To my understanding, C4.5 comes with 4 improvements compared to ID3:
- Handling missing values in both training data and "test" data,
- Handling continuous data
- Handling costs on attributes.
- The pruning
However, not one of all decision tree python modules that I found, even the so-called C4.5, handles missing values.
Do you know of a library that handles them ? And it would be great if it can work with Pandas DataFrames "out-of-the-box"...