This is an attempt to use an entropy-based decision tree to solve the Titanic challenge at Kaggle.com.
The implementation is a pretty standard one, with one exception: it uses a columnar dataset. The CSV is lazily loaded into colums instead of reading it simply as a set of vectors. This makes entropy computation pretty fast.
Refer to src/kaggle_titanic/example.clj
for an overview of how to use the thing.