Recent Releases of gval

gval - v0.2.7

This release addresses memory accumulation and promotion of numerical datatypes.

  • Pairing functions no longer automatically cast numeric datatypes to 64 bit
  • Changed engine of flox xarray_reduce to decrease memory footprint
  • Catalog comparisons no longer accumulate memory
  • Dask dependencies were upgraded and xarray-spatial was removed
  • Changed S3 data source for unit tests

- Jupyter Notebook
Published by GregoryPetrochenkov-NOAA over 1 year ago

gval - V0.2.5

Includes probabilistic statistical data type evaluations, updates to processing integers in continuous evaluations, STAC Client IO, NOX unit testing, and updates to documentation.

- Jupyter Notebook
Published by GregoryPetrochenkov-NOAA about 2 years ago

gval - v0.2.3

Includes functionality and patches for the following:

Catalog Comparisons Subsampling maps for metrics and agreement maps within or excluding regions of interest Adjusted workflow for performance and memory management Cleanup and bug fixes

- Jupyter Notebook
Published by GregoryPetrochenkov-NOAA over 2 years ago

gval - v0.1.2

Includes functionality and patches for the following:

  • Attribute tracking
  • Public rasterize and vectorize data functions
  • Inclusion of new categorical statistics
  • Cleanup and bug fixes

- Jupyter Notebook
Published by GregoryPetrochenkov-NOAA over 2 years ago

gval - v0.1.1

First functional build made for PyPI. Includes:

  • Two-class categorical, multicategorical, and continuous comparisons in raster space.
  • Functionality to create agreement maps, cross-tabulation tables, and both categorical and continuous statistical metrics.
  • Means to convert vector datatypes in to raster space and convert agreement map back to a vector format.
  • Custom classes to batch process agreement maps and statistics as well as allowing registration of custom functions for both.
  • Utilities for loading datasets from S3, visualization, and defining schemas for DataFrame types.
  • Memory optimization, memory and performance benchmarking, unit testing, and code linting.
  • Precommit hooks, pipeline builds, and means to push packages to PyPI.
  • GitHub.io documentation including Tutorials and an API for public facing code.

- Jupyter Notebook
Published by GregoryPetrochenkov-NOAA over 2 years ago