Predict Your Model’s Performance (Without Waiting for the Control Group)

A novel algorithm by NannyML allows estimating the performance of an ML model before the ground truth is available. Here is how it works.

So, your predictive model is up and running. You have developed, tested and finally deployed it. When presenting it to your stakeholders, you mention the model achieved 75% precision and 25% recall on backtesting data. And they go: “But what is the performance on currently active customers?”. The point is that: …

data science hands-on-tutorials machine learning machine learning engineer mlops performance