catboost v0.17.1 Release Notes

Release Date: 2019-09-13 // over 4 years ago
  • ๐Ÿ› Bugs fixed:

    • ๐Ÿ Incorrect estimation of total RAM size on Windows and Mac OS, #989
    • Failure when dataset is a numpy.ndarray with order='F'
    • Disable boost_from_average when baseline is specified

    ๐Ÿ‘Œ Improvements:

    • Polymorphic raw features storage (2x---25x faster data preparation for numeric features in non-float32 columns as either pandas.DataFrame or numpy.ndarray with order='F').
    • ๐Ÿ‘Œ Support AUC metric for CrossEntropy loss on CPU
    • โž• Added datasets.rotten_tomatoes(), a textual dataset
    • Usability of monotone_constraints, #950

    Speedups:

    • โšก๏ธ Optimized computation of CrossEntropy metric on CPUs with SSE3