mcca

This is a package for Multi-category Diagnostic Accuracy.

Installation

The package is available on CRAN and can be installed directly in R using install.packages(). You may want to run install_formats() after the first installation.

R install.packages("mcca") install_formats()

The latest development version on GitHub can be installed using:

R if (!require("remotes")){ install.packages("remotes") } remotes::install_github("gaoming/mcca")

Description

It contains six common multi-category classification accuracy evaluation measures:

Hypervolume Under Manifold (HUM), described in Li and Fine (2008) doi:10.1093/biostatistics/kxm050.

Correct Classification Percentage (CCP), Integrated Discrimination Improvement (IDI), Net Reclassification Improvement (NRI), R-Squared Value (RSQ), described in Li, Jiang and Fine (2013) doi:10.1093/biostatistics/kxs047.

Polytomous Discrimination Index (PDI), described in Van Calster et al. (2012) doi:10.1007/s10654-012-9733-3. Li et al. (2018) doi:10.1177/0962280217692830.

We have published a paper to describe all the above measures and mcca package:

Li J, Gao M, D'Agostino R. Evaluating classification accuracy for modern learning approaches. Statistics in Medicine. 2019;1–27. https://doi.org/10.1002/sim.8103

Citation

Li J, Gao M, D'Agostino R. Evaluating classification accuracy for modern learning approaches. Statistics in Medicine. 2019;1–27. https://doi.org/10.1002/sim.8103

Demo

Here is a little demo.

```r

install.packages("nnet")

install.packages("rpart")

install.packages("e1071")

Basic Setup

install.packages("MASS")

install.packages("devtools")

library(devtools) install_github("gaoming96/mcca") library(mcca)

Help File

?hum

Example 1

rm(list=ls()) str(iris) data <- iris[, 1:4] label <- iris[, 5] hum(y = label, d = data, method = "multinom", k = 3)

[1] 0.9972

hum(y = label, d = data, method = "tree", k = 3)

[1] 0.998

hum(y = label, d = data, method = "mlp", k = 3)

[1] 0.9948

pdi(y = label, d = data, method = "mlp", k = 3)

[1] 0.9976

Example 2

rm(list=ls()) str(iris) data <- data.matrix(iris[, 1:4]) label <- as.numeric(iris[, 5])

multinomial

require(nnet)

model

fit <- multinom(label ~ data, maxit = 1000, MaxNWts = 2000) predict.probs <- predict(fit, type = "probs") pp<- data.frame(predict.probs)

extract the probablity assessment vector

head(pp)

X1 X2 X3

1 1.0000000 7.813676e-11 1.208325e-37

2 1.0000000 2.894707e-08 2.068738e-33

3 1.0000000 2.961086e-09 3.547252e-35

4 0.9999996 3.533713e-07 6.964265e-32

5 1.0000000 5.609367e-11 5.690705e-38

6 1.0000000 2.489396e-11 8.327501e-37

hum(y = label, d = pp, method = "prob", k = 3)

[1] 0.9972

the same result as the first example in Example 1

Example 3

rm(list=ls()) str(iris) data <- iris[, 3] label <- iris[, 5] pdi(y = label, d = data, method = "tree", k = 3)

[1] 0.9082667

pdi(y = label, d = data, method = "mlp", k = 3)

[1] 0.9845333

```