Mr-Geekman

Closed by #875.

It isn't relevant anymore because model was added in #1253.

For MRMR both ModelRelevanceTable and StatisticsRelevanceTable works worse than a baseline. I'll write all the results in the next message to compare. --- View entire conversation on ReviewNB

These are results on my personal laptop.SMAPE baseline: 8.094 Time baseline: 79.353 ModelRelevanceTable + fast_redundancy=True- SMAPE: 7.416- Time: 43.722 StatisticsRelevanceTable + fast_redundancy=True- SMAPE: 13.262- Time: 42.955 ModelRelevanceTable + fast_redundancy=False- SMAPE:...

It isn't relevant anymore.

Setting `disallow_untyped_defs = true` in `pyproject.toml` gives: ``` Found 358 errors in 99 files (checked 195 source files) ``` So, this issue requires a lot of changes. However, most of...

Hello! If you want for `MeanTransform` not to use data for the current timestamp you can do it by applying `MeanTransform` to lagged values. In that case you create lagged...