dummify_input_tables: Turn response column levels into dummy variables.
In RJ333/phyloseq2ML: Connects Phyloseq To Ranger And Keras

View source: R/keras_prepare_functions.r

The factor columns of an input table must be turned into dummy variables so keras can work with it. This function detects factor columns and turns them into dummy columns. First, as many columns are created as there are factor levels in the origin column, so a 2-level factor column would be represented by two complementary columns consisting of 0s and 1s. However, these are dependent on each other, which is known as the dummy trap. Therefore, the first of a given number of dummy columns per factor column is removed. If no factor columns are detected, the list will be returned unmodified along with an info. The response column is ignored, as it will be modified later using 'keras::to_categorical()'.