corrmat: Create a correlation matrix of an input data.frame

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

Description

corrmat returns a correlation matrix of a data.frame. Several different correlation methods can be choosen and the matrix can be created for column or row relations.

Usage

corrmat(matrix, method = "chi2", dim = 1, chi2limit = 0.05,
  rmnegniv = 0)

Arguments

matrix	data.frame with numeric values
method	switch to define which contingency value should be used: "chi2" (default): test decision of the chi-square test for a defined decision niveau. A significant relation of two variables/objects is marked with a numeric "1", a negativ test result with a numeric "0". "phi": Pearson's phi coefficient ("mean square contingency coefficient"). "cc": Pearson's contingency coefficient. "lambda": Goodman and Kruskal's lambda value. Mean of both values calculated depending on what's the dependent and what's the independent variable
dim	switch to define if the correlation matrix should be created for columns or rows. 1 (default): table is created for column (variables) relations. 2: table is created for row (objects) relations.
chi2limit	significance level for the test decision. Just relevant for method "chi2". The higher chi2limit the less results will get removed. default: 0.05 -> 5%
rmnegniv	option allows to remove "negativ relations". If >0 rmnegcorr gets called and applied. A usual value for rmnegniv is 0.1.

Details

The rmnegniv option allows to remove "negativ relations", by activating the rmnegcorr function for values >0. The smaller rmnegniv the weaker the overlap of two variables/objects can be, to still be recognised as the cause of a positiv relation.

See ?rmnegcorr for further info. This function can also be applied later.

Value

correlation matrix

Examples

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

corrmat(testmatrixrand, "chi2", chi2limit = 0.03)

corrmat(matrix = testmatrixrand, method = "lambda", dim = 2)

phicorrtable <- corrmat(
 matrix = testmatrixrand,
 method = "phi",
 dim = 1
 )

# Without negative relations:
phicorrtablewnr <- corrmat(
 matrix = testmatrixrand,
 method = "phi",
 dim = 1,
 rmnegniv = 0.1
 )

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