Exploring the Forecasting Model (Data Mining Tutorial)
Now that you have built the forecasting model, you can explore the results by using the Microsoft Time Series Viewer in the Mining Model Viewer tab of Data Mining Designer. The Microsoft Time Series Viewer contains two tabs: Decision Tree and Charts. For more information about this viewer, see Viewing a Mining Model with the Microsoft Time Series Viewer.
The Microsoft Time Series algorithm builds a model for each distinct series that exists in the dataset. For example, each region in the dataset contains data about sales over a period of time, so the algorithm creates a separate time series for each region. Each time series includes information about both sales quantity and sales amount.
In this task, you will explore time series describing sales in the regions Europe, North America, and the Pacific.
The Decision Tree tab of the Microsoft Time Series Viewer in Data Mining Designer lets you view the decision tree that was created when the model was processed.
Each node in a decision tree displays three pieces of information:
The concentration of cases for the state of the predictable attribute that is specified in the Background control. (Both the Node Legend window and the InfoTip that appears when you pause the pointer over an object in the tree give the exact number of cases.)
The regression formula for the node.
A diamond chart that represents the range of the attribute. The diamond is located at the mean for the node, and the width of the diamond represents the variance of the attribute at that node. A thinner diamond indicates that the node can create a higher quality prediction.
With the Charts tab of the Microsoft Time Series Viewer, you can investigate the time series that the algorithm creates.
The legend on the right side of the viewer lists the series that are selected in the drop-down list box, and includes a check box for each time series. By selecting and clearing the check boxes in the legend, you can control which time series are displayed in the viewer.
The chart displays both historical and future data. Future data is shaded to differentiate it from historical data. Use the Prediction Steps list to control how many future steps of data are displayed. Use the Show Deviations check box to add error bars to the predictions.
As you can see in the viewer, total sales for all regions are generally increasing, with a peak every 12 months, in December. The predictions continue this trend.