A data scientist uses Platt scaling to calibrate a gradient boosting model's output. What does Platt scaling specifically fit to the raw model scores?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Here's the deal — b is correct because Platt scaling fits a two-parameter logistic sigmoid function to the model's raw scores using a calibration dataset, producing probabilities that better reflect true event rates. A describes isotonic regression calibration (non-parametric).
Full explanation below image
Full Explanation
B is correct because Platt scaling fits a two-parameter logistic sigmoid function to the model's raw scores using a calibration dataset, producing probabilities that better reflect true event rates. A describes isotonic regression calibration (non-parametric). C describes Beta calibration. D describes temperature scaling — a single-parameter method used mainly with neural networks.