Groups enable grouping of multiple variables to facilitate applying the same transformations to several variables. The groups section of the recipe has a naming function. Group definition follows this syntax:
"group_name": "group('first_variable','second_variable' )"
Amazon ML has defined a set of default groups based on the type of the variables: ALL_TEXT, ALL_NUMERIC, ALL_CATEGORICAL, ALL_BINARY, and the ALL_INPUTS group for all the variables at once. Let's look at a couple of examples.
Consider the following example where we want to apply the same transformation (normalization) on the age and fare variables. We can define a group and name it TO_BE_NORMALIZED:
"groups" : { "TO_BE_NORMALIZED" : "group('age','fare')", ...