- Documentation of the derived heuristics and what hyperparameters mean (3mr).
- Logging unified
- Benchmark CI for core components as a regression test
- Gradual speedups/rewrites of main algorithms
- Documenting new/extra features (subfeatures etc.)
- Get rid of Pandas, let's use something more efficient instead
- Tree-based explanation for stream-like data