repVarclus: Variable Clustering Diagrams

In harrelfe/rreport: Research Reporting Package

Description

Generate variable clustering diagrams for each time period.

Usage

  repVarclus(data, time, times = sort(unique(time)),
    nmin = 10)

Arguments

data	data.frame. Data used for report.
time	numeric vector. Time for each record.
times	numeric vector. Subset of times to use.
nmin	numeric. Variables must have four responses or at least

Details

Variables are grouped according to how they are correlated with one another, as measured by the square of the Spearman ρ rank correlation coefficient computed on all pairs of variables. Variables connected on lower branches are more highly correlated with one another. Variables missing in more than 75 observations or categorical variables having more than 20 levels are ignored. Categories less than 0.1 prevalent are pooled with other rare categories.

Examples

## Not run: 
  dat <- data.frame(age=sample(41:80, replace=TRUE, 1000), height=round(rnorm(1000, 68, 4)))
  dat$weight <- sample(140:200, replace=TRUE, 1000)+(dat$height-60)*3
  repVarclus(dat, numeric(1000))

## End(Not run)

harrelfe/rreport documentation built on May 17, 2019, 2:48 p.m.