Python Implementation of Optimal Classification Trees.
MIT license. Contributions welcome!
Please have cvxpy installed with there installation guide. A MIP solver should be installed. Please refer to the guide here. Gurobi is recommended, now the repo only supports gurobi.
- Implement scikit-learn compatible APIs.
- Implement it fast (at the expense of memory).
- Implement APIs to interpret/visualize the model.
- Validate performance on 53 UCI ML datasets.
- Complete APIs as DecisionTreeClassifier from sklearn.
- Code refactor and add comments.
- APIs to interpret the model.
- A document to explain how it is implemented.
- Further improvement and experiments.
Now the model only has a naive fit/predict method, which should be validated.