Hierarchical Forecast Reconciliation Workbench

How to use Hierarchical Forecast Reconciliation Workbench

Build a leaf summing matrix, project entered forecasts with exact linear equations and verify every parent sum plus weighted normal-equation residuals.

1. Use the example as the schema; replace its labels and entered model assumptions.
2. Build a leaf summing matrix, project entered forecasts with exact linear equations and verify every parent sum plus weighted normal-equation residuals.
3. Calculate, inspect the complete witnesses and limitations, then copy or download the JSON report.

Good to know

• 1–20 distinct flat nodes, one rooted tree, depth at most 10 and at most 10 leaves. Signed base forecasts and positive variance scales are decimal strings with magnitude at most 1 billion.
• All forecasts must describe the same quantity and horizon. Variance fields are validated in every mode but used only by diagonal-WLS. No forecasting model, covariance estimation, nonnegative optimization, uncertainty interval or accuracy guarantee.
• Local calculation with no upload, input persistence or outside lookup. Model JSON is limited to 30,000 characters and 10,000 values. Decimal quantities are strings with up to six places; count fields use plain JSON integers. Exact fractions are capped at 8,192 bits. The complete JSON output is limited to 1,000,000 characters; approximate numbers preserve their stored values. Unsupported states fail explicitly.