Forecasting

You can now use forecasting in IBM® Planning Analytics Workspace to discover and model trend, seasonality, and time dependence in data.

You can forecast in Planning Analytics Workspace by using automated tools that model time-dependent data. Automated model selection and tuning makes forecasting easy to use, even if you are not familiar with time series modeling.

Forecasts and their confidence bounds are displayed in visualizations as a continuation of historic data. You can also view the statistical details for generated models if you want to see the technical background.

The following example shows forecasting values and confidence bounds in a line visualization.

Specifying time series in forecasts often requires data manipulation. Planning Analytics Workspace supports a wide range of time series without the need for manipulation, ranging from standard date and time types, to nested periodic and cyclical time fields. When data is recognized as a time series, data preparation is automated. Appropriate trend and seasonal periods are detected, and models are selected from a set of nine different model types.

You can forecast in line visualizations. Forecasting allows analysis of hundreds of time series per visualization. Forecasts and confidence bounds are computed for each time series, and displayed in the visualization as extensions of the current data. You can inspect each time series separately, and tailor the forecast and results to your own data and requirements.

If you are familiar with forecasting models, you can view the selected model type, estimated model parameters, standard accuracy measures, and processing summary information.

For complete details on using forecasting in Planning Analytics Workspace, including a complete tutorial, see Forecasting.