Time Series Modeling, Holdout Sandbox & Accuracy

Contact Center Demand Forecasting

Pluggable time series forecasting suite: Year-over-Year (YoY) Seasonal Trend, Ensemble Blended Modeling, Seasonal Decomposition, Holt's Double Exponential Smoothing, SES, and Multi-Variable Regression. Includes walk-forward out-of-sample backtesting, Forecast Holdout Sandbox (target month testing with before-only training & lookback control), accuracy tracking (WAPE, MAPE, Signed Bias %), and multi-year data support.

🌐 Blended View
Viewing combined aggregate across all queues (1 queue)
Data Source & Series Input
History rows (0 periods):
Period / Date Volume
Pluggable Model Architecture
periods
%
Assumed Handle Time (AHT) 180s
seconds
Used for Erlang calculation and Capacity Planning transfer
Historical Volume
--
-- periods
Forecast Total
--
-- future periods
Forecast Avg / Period
--
0% vs baseline
In-Sample MAPE
--%
Mean Absolute % Error
Goodness of Fit (R²)
--%
Variance explained
Peak Future Period
--
Highest demand
Time Series Forecast Projections Daily
Historical Actuals Forecast
Aggregation:
Range:
CI Band:
→
Forecast Breakdown Table (0 periods)
Future Period Base Model Trend Factor Seasonality Index Trend Profile Projected Volume Est. Erlangs (A)