Closed antoinecarme closed 6 years ago
Added a simple interpolation when needed :
time_in_seconds_interpolated = time_in_seconds.interpolate(method='linear', limit_direction='both', axis=0)
before training or forecasting the model.
More sophisticated interpolation is possible , but we let that to the user (can be business related).
Closing
Following issue #19, the dataset can contain missing data in the time column.
Allow the user to interpolate the time column before building the model. Time interpolation should keep the same time numpy dtype.