Popularity
4 Stars
Updated Last
6 Years Ago
Started In
December 2012

Loss.jl

General functions for estimating loss functions inspired by Kaggle's release of code for many common metrics. This package implements the full Cartesian product of two ways of classifying loss functions:

  • Single element loss function definitions:
    • Absolute deviation
    • Squared error
    • 0/1 loss
    • Hinge loss
    • Log loss
  • Aggregation mechanism across elements:
    • Mean
    • Root Mean
    • Median
    • Minimum
    • Maximum

To Do

  • Add other metrics
    • AUC
  • Add convenient abbreviations
    • rmse
    • ...