roc: Measures of Performance

In DMR: Delete or Merge Regressors for linear model selection.

Description

roc is used for calculating measures of performance such as sensitivity and specificity when the true and predicted models can be described using linear hypotheses.

Usage

 roc(Y0, X0) 

Arguments

Y0	matrix of linear constraints for the true model's parameters
X0	matrix of linear constraints for the predicted model's parameters

Value

a list including elements

sensitivity	proportion of the dimension of intersection of linear spaces of parameters for the true and predicted models to the dimension of linear space of parameters for the true model
specificity	proportion of the dimension of intersection of linear spaces that are complements of linear spaces of parameters for the true and predicted models to the dimension of linear space which is a complement of the linaer space of parameters for the true model

Author(s)

Piotr Pokarowski, Agnieszka Prochenka, Aleksandra Maj

See Also

DMR, stepDMR, DMR4glm, plot_bf

Examples

v1 <- factor(rep(1:3, times = 16))
set.seed(1)
x1 <- rnorm(48)
set.seed(2)
y <- x1 + rnorm(48, 0, 3)
m <- lm(y ~ v1 + x1)
out <- DMR(m) 
x0 <- out$Best$Hypotheses
y0 <- matrix(c(0, 0, -1, 1, 1, 0, 0, 0), 2, 4)
measures <- roc(y0, x0)
sen <- measures$sensitivity 
spe <- measures$specificity

DMR documentation built on May 30, 2017, 6:25 a.m.