prepare_cumulative_covid_data.py

data_state.fillna(method='ffill',inplace=True)

max_perc = lambda x: np.max([i/100 for i in x])

data_state_piv = pd.pivot_table(data_state,
                                index='location',
                                values=['total_vaccinations_per_hundred','people_vaccinated','people_fully_vaccinated_per_hundred','people_fully_vaccinated'],
                                aggfunc={'total_vaccinations_per_hundred':max_perc,
                                       'people_fully_vaccinated_per_hundred':max_perc,
                                       'people_vaccinated':np.max,
                                       'people_fully_vaccinated':np.max})

data_state['date'] = pd.to_datetime(data_state['date'])
data_cumsum_doses = pd.pivot_table(data_state,
                                  index='date',
                                  values=['total_distributed','people_vaccinated','people_fully_vaccinated'],
                                  aggfunc=np.sum)