Skip to content

CHANGELOG.md

Latest commit

 

History

History
79 lines (54 loc) · 3.72 KB

CHANGELOG.md

File metadata and controls

79 lines (54 loc) · 3.72 KB

Changelog

All notable changes to this repository are listed here.
Format follows Keep a Changelog.

[Unreleased]

[1.6.0] — 2026-04-27

Added

  • Topic 125 — CluStream: micro-cluster sufficient statistics (CF1x/CF2x/CF1t/CF2t/n), online point assignment with RMSD boundary, staleness-based slot management, offline weighted k-means macro-clustering, pyramidal time frame snapshot storage, time-window drift query, ARI vs q analysis, comparison with DenStream and batch k-means.

[1.5.0] — 2026-04-27

Added

  • Topic 118 — DenStream: density-based stream clustering with fading function, micro-cluster sufficient statistics, online phase, offline DBSCAN phase.
  • Topic 119 — ADWIN (deep dive): full ADWIN algorithm, abrupt/gradual drift detection, δ trade-off table, ADWIN as model-reset controller.
  • Topic 120 — VFDT & CVFDT: Hoeffding bound derivation, VFDT leaf implementation, CVFDT sliding-window adaptation, comparison with batch decision trees.
  • Topic 121 — Concept Drift Detection: DDM, EDDM, and Page-Hinkley Test; visual comparison and detector summary table.
  • Topic 122 — MONIC: cluster overlap computation, transition taxonomy (survive/split/merge/appear/disappear), overlap matrix heatmap, centroid trajectory tracking.
  • Topic 123 — SHAP Values: Shapley value theory, TreeExplainer, LinearExplainer, bar/beeswarm/waterfall/dependence plots, SHAP vs permutation importance.
  • Topic 124 — ARIMA & SARIMA: ADF stationarity test, ACF/PACF order selection, ARIMA baseline, SARIMA with seasonal terms, residual diagnostics, walk-forward validation.
  • shap>=0.46 and statsmodels>=0.14 added to requirements.txt.

[1.4.0] — 2026-04-27

Added

  • Topic 117 — Stream-Based & Adaptive Learning: Gama's adaptive learning definition, sufficient statistics on streams, recursive mean (Welford), ADWIN drift detection, online learning with SGDClassifier.partial_fit, Hoeffding Tree bound, and prequential evaluation.

[1.3.0] — 2026-04-26

Added

  • Topic 115 — End-to-End Real-World ML Project: Heart Disease Prediction capstone notebook covering EDA, preprocessing, PCA, SMOTE, multi-model benchmarking, hyperparameter tuning, ROC-AUC, and feature importance.
  • imbalanced-learn==0.12.4 added to requirements.txt.
  • nbstripout added to requirements.txt and installed as a git filter — notebook outputs are now stripped automatically on every commit.
  • GitHub Actions CI workflow (.github/workflows/ci.yml) — executes changed notebooks on every push/PR and uploads failed notebooks as artifacts.
  • Binder URL validation script (src/utils/check_binder_urls.py).
  • Difficulty tags (🟢 / 🟡 / 🔴) and prerequisite notes added to each README section.
  • First-Time Setup section added to CONTRIBUTING.md.

Fixed

  • Replaced fetch_openml with pandas.read_csv in topic 115 notebook to avoid OpenML redirect errors.

[1.2.0] — 2025-12-01

Added

  • Topics 108–114: Regression Evaluation Metrics, Linear Learning Machines, Kernel Function Matrix, Semi-Supervised Learning (SSL), Safe SSL, S3VM/TSVM, Softmax Activation Function.
  • Deep Learning section (topics 91–107): Introduction through LSTM.

[1.1.0] — 2025-06-01

Added

  • Topics 66–90: Association Rules through Case-Based Reasoning.
  • Active Learning section (topics 47–56).
  • Multi-label & Multi-class Classification section (topics 57–64).

[1.0.0] — 2025-01-01

Added

  • Initial release with topics 01–65.
  • Binder support for all notebooks.
  • requirements.txt, environment.yml, runtime.txt for reproducible environments.
  • CONTRIBUTING.md with branch strategy and naming conventions.