feat(forecast): add Prophet-like additive forecasting model (#248)

[w7-mgfcode]

Summary

Adds ProphetLikeForecaster — a deterministic, feature-aware additive linear forecasting model (MLZOO-C2). It is a scikit-learn Pipeline of SimpleImputer(median) + Ridge(solver="cholesky") over the canonical 14-column feature frame, plus a model-specific decompose() method that splits any forecast into its additive trend / seasonality / holiday-regressor contributions.

This is one of two MLZOO-C review units — the sibling PRPs/PRP-MLZOO-C1-xgboost-model.md ships the XGBoost model on a separate branch. The two are additive and order-independent.

Naming honesty: the model is "Prophet-like", not Prophet. It approximates Prophet's additive decomposition shape with a linear model over already-engineered features. It does not add the real prophet/Stan dependency and does not model changepoint trend, posterior uncertainty intervals, or automatic seasonality discovery. Every docstring and doc section states this plainly.

What changed

• ProphetLikeModelConfig (alpha + feature_config_hash) added to forecasting/schemas.py and the ModelConfig union.
• ProphetLikeForecaster + _PROPHET_LIKE_COMPONENTS constant + ForecastDecomposition dataclass in forecasting/models.py; "prophet_like" added to the ModelType literal and a no-flag model_factory branch.
• Jobs: _execute_train and _execute_backtest gain a prophet_like branch.
• Tests: new test_prophet_like_forecaster.py (contract tests + model-specific invariants — additive invariant, NaN tolerance via the imputer, imputer leakage-safety, determinism); prophet_like coverage extended into the forecasting/jobs/backtesting unit tests and the scenarios model_exogenous integration test.
• Docs: runnable examples/models/prophet_like_additive.py, model_interface.md + feature_frame_contract.md entries, README model-type list.

Pure scikit-learn — ships always-enabled like the regression model: no new dependency, no optional extra, no feature flag. It trains, scenario-re-forecasts, and backtests through the existing requires_features-based dispatch with zero changes to the train/predict/scenarios/backtesting service layers. No pyproject.toml / uv.lock / config.py / routes.py / persistence.py / registry change; no Alembic migration.

Validation

• ruff check — all checks passed
• ruff format --check — 294 files already formatted
• mypy app/ (--strict) — no issues, 272 source files
• pyright app/ (--strict) — 0 errors
• pytest -m "not integration" — 1359 passed, 0 failed
• pytest -m integration (forecasting/scenarios/jobs) — 41 passed, including the new prophet_like model_exogenous test

Closes #248

[sourcery-ai]

Sorry @w7-mgfcode, you have reached your weekly rate limit of 500000 diff characters.

Please try again later or upgrade to continue using Sourcery

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 46af3e1b-a03c-4864-a4a2-c465d18dde9c

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feat/forecasting-prophet-like-model

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}