vblpcmroc: ROC curve plot for vblpcmfit

In VBLPCM: Variational Bayes Latent Position Cluster Model for Networks

ROC curve plot for vblpcmfit

Description

Plot a Receiver Operating Curve to show model fit in terms of link prediction.

Usage

vblpcmroc(v.params, NUM=100)

Arguments

v.params	The fitted values; output from vblpcmfit()
NUM	The number of intervals on the roc curve

Details

A threshold is varied between zero and one. At each point the probability of a link between all pairs of nodes is calculated on the v.params argument containing a fitted vblpcm object. If greater than the threshold the link is "predicted" present, else it is "predicted" absent. A plot of the proportion of true and false positives for each threshold value is thus obtained.

Value

The Area Under the Curve (AUC). The closer to 1 the better the fit.

Author(s)

Michael Salter-Townshend

VBLPCM documentation built on March 31, 2023, 9:21 p.m.