dimarEvaluatePerformance: dimarEvaluatePerformance

In kreutz-lab/DIMAR: Data-driven selection of an imputation algorithm in R

dimarEvaluatePerformance

Description

Evaluates performance of imputation algorithms.

Usage

dimarEvaluatePerformance(
  Imputations,
  ref,
  sim,
  rankby = "RMSE",
  RMSEttest = TRUE,
  group = "cluster"
)

Arguments

Imputations	Imputed data set(s)
ref	Reference data
sim	Simulated patterns of MVs
rankby	Performance measure which should serve as rank criterion
RMSEttest	flag if RMSE of ttest should be calculated
group	indices for ttest

Value

Data frame containing the following performance measures for each imputation method: Deviation, RMSE, RSR, p-Value_F-test, Accuracy, PCC, and in case of RMSEttest=TRUE the RMSE t-test result

Examples

mtx <- matrix(rnorm(1000),nrow=100)
mtx[sample(c(1:1000),100)] <- NA
coef <- dimarLearnPattern(mtx)
ref <- dimarConstructReferenceData(mtx)
sim <- dimarAssignPattern(ref, coef, mtx)
Imputations <- dimarDoImputations(sim, c('impSeqRob', 'ppca', 'imputePCA'))
Performance <- dimarEvaluatePerformance(Imputations, ref, sim)

kreutz-lab/DIMAR documentation built on Aug. 19, 2024, 11:07 a.m.