Skip to content

etna 2.8.0
Compare
Choose a tag to compare
@d-a-bunin d-a-bunin released this 13 Aug 14:15
· 37 commits to master since this release
2.8.0
1f74446
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Learn about vigilant mode.

Highlights

  • Add new anomaly detection functions: get_anomalies_mad, get_anomalies_iqr, get_anomalies_isolation_forest
  • Add new anomaly detection transforms: MADOutlierTransform, IForestOutlierTransform, IQROutlierTransform
  • Add decomposition transforms for anomaly detection: FourierDecomposeTransform ModelDecomposeTransform
  • Add MeanEncoderTransform
  • Set upper bound <2 on numpy version
  • Fix VotingEnsemble, StackingEnsemble, DirectEnsemble to have a valid params_to_tune that returns empty dict

Full changelog

Added

  • Add get_anomalies_iqr function for anomaly detection (#374)
  • Add get_anomalies_isolation_forest method for anomaly detection (#375)
  • Add IForestOutlierTransform (#381)
  • Add IQROutlierTransform (#387)
  • Add num_workers parameter to TS2VecEmbeddingModel (#396)
  • Add get_anomalies_mad function for anomaly detection (#398)
  • Add TSDataset.features property to get list of all features in a dataset (#405)
  • Add MADOutlierTransform class for anomaly detection (#415)
  • Add MeanEncoderTransform (#413)
  • Add FourierDecomposeTransform transform for series decomposition using DFT (#430)
  • Add ModelDecomposeTransform transform for series decomposition using ETNA models (#427)

Changed

  • Allow to change device, batch_size and num_workers of embedding models (#396)
  • Update pipelines documentation (#408)
  • Update formulas for metrics in documentation (#406)
  • Update documentation to explain how to contribute and work with discussions, update templates for issues (#395)
  • Remove "Other issue" template, update links to discussions in issue creation menu (#401)

Fixed

  • Fix rendering in 210 tutorial (#386)
  • Fix typo in 103 tutorial (#408)
  • Remove sorting of ts.df by timestamps in plot_forecast and plot_forecast_decomposition (#410)
  • Fix forecast visualization with horizon=1 (#426)
  • Set upper bound <2 on numpy version (#431)
  • Fix VotingEnsemble, StackingEnsemble, DirectEnsemble have a valid params_to_tune that returns empty dict (#432)
  • Fix passing custom model to STLTransform (#412)
  • Update TSDataset.describe, TSDataset.info to exclude target intervals and target components in num_exogs (#405)
Assets 2
Loading