Extra Recipes for Encoding Predictors
Add WoE in a data frame
Weight of evidence dictionary
embed: Extra Recipes for Encoding Predictors
Objects exported from other packages
S3 methods for tracking which additional packages are needed for steps...
Supervised Collapsing of Factor Levels
collapse factor levels using stringdist
Discretize numeric variables with CART
Discretize numeric variables with XgBoost
Encoding Factors into Multiple Columns
Dummy Variables Creation via Feature Hashing
Supervised Factor Conversions into Linear Functions using Bayesian Lik...
Supervised Factor Conversions into Linear Functions using Likelihood E...
Supervised Factor Conversions into Linear Functions using Bayesian Lik...
Likelihood encoding using analytical formula
Sparse Bayesian PCA Signal Extraction
Sparse PCA Signal Extraction
Truncated PCA Signal Extraction
Supervised and unsupervised uniform manifold approximation and project...
Weight of evidence transformation
tunable methods for embed
Crosstable with woe between a binary outcome and a predictor variable.
Predictors can be converted to one or more numeric representations using a variety of methods. Effect encodings using simple generalized linear models <doi:10.48550/arXiv.1611.09477> or nonlinear models <doi:10.48550/arXiv.1604.06737> can be used. There are also functions for dimension reduction and other approaches.
Useful links