Mathias Brandewinder

When decision tree encounters unknown case, it crashes. It should probably return the most likely branch in that case.

I suspect speed of classification can be improved by iterating over the shortest of words in text and classification tokens, instead of checking for the presence of every token.

classifier function doesn't need to re-compute estimator.

Paks' code includes some performance improvements in Decision Trees, using Maps; include these in the code. https://github.com/paks/Machine-Learning-In-Action/network

On the left-hand side, there are two buttons, an arrow and a cercle:  So far I have not been able to figure out how they are different. I initially...