metrics: Computes several clasification metrics

In mAPKL: A Hybrid Feature Selection method for gene expression data

Description

This function calculates several classification related metrics. It uses the original and the predicted samples' labels to quantify the quality of the classification process. Those meassures give us a direct outlook of the selected "genes" and how well discriminate between two phenotypes.

Usage

    metrics(classLbls, predLbls)

Arguments

classLbls	The initial class labels.
predLbls	The predicted class labels.

Value

AUC	The Area Under the ROC curve as a degree of samples discrimination
Accuracy	The classification accuracy
MCC	The MCC classification meassure
Specificity	The degree of true negative's identification
Sensitivity	The degree of true positive's identification

Author(s)

Argiris Sakellariou

Examples

## Suppose 'val' represent the correct validation set labels
## and 'predictions' the predicted labels according to an SVM model

    val <- c(rep(0,8),rep(1,4))
    predictions <- c(rep(0,6),1,1,rep(1,3),0)
    perfMetrics <- metrics(classLbls=val, predLbls=predictions)

mAPKL documentation built on Nov. 8, 2020, 4:57 p.m.