Martin Gabdushev

```python from etna.datasets import TSDataset, generate_ar_df from etna.pipeline import Pipeline from etna.ensembles import StackingEnsemble from etna.models import CatBoostModelMultiSegment, LinearPerSegmentModel, NaiveModel from etna.transforms import StandardScalerTransform, LagTransform, SegmentEncoderTransform, TimeSeriesImputerTransform, DateFlagsTransform from etna.metrics...

Hi, @mirik123 ! Thank you for your bug report. We have no explicit type checking here but there is typing support in the signature `DeepARModel(..., loss: Optional["DistributionLoss"], ...)` . So...

@mirik123 Warnings are ok. They don't affect final results. But we're going to resolve them all if it would be possible `checkpoint_callback=False` - could cause errors in case of using...

Hi! Seems it coudn't be done now. Most similiar way to do it with existed code base is to use `AutoRegressivePipeline` - in case of known features and lags with...

sudo py-spy record -o speedscope.json -f speedscope python f.py --rate 50 --nonblocking ```python # %% [markdown] # # Feature selection # # This notebook contains the simple examples of using...

[speedcope.json](https://gist.githubusercontent.com/martins0n/fe172360880af2a140328e8f4987d20a/raw/af29ab3358fcffe45ef5f594027fe8691613f7d9/etna-speedscope.json) Maybe we can pass all columns to mann whitney test. current implementation compares features separately.

We should add comment to changelog

Actual PR https://github.com/tinkoff-ai/etna/pull/1253