feat(forecasting): add baseline model zoo with security validations#30
Merged
Conversation
* docs: update INITIAL-4 and INITIAL-5 with additional references - Add scikit-learn, mlforecast, and sktime documentation links - Add considerations for imputation logic, agent tooling, and computation overhead - Add model persistence documentation references Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * feat(featuresets): implement time-safe feature engineering layer Add complete feature engineering module with: - Pydantic schemas for feature configuration (lag, rolling, calendar, exogenous, imputation) - FeatureEngineeringService with CRITICAL leakage prevention: - Lag features use positive shift() only - Rolling features use shift(1) BEFORE rolling to exclude current observation - Group-aware operations prevent cross-series leakage - Cutoff date filtering before any computation - FastAPI endpoints: POST /featuresets/compute and /featuresets/preview - Comprehensive test suite (55 tests) including leakage prevention tests - Example demo script Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * docs: update INITIAL-5.md * docs: update documentation for Phase 3 Feature Engineering - README.md: Add featuresets module to project structure and API endpoints - docs/ARCHITECTURE.md: Add Feature Engineering section (section 6) - docs/PHASE-index.md: Mark Phase 3 as completed with summary - docs/PHASE/3-FEATURE_ENGINEERING.md: Create detailed phase documentation Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * fix(featuresets): address code review feedback and prevent data leakage Routes: - Validate store_id/product_id presence (no silent defaults to 0) - Convert ValueError for unsupported date types to HTTP 400 Service: - Add expanding_mean imputation strategy (time-safe alternative) - Add warnings when bfill/mean strategies are used (leakage risk) - Fix price_pct_change_7d to use shift(1) before pct_change Schemas: - Add expanding_mean to ImputationConfig Literal type - Document time-safety of each imputation strategy - Fix PreviewFeaturesRequest docstring: GET → POST Documentation: - Convert bare URLs to markdown links in INITIAL-4.md, INITIAL-5.md - Fix PRP-4 to show POST for preview endpoint Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * style: format schemas.py Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> --------- Co-authored-by: Gabe@w7dev <gabor@w7-7.net> Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
Service: - Fix min_periods falsy check to explicit None check (preserves 0) Tests: - Add expanding_mean to test_valid_strategies, expect 6 strategies Documentation: - Update PR reference from #24 to #25 - Fix all GET /featuresets/preview to POST in PRP-4 Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
* docs: update DAILY-FLOW.md for Phase 4 Forecasting Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * docs: add PRP-5 for Forecasting module Comprehensive PRP including: - Model zoo (naive, seasonal naive, moving average) - Unified BaseForecaster interface (fit/predict/serialize) - ModelBundle persistence with joblib - 15 ordered implementation tasks - 40+ test cases specified - Integration with FeatureEngineeringService Confidence: 8/10 Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> --------- Co-authored-by: Gabe@w7dev <gabor@w7-7.net> Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
…#28) * feat(forecasting): implement baseline model zoo and unified interface Add forecasting module (PRP-5) with: - BaseForecaster ABC with scikit-learn-style interface (fit/predict) - NaiveForecaster, SeasonalNaiveForecaster, MovingAverageForecaster - ModelBundle persistence with joblib serialization - POST /forecasting/train and /forecasting/predict endpoints - ForecastingService for orchestration - 81 unit tests covering schemas, models, persistence, and service - Example scripts demonstrating each baseline model - LightGBM placeholder (feature-flagged, not yet implemented) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * docs: Update documentation for forecasting module (PRP-5) - Add forecasting API endpoints to README.md with examples - Update ARCHITECTURE.md with forecasting implementation details - Add scikit-learn and joblib to dependencies list - Add forecasting config variables to .env.example - Mark forecasting module as IMPLEMENTED in architecture docs Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * fix: address CI lint and type check failures - Add type: ignore for intentional type mismatch in frozen config test - Add S101 ignore for examples/ to allow assert statements Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> --------- Co-authored-by: Gabe@w7dev <gabor@w7-7.net> Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
Security improvements: - Add constructor validation for season_length >= 1 in SeasonalNaiveForecaster - Add constructor validation for window_size >= 1 in MovingAverageForecaster - Add path traversal prevention in ForecastingService.predict() - Validate .joblib extension and artifacts directory containment - Log rejection reasons for security auditing Test improvements: - Fix get_settings patching to wrap ForecastingService construction - Add tests for constructor validation - Add tests for path traversal and extension validation Documentation fixes: - Fix config parameter names in ARCHITECTURE.md (season_length, window_size) - Fix README example to use season_length instead of seasonal_period - Fix markdown issues in PRP-5 (code fences, ATX headings) Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
w7-learn
approved these changes
Feb 1, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
naive,seasonal_naive,moving_averageBaseForecasterinterface following scikit-learn conventionsModelBundlepersistence with joblib serializationPOST /forecasting/trainandPOST /forecasting/predictWhat's New
Models
NaiveForecaster- predicts last observed valueSeasonalNaiveForecaster- predicts value from same season in previous cycleMovingAverageForecaster- predicts mean of last N observationsSecurity
season_length >= 1andwindow_size >= 1.joblibextension validationDocumentation
season_lengthTest plan
🤖 Generated with Claude Code