Effective Amazon Machine Learning

Naming variables with assignments

The main purpose of assignments is naming facilitation. You can choose to name the transformed variable or group of variables in the assignments section or directly in the output section. Assignments are only for convenience and readability. Assignments follow this syntax:

"assignment_name": "transformation('group_name' )"

For instance, you could rename and normalize the numeric variables as follows:

"assignments": {    "normalized_numeric": "normalize(TO_BE_NORMALIZED)",}

Or rename and process the subject and body of your e-mails:

"assignments": {    "bigrams": "ngram(N_GRAM_TEXT,2)",}

You can also leave the assignments section empty and apply the transformations to the variables groups in the output section. ...