Open ramdhan1989 opened 4 years ago
I would like to implement fbprophet in gluonts with this configuration :
model = Prophet(growth='linear',holidays=us_public_holidays,daily_seasonality=False,mcmc_samples=50,weekly_seasonality=True) #model.add_country_holidays(country_name='US') model.add_seasonality(name='hourly_on_summer', period=1, fourier_order=2, condition_name='on_summer') model.add_seasonality(name='hourly_off_summer', period=1, fourier_order=2, condition_name='off_summer') model.add_seasonality(name='on_anomaly', period=1, fourier_order=2, condition_name='on_anomaly_seasonality') model.add_seasonality(name='off_anomaly', period=1, fourier_order=2, condition_name='off_anomaly_seasonality') model.add_seasonality(name='monthly', period=30.5, fourier_order=2) model.add_regressor('temp', prior_scale=0.5, mode='multiplicative') model.add_regressor('rain', prior_scale=0.5, mode='multiplicative') #model.add_regressor('sunhr', prior_scale=0.5, mode='multiplicative') model.add_regressor('windspeed', prior_scale=0.5, mode='multiplicative') model.add_regressor('winddeg', prior_scale=0.5, mode='multiplicative') model.add_regressor('vis', prior_scale=0.5, mode='multiplicative') model.add_regressor('windchill', prior_scale=0.5, mode='multiplicative') model.add_regressor('windgust', prior_scale=0.5, mode='multiplicative') model.add_regressor('feelslike', prior_scale=0.5, mode='multiplicative') #model.add_regressor('snow', prior_scale=0.5, mode='multiplicative') model.fit(train_all)
how to implement in gluonts? I know from the doc is mentioned there about "def configure_model" but I didn't find any detail example code to explain the implementation. then, how to train the model ? Sorry, I didn't find example of this
thank you
I think you can consult the link https://github.com/awslabs/gluon-ts/issues/338
I would like to implement fbprophet in gluonts with this configuration :
how to implement in gluonts? I know from the doc is mentioned there about "def configure_model" but I didn't find any detail example code to explain the implementation. then, how to train the model ? Sorry, I didn't find example of this
thank you