There are implementation not working as expected, so further debug is needed.
- Boltzmann machine and DBN
- Apriori
- DBSCAN
- FP-Growth
- Partitioning Around Medoids
Feel free to use the frame.jl to develop your version of ML algorithm.
- Eliminate ForwardDiff dependency, this involves linear regression code and should be easy to do.
- Eliminate PyCall dependency, this involves eliminating the dependency on python scikit-learn's datasets for testing purpose.