Group correlated features
Usage
group_correlated_features(
features,
corr_thresh = 1,
group_neg_corr = TRUE,
corr_method = "spearman"
)
Arguments
- features
a dataframe with each column as a feature for ML
- corr_thresh
For feature importance, group correlations above or equal to
corr_thresh
(range0
to1
; default:1
).- group_neg_corr
Whether to group negatively correlated features together (e.g. c(0,1) and c(1,0)).
- corr_method
correlation method. options or the same as those supported by
stats::cor
: spearman, pearson, kendall. (default: spearman)
Author
Kelly Sovacool, sovacool@umich.edu
Examples
features <- data.frame(
a = 1:3, b = 2:4, c = c(1, 0, 1),
d = (5:7), e = c(5, 1, 4), f = c(-1, 0, -1)
)
group_correlated_features(features)
#> [1] "a|b|d" "c|f" "e"