- The inclusion of an upper bound on the error rate of Adaboost: crucial, because the error rate at each iteration has to be at least as good as random guess’.
- New quasi-randomized networks models for regression and classification, with two shrinkage parameters (for model regularization).
The full list of changes can always be found here on Github and a notebook describing some of the new models (for classification) here for 4 datasets (with a snippet below on a wine classification dataset).
Contributions/remarks are welcome as usual, you can submit a pull request on Github.
Note: I am currently looking for a gig. You can hire me on Malt or send me an email: thierry dot moudiki at pm dot me. I can do descriptive statistics, data preparation, feature engineering, model calibration, training and validation, and model outputs’ interpretation. I am fluent in Python, R, SQL, Microsoft Excel, Visual Basic (among others) and French. My résumé? Here!