Problem
A production loan model has started declining applications it used to approve, and a regulator wants a straight answer: exactly what data, code, and parameters produced the model making these decisions. Nobody on the team can answer. Runs happen roughly 200 times a week, models get promoted, and the connection between a live production model and the run that actually produced it evaporates somewhere between training and deployment.
Required components
Click to add a component to the canvas.
Additional components
Browse by category. Click or drag a component onto the canvas.