Distributable feature analysis in model training system

Abstract

A computing system defines transformed variable values for training a machine learning model. A data description is determined for each variable of a plurality of variables from observation vectors. A number of rare-levels is determined for any variable of the plurality of variables that has a nominal variable type. Bins that describe a cumulative distribution function are defined for each variable based on the data description determined for each variable and based on the number of rare-levels determined for any variable of the plurality of variables identified as the nominal variable type. A transformed value is determined for each variable and for each observation vector of the observation vectors using the bins defined for a respective variable of the plurality of variables. Each determined transformed value is written to a transformed dataset with a respective observation vector of the observation vectors.