Data Discretization for Risk Score Models

About this Repository

discretize.R contains the main function, discretize().

example_usage.Rmd decribes the functionality and usage of discretize().

test_adherence.Rmd and test_discretize.Rmd are test files used during the development of discretize() to compare performance between risk_mod() models using more involved preprocessing and discretization, respectively.

Future Improvements

• Perform more error checking for arguments to discretize(). For example, check that threshold is numeric, or check that all strings in continuous_cols are indeed column names in X.
• Implement a method to determine the exact range of values in a bucket. For example, if a column called age was split into 3 buckets, we would be able to see the precise ranges that each bucket represented (say, <18, 18-24, 25+).