Skip to content

v0.2.0

Choose a tag to compare

@ethqnol ethqnol released this 27 Sep 19:44
· 10 commits to main since this release
73ee112

What's Changed

  • Support Vector Machines (SVM): Added LinearSVC (L1/squared hinge loss classification via Dual Coordinate Descent) and LinearSVR (epsilon-insensitive loss regression with active-set shrinking).
  • Spatial Clustering: Added DBSCAN with KDTree radius queries, core sample indexing, and non-parametric noise rejection.
  • Naive Bayes Classifiers: Added BernoulliNB (multivariate binary features) and ComplementNB (imbalanced text classification).
  • Benchmarks & Parity: Comprehensive benchmark suite against Scikit-Learn (1.1x–3.0x speedups with exact mathematical parity).
  • CI & Tooling: Modernized GitHub Actions workflow with parallel test execution.