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
  • ...