predictvo: Predict the relation of objects to a list of variables of...

In nevrome/varnastats: Analyse some aspects of prehistoric burial sites

Description

By comparison of significant correlations within variables and the variables the objects incorporate predictvo() makes an prediction about the relation to given variables for every object.

Usage

predictvo(matrix, reltable, mvars, level = 2)

Arguments

matrix	data.frame with numeric values
reltable	table of correlation values (e.g. produced by reltable())
mvars	vector of variables of interest (full name)
level	switch to define if the prediction should be based on level 1 or on level 1 + level 2 variables. A level 1 variable is directly linked to a variable of interest, a level 2 variable is linked to the level 1 variables of said variable of interest 1: prediction is based only on level 1 variables. 2 (default): prediction is based on level 1 + level 2 variables.

Value

table with predicted, normalized relation values of every object and given variables of interest. If no variable of interest has correlations to other variables, predictvo() returns FALSE.

Examples

testmatrixrand <- data.frame(
   matrix(base::sample(0:1,400,replace=TRUE), nrow=20, ncol=20)
)

testcorr <- corrmat(testmatrixrand, "chi2", chi2limit = 0.1, dim = 1)

rel <- reltable(testcorr)

predictvo(testmatrixrand, rel, c("X2", "X3"))

nevrome/varnastats documentation built on May 9, 2019, 10:43 a.m.