Table 2 Predictive accuracy measures and the corresponding R package in which they can be computed
Predictive accuracy measure | R package | Version |
|---|---|---|
(1) The confusion matrix | ||
Receiver Operating Characteristic Curve and area under the curve | prediction and performance in ROCR [11] performance (prediction.object, “auc”) | 1.0-7 |
Positive predictive value and negative predictive value | prediction and performance in ROCR performance (prediction.object, “ppv”) performance (prediction.object, “npv”) | 1.0-7 |
(2) Visualization of the distribution of prediction values | ||
Histograms of the prediction values separated by class | multhist in plotrix [12] | 3.5-11 |
Box plots | boxplot in graphics | Base package |
Violin plots | vioplot in vioplot | |
Quantile-quantile plots | qqplot in stats | Base package |
(3) Statistical tests | ||
Hypergeometric test | phyper in stats | Base package |
Mann–Whitney U test | wilcox.test in stats | Base package |
Asymptotic Generalized Cochran-Mantel-Haenszel Test | cmh_test in coin [13] | 1.0-24 |