In this issue:

• we discuss Pre-Train Model Testing; 

• we overview the pillars of robust machine learning;

• we explore Great Expectations.

Enjoy the learning!  

💡 ML Concept of the Day: Pre-Train Model Testing 

In the previous edition of our machine learning (ML) testing series (Edge#213), we discussed the main methods for testing trained ML models. Testing models post-training is the most common approach in ML pipelines but certainly not the only one. Pre-train tests are another type of testing method targeted to identify potential errors in ML models prior to executing expensive training jobs.  

Typically, pre-train model testing is targeted to validate the correct composition of the datasets and some key verifications in the model behavior. Without attempting to provide an exhaustive list, there are some pre-train testing tasks that are regularly present in ML testing workflows.  ...