EX

Simulation & Prediction

Back to journey: Fraud & Revenue Assurance

Forecasting

Historical actuals with run-rate, no-action and recommended-action projections against target.

Latest actual

ACTUAL

USD 1.9M

Run-rate (6 mo)

FORECAST

USD 1.9M

No action (6 mo)

FORECAST

USD 1.13M

Recommended (6 mo)

SCENARIO

USD 1.13M

Target

TARGET

Not stored

Fraud + revenue leakage — Fraud & Revenue Assurance

Solid = stored actuals (Historical KPI trend data). Dashed = deterministic projections in the shaded period. Band = ±1σ of historical residuals around the no-action line, widening with horizon. Direction of good: lower.

Forecast table

PeriodTypeActualRun-rateNo actionRecommendedSpread (±1σ)
Apr 26ACTUALUSD 2.6M
May 26ACTUALUSD 2.5M
Jun 26ACTUALUSD 2.4M
Jul 26ACTUALUSD 2.3M
Aug 26ACTUALUSD 2.2M
Sep 26ACTUALUSD 1.9M
Oct 26FORECAST—USD 1.9MUSD 1.77MUSD 1.77MUSD 1.7M – USD 1.84M
Nov 26FORECAST—USD 1.9MUSD 1.64MUSD 1.64MUSD 1.54M – USD 1.74M
Dec 26FORECAST—USD 1.9MUSD 1.51MUSD 1.51MUSD 1.39M – USD 1.63M
Jan 27FORECAST—USD 1.9MUSD 1.39MUSD 1.39MUSD 1.25M – USD 1.53M
Feb 27FORECAST—USD 1.9MUSD 1.26MUSD 1.26MUSD 1.11M – USD 1.41M
Mar 27FORECAST—USD 1.9MUSD 1.13MUSD 1.13MUSD 0.96M – USD 1.3M

Source data, assumptions, methodology & limitations

Source data

  • Telecom KPI data (current period, target, prior)
  • Historical KPI trend data (monthly history Apr–Sep 2026)

Assumptions

  • Run-rate baseline: holds the latest stored actual flat.
  • No action: continues the 6-month stored trend (-0.13 per month).
  • Recommended action: reverses the trend at the same monthly rate (no stored target).

Methodology

  • Least-squares straight line through 6 stored months (slope -0.129 per month).
  • Residual standard deviation σ = 0.069; spread = σ × √(months ahead).
  • Identical arithmetic feeds Decision Journey Stage 5 — Predict.

Limitations

  • Forecasts are straight-line projections of six stored months; they ignore seasonality, shocks and interactions between measures.
  • Many KPIs have no stored target; where absent, the recommended path reverses the trend at the same monthly rate.
  • No downside scenario is stored for any journey; downside coverage is shown as a gap rather than invented.
Forecast horizon: 6 months (Oct 2026 – Mar 2027)Model status: deterministic arithmetic — no AI or machine-learning model.
Evidence typeItemSource
model output

SIM-box detection output

1,240 SIM-box incidents flagged; 91% detection rate.

Revenue assurance data
methodology

Fraud & Revenue Assurance methodology

Run CDR anomaly detection; Reconcile usage, rating and billing; Value prevented losses.

Finance Planning Dataset