forecast-models
Guidance for the demand forecasting models that predict parcel volume per depot for the Mossback capacity planner. Covers model training, model features, feature engineering, training data, holdout wi
SKILL.mdskills/agent-skills-creator/evals/fixtures/crowded-skill-listing/base/.agents/skills/forecast-models/SKILL.md
GitHubForecast Models
Rules
- Training code lives in
forecast/train.py; features inforecast/features/. - The holdout window is always the most recent 28 days. Do not tune on it.
- A new model version is promoted with
python -m forecast.promote <version>only when its error on the holdout beats the live version for at least 80% of depots. - A manual override in
planner_overridesalways wins over a forecast for its date range.