In the process of debugging why the pipeline was doing poorly on a specific day, we needed to adjust the start date. Eventually we used a pair of browsers, one in slide_forecaster and one in the actual forecast, with manual start date adjustment. This is jank.
In general, slide_training and n_training_pad are pretty awkward to use as-is, and should probably get refactored.
In the process of debugging why the pipeline was doing poorly on a specific day, we needed to adjust the start date. Eventually we used a pair of browsers, one in
slide_forecasterand one in the actual forecast, with manual start date adjustment. This is jank.In general,
slide_trainingandn_training_padare pretty awkward to use as-is, and should probably get refactored.