getCatCorrMatrix: Get Categorical Correlation Matrix

In dominikkirchhoff/CCKriging: Kriging for Mixed Continuous and Categorical Input Variables

Description

This function returns a matrix of correlations between the categorical level-combinations, which is needed for methods that split the correlation function in one product for the continuous and one for the categorical inputs (i.e., EC, MC, or UC).

Usage

getCatCorrMatrix(x, cat.type, par, q, design.corr = TRUE, info = TRUE)

Arguments

x	[data.frame] Categorical inputs of the design matrix.
cat.type	[character(1)] Which method should be used? Possible choices: “EC”: Exchangeable Correlation “MC”: Multiplicative Correlation “UC”: Hypersphere Decomposition-Based Unrestrictive Correlation “TMC”: Toeplitz Matrix Multiplication-Based Correlation
par	[numeric] Vector of the Kriging model's parameters. The first values belong to the continuous variables, the rest to the categorical variables.
q	[numeric] Number of continuous variables in the original design. This is needed in order to remove the unnecessary elements of the parameter vector par.
design.corr	[logical(1)] Should the n-by-n categorical correlation matrix of the design points be returned? If FALSE, the s-by-s matrix with correlations between the level combinations is returned, where s denotes the number of level combinations of the categorical variables. Default is TRUE.
info	[logical(1)] Should extra information be printed to the console? Default is TRUE.

Value

[matrix] The correlation matrix.

dominikkirchhoff/CCKriging documentation built on May 19, 2019, 4:05 p.m.