Calculate pairwise association of data with mixed types of variables

Usage

pairwise_cor(df, var_type = NULL)

Arguments

df: dataframe with mixed types of variables
var_type: a character vector corresponding to types of variables in df. If not provided, will guess based on column classes.

Value

A visx_cor object (S3 class) containing:

cor_value: numeric matrix of pairwise association values
cor_type: character matrix of association types (spearman, pseudoR2, GKgamma)
cor_p: numeric matrix of p-values
var_type: character/factor vector of variable types
data: the original data.frame

Use print(), summary(), plot(), and as.data.frame() methods on the result.

Details

The following associated measures and tests are implemented dependent on variable type:

factor vs numeric, factor or ordinal: Pseudo R^2 and p value from multinomial regression

ordinal vs ordinal or numeric: GK gamma and GK gamma correlation test

numeric vs numeric: Spearman correlation and p value

Examples

data1 <- data.frame(x = rnorm(10),
 y = rbinom(10, 1, 0.5),
z = rbinom(10, 5, 0.5))
type1 <- c("numeric", "factor", "ordinal")
result <- pairwise_cor(data1, type1)
result
#> Pairwise associations for 3 variables (1 numeric, 1 factor, 1 ordinal)
#> 1 of 3 pairs significant at p < 0.05
#> 
#> Variables: x, y, z 
summary(result)
#> Correlation/Association Matrix
#> Significance: **** p<0.0001, *** p<0.001, ** p<0.01, * p<0.05
#> 
#>          y        z
#> x 0.34     0.28*   
#> y          0.05    
#>