DDoS-Detector v32 – AutoML Complete Pipeline

This release focuses on delivering a fully integrated AutoML pipeline with end-to-end search, stacking, and evaluation capabilities.
It introduces complete model search orchestration, hyperparameter optimization, automated stacking construction, and structured cross-validation workflows, while improving determinism and strengthening reproducibility across the DDoS-Detector framework.

These changes enhance the robustness of experimental pipelines while maintaining methodological consistency and improving maintainability across modules.

---

Key Features

End-to-End AutoML Search Pipeline

• Added run_automl_model_search to orchestrate automated model discovery.
• Implemented automl_objective for structured optimization logic.
• Standardized hyperparameter suggestion via suggest_hyperparameters_for_model.
• Improved reliability of model instantiation through create_model_from_params.
• Extended support for configurable search spaces with get_automl_search_spaces.

AutoML Cross-Validation and Evaluation Framework

• Added automl_cross_validate_model for structured cross-validation.
• Implemented evaluate_automl_model_on_test for post-search validation.
• Fixed return type to ensure Python float output from cross-validation results.
• Improved error handling for:
  • invalid hyperparameter combinations,
  • inconsistent scoring outputs,
  • incompatible model configurations.
• Enhanced cross-module consistency across:
  • stacking,
  • evaluation,
  • optimization workflows.

Automated Stacking Search and Optimization

• Added run_automl_stacking_search for automated stacking optimization.
• Implemented automl_stacking_objective for meta-model evaluation.
• Added build_automl_stacking_model for dynamic stacking construction.
• Strengthened integration between:
  • base model search,
  • stacking meta-model optimization.
• Reduced runtime inconsistencies in long-running experiments.

Architecture and Refactoring

• Standardized internal interfaces and configuration flow.
• Removed fragmented AutoML logic in favor of a unified search structure.
• Renamed and clarified objective and search-related functions.
• Consolidated search, evaluation, and stacking orchestration within the stacking module.
• Improved separation of concerns between:
  • preprocessing,
  • feature selection,
  • optimization,
  • stacking,
  • evaluation,
  • reporting.

Dataset and Pipeline Enhancements (if applicable)

• Improved dataset handling during cross-validation loops.
• Standardized scoring aggregation across AutoML evaluations.
• Enhanced determinism in model ranking and selection.
• Reduced variability in repeated search executions.
• Strengthened reliability of test-set validation procedures.

Tooling and Infrastructure (if applicable)

• Enhanced extensibility for future model families and search strategies.
• Improved structural clarity of optimization workflows.
• Strengthened resource handling within iterative search loops.
• Reduced type-related inconsistencies in evaluation returns.
• Improved compatibility with existing stacking infrastructure.

Documentation Updates

• Updated stacking module documentation to reflect full AutoML integration.
• Clarified search-space definition and objective responsibilities.
• Improved inline documentation for optimization and stacking functions.
• Enhanced maintainability through clearer orchestration structure.
• Updated usage patterns for complete AutoML experimentation workflows.

---

Impact

This release improves the scalability, reproducibility, and optimization capabilities of the DDoS-Detector framework.

By implementing a complete AutoML pipeline with structured search, cross-validation, and automated stacking optimization, the system becomes more predictable, easier to maintain, and more resilient during large-scale experimental executions.

These changes strengthen the overall framework foundation without altering the experimental methodology, ensuring safer and more reproducible long-running machine learning pipelines.

---

Full Changelog: v31-AutoMLFoundation...v32-AutoMLCompletePipeline