ScrubR: ScrubR'

In womta/PurifyR: A R Package for Highly Automated Reproducible Variable Extraction and Standardization

Description

This function 'scrubs' the data, it massages the data in checking and fixing missing data, transformations, scaling and outlier removal or imputation

Usage

ScrubR(DT, predictors, rowWiseMissingPercentage, SDs, transformTable,
  scalingMethod, imputationMethod)

Arguments

DT	This is a data.table object
predictors	These are the clean predictors provided manually or from the output of ScanR
rowWiseMissingPercentage	This the the allowed threshold of the percentage missing data per row
SDs	This the the allowed threshold of standard deviations away from the mean
transformTable	This is the table received from ScanR that provides information what features and how they need to be transformed
scalingMethod	This is the way the data will be scaled
imputationMethod	This is the way missing data will be treated

Value

a data.table with all the preprocessed data

Examples

DT = data.table::data.table(plyr::baseball)
DT = LoadR(DT)
rs = scanr(DT, correlationCutoff = 0.99, SamplingPercentage=15, percentUniqueCutoff=5)
if (length(names(DT)) == length(unique(names(DT)))) {
data.table::setnames(DT, names(DT), rs$Predictors)
} else {
	names(DT) = rs$Predictors
}
predictors = rs[Status=='Healthy' | Status=='Uniform' ,Predictors]
transformTable = rs[abs(Skewness)>abs(selectedTransformationBoundary),
c('Predictors', 'Skewness'), with=F]
results = scrubr(DT, predictors, rowWiseMissingPercentage=100, SDs=8,
transformTable=transformTable, scalingMethod='robustZscore', imputationMethod='CWD')

womta/PurifyR documentation built on May 21, 2019, 11:11 a.m.