Model Calibration
Loved this definition:
Generally, for any classification problem, we predict the class value that has the highest probability of being the true class label. However, sometimes, we want to predict the probabilities of a data instance belonging to each class label
So the main object of using model calibration is to have a more precise probability related with what we want
Types
- Platt Scaling
- Isotonic Regression
- Spline Calibration
- Temperature Scaling
Links
- https://www.geeksforgeeks.org/calibration-curves/?utm_source=pocket_reader
- https://towardsdatascience.com/a-comprehensive-guide-on-model-calibration-part-1-of-4-73466eb5e09a
- https://geoffpleiss.com/nn_calibration#:~:text=What%20is%20Temperature%20Scaling%3F,a%20learned%20scalar%20parameter%2C%20i.e.
- https://www.youtube.com/playlist?list=PLeVfk5xTWHYBw22D52etymvcpxey4QFIk
- On Calibration of Modern Neural Networks
- https://colab.research.google.com/github/nplan-io/kdd2020-calibration/blob/master/tutorial/KDD%202020%20-%20nPlan%20calibration%20session%20(completed).ipynb