Demand Forecast Strategy Alignment Index (DFSAI)

GMI-AVI-032AviationIntegration-Required

Definition

Degree to which the parts demand forecasting model's predictions align with actual demand patterns under the current maintenance strategy: specifically measuring forecast accuracy during transitions from calendar/hard-time maintenance to predictive and on-condition maintenance, where the demand timing and quantity patterns shift in ways the trained model doesn't anticipate.

The Measurement Gap

The measurement gap DFSAI addresses: parts fill rate and inventory carrying cost are the standard metrics for supply chain performance, but neither reveals whether demand forecast errors are caused by maintenance strategy changes. An airline transitioning 30% of its fleet to on-condition maintenance over 24 months will see its demand patterns change in that population while the AI model continues to forecast based on legacy hard-time demand patterns. Fill rate may degrade, expedite costs may rise, and stock-outs may increase, but the connection to the maintenance strategy transition is invisible in standard supply chain reports.

The structural cause: demand forecasting models are trained on historical consumption data that reflects a specific maintenance strategy mix. As the strategy mix changes, demand timing shifts. Hard-time components are removed at fixed intervals regardless of condition; on-condition components are removed when condition triggers removal. The two populations have different demand timing patterns. If the model's assumed strategy mix is 70% hard-time and the actual mix is 40% hard-time / 30% on-condition / 30% predictive, the demand timing assumptions are wrong for the majority of demand.

The decision DFSAI supports: should the demand model be retrained with recent data that reflects the current strategy mix? Should manual adjustment factors be applied to forecasts for part families undergoing strategy transitions? Are the current stocking levels calibrated for how maintenance is actually executed today, or for how it was executed two years ago when the model was last trained? DFSAI makes this misalignment visible by measuring forecast accuracy segmented by removal reason type: revealing which part families the model has lost alignment with.

Formula

Plain language: For each major part number, compare the AI's demand forecast (quantity and timing) to actual demand over the measurement period. Segment by removal reason type. Measure forecast error by strategy type to reveal where the model has lost alignment.

Primary notation:
DFSAI = 1 − (1/P) × Σ |Forecast_demand_p − Actual_demand_p| / Actual_demand_p

Also compute the strategy alignment gap:
Strategy_alignment_gap = |Forecast_strategy_mix − Actual_strategy_mix|

Variables:

  • Forecast_demand_p = AI-predicted demand for part p in the measurement period (units)
  • Actual_demand_p = actual demand for part p from requisition records (units)
  • P = number of major part numbers tracked
  • Forecast_strategy_mix = the model's assumed distribution of removal reasons (% hard-time, % on-condition, % predictive), often implicit in the model's training data
  • Actual_strategy_mix = the actual distribution of removal reasons in the period, from MRO system removal reason codes

Units: DFSAI is 0 to 1 (1 = perfect forecast). Strategy_alignment_gap is a separate diagnostic showing where the model's assumptions diverge from current reality. Report both together: DFSAI quantifies the accuracy gap; Strategy_alignment_gap explains it.

Edge case handling: parts with very low demand (fewer than 5 units per year) produce unreliable individual MAPE estimates. Aggregate into ATA chapter families for DFSAI computation. New-to-fleet parts with no demand history cannot be included: track separately. Campaign-driven demand spikes are inherently hard to forecast and should be excluded from DFSAI or tracked as a separate category.

Proposed formula, pending validation pilots.

At a Glance

Category
Parts & Supply Chain Prediction
← MEL Interaction Risk Score (MIRS)Pool TAT Distribution Coverage Score (PTDCS) →

Assess Your Organization

The AI-GPS Assessment evaluates your AI governance maturity and recommends KPIs like this one for your implementation roadmap.

Take the Free AI-GPS Assessment