Skip to content

v0.2.0: Elastic Net, Cross-validation, Model Persistence

Choose a tag to compare

View all tags
@kolkov kolkov released this 25 Nov 17:34
· 6 commits to main since this release
v0.2.0
aa4a4f3

What's New

Added

  • Elastic Net regularization with Alpha parameter (L1+L2 mix)
  • K-fold cross-validation for automatic lambda selection
  • Model serialization (Save/Load to JSON)
  • Input validation for NaN/Inf values
  • Comprehensive benchmarks
  • Project infrastructure (CI/CD, linting, documentation)

Changed

  • Refactored to sequential coordinate descent (better cache locality)
  • Optimized matrix access with RawMatrix() direct slice operations
  • Reduced per-iteration allocations with predictInto()
  • Lambda now scaled by nSamples for sklearn compatibility
  • Fit() returns error instead of panic
  • Updated to Go 1.25

Fixed

  • Population variance (n) instead of sample variance (n-1) for sklearn consistency

Installation

go get github.com/CausalGo/lasso@v0.2.0

Full Changelog: v0.1.0...v0.2.0

Assets 2
Loading