Explaining Titanic hypothesis with decision trees
A common argument against linear classifiers and against statistical learning methods is that it is difficult to explain how the built model decides its predictions for the target classes. If you have a highly dimensional SVM, it is impossible for a human being to even imagine how the hyperplane built looks like. A Naïve Bayes classifier will tell you something like: "this class is the most probable, assuming it comes from a similar distribution as the training data, and making a few more assumptions" something not very useful, for example, we want to know why this or that mail should be considered as spam.
decision trees are very simple yet powerful supervised learning methods, which constructs ...
Get scikit-learn : Machine Learning Simplified now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
