5.8 Summary

With the exception of tree-based models, categorical predictors must first be converted to numeric representations to enable other models to use the information.

The simplest feature engineering technique is to convert each category to a separate binary dummy predictor.

Some models require one fewer dummy predictors than the number of categories.

Creating dummy predictors may not be the most effective way. If, for instance, the predictor has ordered categories, then polynomial contrasts may be better.

Text fields, too, can be viewed as an agglomeration of categorical predictors and must be converted to numerics.