Forecasting 1000's of items for greater accuracy with Forecast PRO
Exception reporting is ideal where the data volumes are huge and improved accuracy. Look into the really serious issues first sorting the data by priority. 1000's or 10's of thousands of items to manage.
This expert statisical forecast from Forecast PRO shows 6 items requiring attention as they have changed more 25% since the last time Forecast.
To see and discuss FORECAST PRO - Contact MXI.
Forecast PRO selected a forecasting method to generate a sales forecast. The Business Intelligance within Forecast PRO simply highlights the SKU/items where some attention is required. The Supply Chain people or Sales & Operations planning guys can focus on the issues at hand here.
FORECAST PRO - Brochure Page & Case Studies
The Forecast can be different to the actual recorded sales data or simply last statistically generated sales forecast. Either exception method is fine. Another benefit here is that some thought can be put into your planning approach such as using a different statisical planning method (trendline, box jenkins, exponential smoothing) in Forecast PRO.
Monitoring within-sample error statistics such as the MAPE and MAD. This type of exception reporting is often misused by individuals who assume that large within-sample errors indicate poor forecasting models-usually they don't. In fact, large within-sample errors reflect scale and volatility of the data rather than the accuracy of the forecasting model. Highly volatile data sets always produce large within-sample errors because they are volatile-not because the forecasting model is doing a poor job. Similarly, high-volume series generate larger MADs (unit errors) than low-volume series because they are higher volume-not because the forecasting model is inferior. Thus, monitoring within-sample statistics can be useful to understand the scale and volatility in the data, but since it is not monitoring the actual forecasts, it is not very useful in terms of finding potentially poor forecasts where action may be needed.
Selecting Thresholds?
The above example selected 6 items where the forecast differed by over 25% its previous forecast. Obviously a lower threshold of 10% would generate more exceptions and 50% far fewer for manual intepretation. The planners then intervene accordingly changing the item/reporting group forecast to manually adjust.
There is a cost to reduced thresholds-managing additional false positives requires time and resources and does not improve the final forecast. There is a benefit to reducing thresholds-they somtimes generate additional true positives as taking action improves forecast accuracy & saves money. Thus, the thresholds need to be set to values that balance the cost of reviewing the false positives with the cost of missing true positives.
Expermenitation is the best way to develop this iterative planning process which will improve item level accuracy.
It is not a universal solution either. High-value items warrant lower thresholds (and thus more weeding through false positives) than lowervalue items owing to the higher cost to missing the true positives. Best forecasting practice is to categorise items based on their importance and vary the thresholds for different categories accordingly.