search

tblume1992 / MFLES

MIT License
109 stars 4 forks source link

Conformal #1

Closed firmai closed 3 months ago

firmai commented 8 months ago

Thanks again for a nice package, it would be great if we can weave conformal prediction into this function and they can be returned as columns.

from MFLES.Forecaster import fit_from_df

output = fit_from_df(df_signal,
                      forecast_horizon=210,
                      freq='W',
                      seasonal_period=[13, 26, 52, 104,210,420, 840],
                      id_column='ticker',
                      time_column='date',
                      value_column='total_revenue',
                      floor=0)
tblume1992 commented 8 months ago

@firmai I totally agree! Should not be tough, I'll add it.

tblume1992 commented 5 months ago

@firmai I was not ignoring this request! Nixtla just released MFLES as part of their statsforecast library which enables this feature and should be the go-to for MFLES forecasting. I will keep this package around more for experimentation. https://github.com/Nixtla/statsforecast