset_index().sort_values(by=['location', 'date'])
    df_final = df_final.fillna(0)

    # select start point and sort date in ascending order
    df_final = df_final.sort_values(by='date', ascending=True)
    df_final = df_final[df_final.date >= '2020-02-09']

    # date needs to be in string format for plotly animations to work
    df_final.date = df_final.date.astype(str)

    # get rid of any duplicate columns
    df_final = df_final.loc[:, ~df_final.columns.duplicated()]
    return df_final