Is it obligatory converting categorical data to numerical data to use interpret (Microsoft package)?

Question

我是开源 InterpretML Python 包的新手，我想知道是否有必要将分类数据转换为数字数据以便使用任何玻璃/黑匣子，谢谢！

Answer 1

For training EBMs (Explainable Boosting Machines), which are glassbox models, you do not need to convert categorical data to numeric. EBM can accept Pandas dataframes containing a mixture of types (numeric and strings representing categoricals), and will handle conversions implicitly.

For other methods in InterpretML, including blackbox methods, you will need to handle conversions yourself at this time (through a technique like One Hot Encoding ).