Summary
Current Period Prediction is an optional feature within Forecast Manager that allows you to try and respond more quickly to changes evidenced by demand in your current forecasting period.
This feature is most useful when doing monthly forecasting, since it allows you to get updates to your forecast more than once per month.
The feature works by examining the level of demand within the current month, and comparing to how far through the month you are, and what level of demand was expected, based on forecast, for where you are in the month. It then extrapolates the current level of demand forward and projected total demand for the current period.
Then, this projected total for your current month is added as a new point in the demand time series fed to the forecasting algorithm, and that allows the forecast to adjust.
This process is repeated day-by-day as you move through the month, refining your forecast each day until the month is complete.
As you get closer and closer to the end of the month, your statistical forecast converges on what your statistical forecast will be on the first of the next month, thus allowing you to react more quickly to changes in demand.
Process Setup
Enable Current Period Prediction by checking the "Use Current Period Prediction in Statistical Forecast" checkbox in the "Current Period Settings" tab in Forecast Manager.
Like other forecast settings, this checkbox inherits from higher levels of your hierarchy, so if you have selected a node higher up in your "tree view" of Forecast Manager, such as a product group or product family, this setting will inherit downwards, allowing you to make this change for large groups of items at once.
Only a few settings are necessary to configure the Current Period Projection feature.
- Min % Through Period: Once enabled, you then have the option to decide how far through the period to begin making these projections. Typically this should be no fewer than 10 days through the month, to avoid large swings in the predicted monthly total due to unusual demand being randomly high or low in the first few days of the month.
-
Max Adjustment Range: As well, you can set a max adjustment range to limit how much the prediction is allowed to vary from your baseline forecast, to prevent any numbers from getting too large. A 200% variation is a reasonable range here. (e.g. enter 2.0 = 200%), but we have seen as high as 500% (5.0) be reasonable in some situations.
- If you put no value here, there is no limit on the allowed adjustment from the current period prediction.
Process Output
You can see how these predictions are being made in the Current Period tab of Forecast Manager.
- Period-To-Date Sales - the actual sales that you have had in this period so far
- Expected PTD Sales - This is SIQ’s expectation of what your period-to-date sales should be, based on how far through the period you are. For example, if we forecasted that you would sell 100 units for the month, and you are 50% through the month, your expected sales would be 50 units. The reality is slightly more involved, since SIQ accounts for the demand pattern (e.g. if more is bought at the end of the month vs beginning etc.).
- Projected Period Sales - Based on your current period-to-date sales, and your position in the month, we project what we now believe your current period total will be. Using the example above of 100 units per month forecasted, and 50% through the month, if you’ve only sold 35, then our new projected total would be 70 instead of 100, since you’re under-performing the forecast. So, this number can be higher or lower than what your forecasted units is.
-
Forecasted Units - The forecasted units for this item as of the beginning of the month. This is based on the forecast snapshots, which you can view in the Snapshots tab.
- NOTE: If there is not a snapshot for the beginning of the month (or week) the most recent snapshot is used instead as the basis for comparison.
Troubleshooting
- Current Period Prediction is only enabled for items with a recurring usage pattern. Items that are sporadic or slow do not have predictable enough demand to make predicting total demand for the current period a useful feature.
- If you have the Demand Patterns feature enabled, your "% of Demand Expected" will be sensitive to demand patterns throughout the month. For example, if you are 50% through the month, but 90% of your demand is in the last 5 days of the month, StockIQ takes this into account in the "% of Demand Expected" number that is used to project the period total.