maiven gateway · online··
manifest · 1d ago
Sign in Back to catalog
Definition
Monthly forecasted amount vs posted GL actual by scenario and account type. error = forecast - actual; `accuracy_pct` is a formula measure (1 - sum(abs error) / abs(sum(actual))) so it stays window-correct.
What it means
Watch out for
- highComparing scenarios as if one is the truth
scenario (base / upside / downside) are alternative plans. Pick one scenario before judging accuracy; blending them is meaningless.
- mediumRecomputing accuracy_pct client-side
accuracy_pct is a formula over summed absolute error and actual. Dividing pre-aggregated outputs across a window biases it.
Questions this answers
- “How accurate was the base forecast by account type this year?”
- “Forecast error by scenario last quarter.”
Related metrics
- budget_vs_actual
The budget counterpart to this forecast comparison.
- gl_balances_by_period
The posted actual side of the comparison.
7 of 7
Column
Type
Description
Tests · PII
period_month_start
Date
Month bucket — toStartOfMonth(period start). Grain + sort key.
1 test
scenario
String
base / upside / downside. Dimension + sort key.
1 test
account_type
String
asset / liability / equity / revenue / expense; `unknown` when unset. Dimension.
1 test
forecast_amount
Decimal(18, 2)
Summed forecasted amount. Decimal(18,2).
1 test
actual_amount
Decimal(18, 2)
Summed posted GL actual. Decimal(18,2).
1 test
error_amount
Decimal(18, 2)
forecast - actual. Decimal(18,2).
1 test
abs_error_amount
Decimal(18, 2)
abs(forecast - actual). Decimal(18,2).
1 test