Apply data spreading to a forecast

For predictions that are written to consolidated cells, you can use either the Proportional (default) or Relative proportional data spreading methods.

Note: Data spreading in a forecast requires TM1 server 11.8.7 (PA 2.0.9.8) or later. If Planning Analytics Workspace is connected to a TM1 server prior to 11.8.7, the data spreading option for a forecast is not available.

See Proportional data spreading method for more information on Proportional data spreading.

See Relative proportional data spreading method for more information on Relative proportional data spreading.

The following example shows how you can define and use data spreading. In this case, we are performing a forecast on three time series on the context: UK, Germany, and the consolidated member Europe. If all items on the context were not consolidated, then data spreading would be applied only to Europe. However, in the following scenario, given there are consolidated items on the context ('All Products' and 'Whole year'), then all three time series use the selected spreading option.

Click the Edit button for the Spread forecast values option, then select the type of data spreading you want to use on the Spread Data Options page.

Proportional - Proportional data spreading distributes a specified value among cells proportional to existing cell values. When you select the Proportional method, no further configuration is required. You can click Apply to set the data spreading method.

Relative proportional - Relative proportional spreading distributes values to the leaves of a consolidation cell proportional to the leaves of a reference cell or cells. When you select the Relative proportional method, you must define the Reference type for the spreading operation. Select one of the following reference types.

Seasonality (Auto): For each time series, after the prediction is made but before the forecasted values are written, the forecasting process attempts to identify a seasonality period. If one is identified (for example, for a retailer with a month granularity time series, may be the number 12, representing a yearly, repeated cycle of ups and downs in the data), then the reference cell used for the relative proportional spread is 12 cells back along the time series. If no seasonality is detected, then the last historical cell is used for reference.
Last historical time period: For every consolidated cell written to (that is, forecasted values along a time series), the reference cell for the relative proportional spread is the latest non-ignored value before the Start Date along the time series. On a cell-by-cell basis, the behavior is identical to that of doing a relative proportional spread on that cell manually with the aforementioned reference cell.
Specific time period: For every consolidated cell written to (that is, forecasted values along a time series), the reference cell for the relative proportional spread is the one you select with the Reference time period option for each given time series. On a cell by cell basis, the behavior is identical to that of doing a relative proportional spread on that cell manually with the aforementioned reference cell
Range of time periods: This works similar to the Seasonality (auto) reference type in that the reference cell cycles on a range along the time series. Namely, for every consolidated cell written to (that is, forecasted values along a time series), the reference cell for the relative proportional spread is the one identified within the range defined by the Start of time period and End of time period that you specify. When the specified End of time period is reached, the next reference cell will be the ‘Start of time period’ and the cycle repeats until all forecasted cells are written to.
Click Apply after you define a reference type.