Skip to content

Releases: philipdarke/torchtime

v0.6.1

13 Jun 11:18
Compare
Choose a tag to compare

Fixed

  • Support for PyTorch 1.13.x to 2.0.x
  • GitHub Actions badge

v0.6.0

12 Jun 18:51
Compare
Choose a tag to compare

Changed

  • Updated dependencies

Added

  • Support for Python 3.10 and 3.11

v0.5.1

03 Aug 09:50
Compare
Choose a tag to compare

Changed

  • Adopt updated sktime version

v0.5.0

01 Aug 17:19
Compare
Choose a tag to compare

Changed

  • One-hot encoded PhysioNet2012 ICUType channel
  • Updated pre-commit hooks

Added

  • "zero" value imputation
  • Download progress bars

Fixed

  • overwrite_cache now re-downloads data
  • scikit-learn dependency

v0.4.2

06 Jul 18:57
Compare
Choose a tag to compare

Fixed

  • Continuous deployment

v0.3.0

25 Apr 11:27
Compare
Choose a tag to compare

Changed

  • More efficient PhysioNet data set downscaling
  • Updated documentation

Added

  • PhysioNet2012 data set
  • PhysioNet and UEA/UCR unit tests
  • Utility function module
  • Better console messages

Fixed

  • Replace PhysioNet2019 missing data indicator with NaNs
  • Code coverage badge

v0.2.0

08 Apr 16:43
Compare
Choose a tag to compare

Changed

  • Processed data are now cached in the .torchtime directory
  • train_split and val_split arguments are renamed train_prop and val_prop respectively
  • Introduced generic torchtime.data_TimeSeriesDataSet class behind the scenes - note training/validation/test data splits have changed for a given seed
  • torchtime.collate.packed_sequence now returns both X and y as a PackedSequence object
  • Expanded unit tests - note coverage is currently limited as PhysioNet2019 tests cannot be run under CI
  • Updated documentation

Added

  • impute argument to support missing data imputation using mean and forward imputation methods or a custom imputation function
  • downscale argument to reduce the size of data sets for testing/model development
  • torchtime.data.TensorTimeSeriesDataset class to create a data set from input tensors

Fixed

  • Use float32/torch.float and int64/torch.long precision for all data sets
  • Shape of y data in PhysioNet2019 data
  • Bug when adding time delta channels without a missing data mask

v0.1.1

31 Mar 15:35
Compare
Choose a tag to compare

Changed

  • Simplified training/validation/test split approach
  • Default file path for PhysioNet2019 data set is now data/physionet2019
  • Refactored torchtime.data to share utility functions across data classes
  • Expanded unit tests
  • Updated documentation

Added

  • Missing data simulation for UEA/UCR data sets
  • Support appending missing data masks and time delta channels
  • packed_sequence collate function
  • Documentation now includes a tutorial
  • Automated releases using GitHub Actions
  • DOI

v0.1.0

28 Mar 17:19
Compare
Choose a tag to compare

First release to PyPi