backpain_2.py

backpain_2.py

from scipy import stats

def create_future():
  """Code to create datetime index for the covid WFH period."""
  future = list()
  for y in [2020]:
    for m in range(4, 13):
      date = f'{y}-{m:02d}' 
      future.append([date])
  for y in [2021]:
    for m in range(1, 12):
      date = f'{y}-{m:02d}' 
      future.append([date])
  future = pd.DataFrame(future)
  future.columns = ['ds']
  future['ds'] = pd.to_datetime(future['ds'])
  return future

for variableOfInterest in ['back pain', 'lower back pain', 'chiropractor near me']:
  ts = pd.DataFrame({'ds':df.loc[:'2020-03-01'].index,'y':df.loc[:'2020-03-01'][variableOfInterest]})

  prophet = Prophet(weekly_seasonality=False,
                    daily_seasonality=False, 
                    changepoint_range=1,
                    changepoint_prior_scale=0.1,
                    )
  prophet.fit(ts)

  future = create_future()
  forecast = prophet.predict(future)

  f, ax =plt.subplots()
  covid_df = df.loc['2020-04-01':][[variableOfInterest]]
  covid_df.plot(ax=ax)
  forecast_covid_df = forecast[['yhat']]
  forecast_covid_df.index = covid_df.index
  forecast_covid_df.plot(ax=ax, color='r')
  ax.set_ylabel("searches (%)", rotation=0, labelpad=60)
  ax.set(title=f'{variableOfInterest}')
  plt.legend(['ground truth searches (y)', 'predicted searches (yhat)'], bbox_to_anchor=(1.05, .65), loc=2, borderaxespad=0.)
  plt.show()
  diff = covid_df.values - forecast_covid_df.values
  print(stats.ttest_1samp(diff.ravel(), 0))