metan (version 1.2.1)

can_corr: Canonical correlation analysis

Description

Performs canonical correlation analysis with collinearity diagnostic, estimation of canonical loads, canonical scores, and hypothesis testing for correlation pairs.

Usage

can_corr(
  .data = NULL,
  FG = NULL,
  SG = NULL,
  by = NULL,
  means_by = NULL,
  use = "cor",
  test = "Bartlett",
  prob = 0.05,
  center = TRUE,
  stdscores = FALSE,
  verbose = TRUE,
  collinearity = TRUE
)

Arguments

.data

The data to be analyzed. Must be a dataframe containing the numeric variables that will be used in the estimation of the correlations. The data can also be passed directly by the arguments FG and SG. Alternatively, .data may be passed from the function split_factors. In such case, the canonical correlation will be estimated for each level of the grouping variable in that function.

FG, SG

If a dataframe is informed in .data, then FG and SG is a comma-separated list of unquoted variable names that will compose the first (smallest) and second (highest) group of the correlation analysis, respectively. Select helpers are also allowed.

by

One variable (factor) to split the data into subsets. The function is then applied to each subset and returns a list where each element contains the results for one level of the variable in by. To split the data by more than one factor variable, use the function split_factors to pass subsetted data to .data.

means_by

The argument means_by is a grouping variable to compute the means by. For example, if means_by = GEN, then the means of the numerical variables will be computed for each level of the grouping variable GEN, and the canonical correlation analysis will be computed using these means.

use

The matrix to be used. Must be one of 'cor' for analysis using the correlation matrix (default) or 'cov' for analysis using the covariance matrix.

test

The test of significance of the relationship between the FG and SG. Must be one of the 'Bartlett' (default) or 'Rao'.

prob

The probability of error assumed. Set to 0.05.

center

Should the data be centered to compute the scores?

stdscores

Rescale scores to produce scores of unit variance?

verbose

Logical argument. If TRUE (default) then the results are shown in the console.

collinearity

Logical argument. If TRUE (default) then a collinearity diagnostic is performed for each group of variables according to Olivoto et al.(2017).

Value

If .data is an object of class split_factors then the results will be returned into a list where each element has the following values.

• Matrix The correlation (or covariance) matrix of the variables

• MFG, MSG The correlation (or covariance) matrix for the variables of the first group or second group, respectively.

• MFG_SG The correlation (or covariance) matrix for the variables of the first group with the second group.

• Coef_FG, Coef_SG Matrix of the canonical coefficients of the first group or second group, respectively.

• Loads_FG, Loads_SG Matrix of the canonical loadings of the first group or second group, respectively.

• Score_FG, Score_SG Canonical scores for the variables in FG and SG, respectively.

• Crossload_FG, Crossload_FG Canonical cross-loadings for FG variables on the SG scores, and cross-loadings for SG variables on the FG scores, respectively.

• SigTest A dataframe with the correlation of the canonical pairs and hypothesis testing results.

• collinearity A list with the collinearity diagnostic for each group of variables.

References

Olivoto, T., V.Q. Souza, M. Nardino, I.R. Carvalho, M. Ferrari, A.J. Pelegrin, V.J. Szareski, and D. Schmidt. 2017. Multicollinearity in path analysis: a simple method to reduce its effects. Agron. J. 109:131-142. doi:10.2134/agronj2016.04.0196. 10.2134/agronj2016.04.0196

Examples

# NOT RUN {
library(metan)

cc1 <- can_corr(data_ge2,
               FG = c(PH, EH, EP),
               SG = c(EL, ED, CL, CD, CW, KW, NR))

cc2 <- can_corr(FG = data_ge2[, 4:6],
                SG = data_ge2[, 7:13],
                verbose = FALSE,
                collinearity = FALSE)

# Canonical correlations for each environment
cc3 <- data_ge2 %>%
       can_corr(FG = c(PH, EH, EP),
                SG = c(EL, ED, CL, CD, CW, KW, NR),
                by = ENV,
                verbose = FALSE)


# }