Forecast Time Series Explorer Insight

Use the Forecast Time Series Explorer to drill down into individual forecast series. Planners can filter by accuracy range, key values, or winning method, and visually compare History, Forecast, and Actuals data in an interactive time series chart.

The DMP Forecast Quality KPI Dashboard shows a high-level overview of forecast health, but planners need a way to investigate individual series when they identify problems. To understand why a specific series has good or poor forecast quality, planners must manually extract data. This companion dashboard bridges the gap between summary KPIs and series-level investigation.

These are the components of the Forecast Explorer Dashboard:

• Five accuracy range cards (Excellent, Good, Fair, Poor, and Bad) show the series count for each sMAPE band. You can select a card to filter all grids by that accuracy range.
• A time series line chart shows History (AccID 95), Forecast (AccID 20), Actuals (AccID 32), and a Difference line that represents the difference between Forecast and Actuals. You can view the difference in absolute (ABS) mode or relative percentage (REL) mode.
• Optional fill-under-lines rendering for improved visual clarity when lines overlap.
• Use search filters for Key 1, Key 2 (partial match), and winner Method ID (exact match) to narrow results.
• The Best ten and Worst ten grids show the top and bottom series in the selected accuracy range. You can select a BaseID link to load the corresponding series in the chart.
• You can configure parameters, such as the dataset, period range (pcfrom/pcto), difference method (ABS/REL), fill mode, global filter, key column visibility, and method column visibility.
• The dashboard uses data from the Dataset table and related forecast data for the configured dataset and period range.
• You can open this view from the Forecast Quality KPI Dashboard.