Make tests more modular. split data prep from analysis.

[aiqc]

Difficulty: ★☆☆☆☆

Background

tests.py has tests for each:

• data type (tabular, image)
• analysis type (binary classify, multi label classify, regression)
• library (pytorch, keras)

The make_test_batch_* function runs all of the low level api calls needed to prepare data to test that type of analysis.

Problem

When we add tensorflow, there will be 3 different libraries all making duplicate data preparation steps for make_test_batch_*

Solution

Make a functions that prepare low level api objects for:

• tabular classify binary
• tabular classify multi
• tabular classify regression
• image classify binary

So that they can be used by either keras, pytorch, or tensorflow