eval_func: Evaluation function, return the performance of simulation...

In VARDetect: Multiple Change Point Detection in Structural VAR Models

Evaluation function, return the performance of simulation results

Description

Evaluation function, return the performance of simulation results

Usage

eval_func(true_mats, est_mats)

Arguments

true_mats	a list of true matrices for all segments, the length of list equals to the true number of segments
est_mats	a list of estimated matrices for all simulation replications, for each element, it is a list of numeric matrices, representing the estimated matrices for segments

Value

A list, containing the results for all measurements

sensitivity

A numeric vector, containing all the results for sensitivity over all replications

specificity

A numeric vector, including all the results for specificity over all replications

accuracy

A numeric vector, the results for accuracy over all replications

mcc

A numeric vector, the results for Matthew's correlation coefficients over all replications

false_reps

An integer vector, recording all the replications which falsely detects the change points, over-detect or under-detect

Examples

true_mats <- vector('list', 2)
true_mats[[1]] <- matrix(c(1, 0, 0.5, 0.8), 2, 2, byrow = TRUE)
true_mats[[2]] <- matrix(c(0, 0, 0, 0.75), 2, 2, byrow = TRUE)
est_mats <- vector('list', 5)
for(i in 1:5){
    est_mats[[i]] <- vector('list', 2)
    est_mats[[i]][[1]] <- matrix(sample(c(0, 1, 2), size = 4, replace = TRUE), 2, 2, byrow = TRUE)
    est_mats[[i]][[2]] <- matrix(sample(c(0, 1), size = 4, replace = TRUE), 2, 2, byrow = TRUE)
} 
perf_eval <- eval_func(true_mats, est_mats)

VARDetect documentation built on May 10, 2022, 9:07 a.m.