rocCurves: A function to calculate and plot ROC curves

In melmasri/HPprediction: Host-parasite prediction based on phylogeny

Description

Computes the ROC curves as x = false positive rate (FPR) and y = true positive rate (TPR)

FPR = False Positives / (False Positives + True Negatives) TPR = True Positives / (True Positives + False Negatives)

Usage

rocCurves(Z.test, Z.train, P, plot = TRUE, bins = 400, all = FALSE)

Arguments

Z.test	the bipartite interaction matrix used for the test set
P	the posterior probability matrix output by sample_parameter
plot	TRUE/FALSE to plot the ROC curve.
bins	the number of bins that the interval (0,1) is divided into (default is 400)
all	TRUE/FALSE to calculate the ROC curve based on the whole dataset, or only the held-out portion
Z_est	the estimated bipartite interaction matrix used

Value

Returns: 'auc': the maximum AUC value 'threshold': the threshold where P > threshold has the maximum AUC value 'roc': a matrix containing the threthold, FPR, and TPR

melmasri/HPprediction documentation built on May 2, 2020, 11:09 a.m.