3. advanced methods

[ashlinrichardson]

Implement convenient interfaces to, or occasionally directly implement, contemporary methods for general purpose use within R or python. Caveat: need to be robust to moderately large data. Must be accessible from R and python

• low-d embedding for visualization: isomap, tsne
• combinatorical pattern explo: correl and co-occur analysis
• unsuperv. classif: kmeans, dbscan, hdbscan, mean-shift clust, HAC
• superv. classif: c45 decision tree classifier, random forest classiifer
• encodings for categorial data: one hot, etc...