[MADLIB-1358] DL: make types consistent

• Type: Improvement
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Fix Version/s: v1.16
• Assignee: Unassigned
• Reporter: Nikhil
• Created: 13/Jun/19 17:50
• Updated: 19/Jun/19 19:14
• Resolved: 19/Jun/19 19:14

Description

Under deep_learning module (this includes preprocessor, keras model arch table, fit, evaluate and predict):

1. We would like for independent_var to be float32 (REAL[]) everywhere.
2. We would like for dependent_var to be a 16-bit integer (SMALLINT) everywhere.

This will require changing schemas in CREATE TABLES, and argument types to UDA functions. We should also create a function to wrap `np.array()` that automatically uses `dtype=np.float32`

• Add dev-check assert for types of class_values, dependent_var, independent_var, and model_weights.