Intelligent Forecasting in Sheets

The Intelligent Forecasting feature provides predictive analytics capabilities within JustPerform sheets. This feature offers improved algorithm selection and configuration options.

The users can leverage historical trend data from your sheets to predict future values for metrics across different versions and time periods.

For example: If you have one year of actual data for 2024, you can use this to predict values for 2025 or for different versions like forecast. It provides clear guidance on optimal data requirements: ideally, you need at least five years of historical data to predict one year ahead, while one year of data can reliably forecast up to three months. Understanding these ratios helps set realistic expectations for prediction accuracy.

The enhanced interface streamlines the forecasting workflow while providing comprehensive configuration options. Users specify the source version containing historical data (such as "Actual"), select the metric to analyze (such as "Amount"), define the historical time period (like January 2024 to December 2024), choose the target version for predictions (such as "Forecast"), and specify the forecast horizon in months. The system then processes this configuration using the selected algorithm and generates predictions that integrate seamlessly with your planning workflows.

Using Intelligent Forecasting

To access and configure intelligent forecasting

1. Navigate to the desired sheet containing your historical data.

2. Select Sheet Actions from the top right menu.

3. Choose Intelligent Forecasting under actions.

   

4. The Intelligent Forecasting tab is displayed as shown below:

   

5. Configure the following in the Intelligent Forecasting tab:

   • Configure the algorithm selection.

     1. Select Automatic to let the system choose the optimal algorithm.

   • Or select a specific AutoML algorithm -AutoArima, AutoTheta or AutoETS depending on how your data is defined.Define the Version, and Metric values containing historical data in the Input Data parameters:

   • Define the Time From and Time To values containing historical data in the Input Time parameters:

   • Define the Version, and Metric values containing historical data in the Post To parameters:

   • Select the Enable forecasting bounds checkbox.

   • Enter the Forecast Horizon details.

6. Click Run to execute the forecasting process.

   The forecast analyses the data and predicts the result.

   

7. Review the generated predictions:

   • Verify the forecast values appear in the specified version.

   • Check that the time periods align with your forecast horizon.