[Autoloop: build-tsikit-learn-scikit-learn-typescript-migration] #7

tsikit-learn: scikit-learn → TypeScript Migration

🤖 This PR is maintained by Autoloop. Each accepted iteration adds a commit to this branch.

Program Goal: Build tsikit-learn, a complete TypeScript port of scikit-learn, one feature at a time.

Issue: #5

Current Best Metric: sklearn_features_ported = 11 (up from 2)

Iteration 2 Summary

This iteration establishes the complete project foundation and ports Phase 1 (Foundation) and Phase 2 (Preprocessing & Metrics) of the migration plan:

Project infrastructure:

• package.json, tsconfig.json (strictest settings), biome.json, bunfig.toml
• .github/workflows/ci.yml — CI with Bun test, typecheck, GitHub Pages deploy
• AGENTS.md — agent instructions and project conventions
• playground/index.html — interactive feature roadmap showing ported vs pending modules

Phase 1 — Foundation (6 modules):

• src/exceptions.ts — NotFittedError, ConvergenceWarning, DataConversionWarning, etc.
• src/base.ts — BaseEstimator, ClassifierMixin, RegressorMixin, TransformerMixin, ClusterMixin, Matrix2D interface, checkIsFitted, clone
• src/utils/extmath.ts — rowNorms, softmax, logLogistic, safeSparseDot, matVecDot, colMean, colVar, logSumExp
• src/utils/validation.ts — checkArray, checkXy, checkNFeatures, isMatrix2D, validateSampleWeight, columnMinMax
• src/utils/multiclass.ts — typeOfTarget, uniqueLabels, labelBinarize, nClasses
• src/utils/class_weight.ts — computeClassWeight, computeSampleWeight

Phase 2 — Preprocessing & Metrics (5 modules):

• src/preprocessing/standard_scaler.ts — StandardScaler (fit, transform, inverseTransform, fitTransform)
• src/preprocessing/minmax_scaler.ts — MinMaxScaler (configurable feature range)
• src/preprocessing/label_encoder.ts — LabelEncoder, Normalizer (L1/L2/max)
• src/metrics/regression.ts — MSE, RMSE, MAE, R2, MAPE, explainedVariance, medianAbsoluteError
• src/metrics/classification.ts — accuracy, confusionMatrix, precision, recall, F1, logLoss, zeroOneLoss
• src/model_selection/index.ts — trainTestSplit (seeded shuffle), KFold

Tests: 25+ test cases covering all ported modules with property-based assertions.

Metric: sklearn_features_ported: 2 → 11 (+9)
Actions run: §25721028611

network:
  allowed:
    - defaults
    - "releaseassets.githubusercontent.com"

Generated by Autoloop · ● 42.7M · ◷

To install this agentic workflow, run

gh aw add .github/workflows/autoloop.md

To create a pull request with the changes:

# Download the artifact from the workflow run
gh run download 25721028611 -n agent -D /tmp/agent-25721028611

# Fetch the bundle into a local branch
git fetch /tmp/agent-25721028611/aw-autoloop-build-tsikit-learn-scikit-learn-typescript-migration.bundle refs/heads/autoloop/build-tsikit-learn-scikit-learn-typescript-migration:refs/heads/autoloop/build-tsikit-learn-scikit-learn-typescript-migration
git checkout autoloop/build-tsikit-learn-scikit-learn-typescript-migration

# Push the branch to origin
git push origin autoloop/build-tsikit-learn-scikit-learn-typescript-migration

# Create the pull request
gh pr create --title '[Autoloop: build-tsikit-learn-scikit-learn-typescript-migration]' --base main --head autoloop/build-tsikit-learn-scikit-learn-typescript-migration --repo githubnext/tsikit-learn