neg_gmpa: Negative geometric mean of class specific predictive accuracy
In emil: Evaluation of Modeling without Information Leakage

Description Usage Arguments Value Author(s) References See Also

When dealing with imbalanced classification problem, i.e. where the class sizes are very different, small classes tend to be overlooked when tuning parameters by optimizing error rate. Blagus and Lusa (2013) suggested to remedy the problem by using this performance measure instead.

1	neg_gmpa(truth, prediction, na.rm = FALSE)

`truth`	See `error_fun`.
`prediction`	See `error_fun`.
`na.rm`	Whether to remove missing values or not.

A numeric scalar.

Christofer Bäcklin

Blagus, R., & Lusa, L. (2013). Improved shrunken centroid classifiers for high-dimensional class-imbalanced data. BMC bioinformatics, 14, 64. doi:10.1186/1471-2105-14-64

error_fun

emil documentation built on Aug. 1, 2018, 1:03 a.m.