Run Predictive Forecasts on Table Cells

When working with planning models in SAP Analytics Cloud, you can run predictive time series forecasts on your data within a table.

Prerequisites

You will need planning rights and a planning license to run a predictive time series forecast. You'll also need access to a table with planning model data and a date dimension in the columns.

Context

A predictive time-series forecast runs an algorithm on historical data to predict future values for a specific measure in a planning model. The forecasted values can be later used in the planning process. The prediction is only applied to the specific selected measure; however, booked data will be overwritten if you choose to run a forecast over a cell containing historical data. Other data in the version will not be impacted.

While running a predictive time series forecast, you can determine the time range for the prediction and the past time periods to use as historical data.

Note

You currently cannot run predictive forecasts on the following:

calculated measures
formulas
blended datasets
local currencies
on models with multiple time dimensions
in tables with locked cells
on measures with aggregation types Label or None
on time granularities less than monthly

Procedure

Select a single table cell that can be edited. From the toolbar, select Tools Predictive Forecast.

The Create Predictive Forecast dialog is displayed.
Under Predictive Forecast Settings, specify the period you want to forecast.

Select an option under Granularity to specify the time granularity for the predictive forecast. Use the Predict From field to specify the starting point for the predictive forecast calculation. Under Predict To, specify a time period to end the calculation.

Restriction
By default, the lowest Granularity is Month. For fiscal years, Period is the lowest Granularity. Daily or weekly granularity are not supported.
Under Advanced Options, specify options on how you want to run the predictive forecast.
1. Under Algorithm, select one of the following options to create the forecast:
  - Automatic Forecast
  - Linear Regression
  - Triple Exponential Smoothing
2. If you do not want to have any negative values included in your results, under Output Values select Positive Only.
3. Under Use Data From, specify on what historical data you want to run the algorithm.
4. Select which Version to use for the forecast input data.
  By default the Actual version is used.
When you are ready to run the predictive forecast, select Preview.

The time series forecast is displayed in the Preview Predictive Forecast dialog. The preview contains the past actual values together with predicted values displayed as a graph and a table row. Upper and lower limits for the confidence interval are also provided for the graph. The table rows contain the predicted values that will be added to the table. A dashed line (originating from the first point of the forecast) represents the predictive model applied to past periods.

Note
You can preview the predicted values for booked cells after running a time series forecast.

For cells containing values, the cell's existing weighting is respected. For unbooked cells, the weighting of the reference period is used.
To populate the table with the forecasted values, select OK.

A message indicates that the forecasted values were added to the table.