Split Model Viewer

The Model tab lists all the models contained in the nugget, and provides statistics in various forms about the split models. It has two general forms, depending on the modeling node.

Sort by. Use this list to choose the order in which the models are listed. You can sort the list based on the values of any of the display columns, in ascending or descending order. Alternatively, click on a column heading to sort the list by that column. Default is descending order of overall accuracy.

Show/hide columns menu. Click this button to display a menu from where you can choose individual columns to show or hide.

View. If you are using partitioning, you can choose to view the results for either the training data or the testing data.

For each split, the details shown are as follows:

Graph. A thumbnail indicating the data distribution for this model. When the nugget is on the canvas, double-click the thumbnail to open the full-size graph.

Model. An icon of the model type. Double-click the icon to open the model nugget for this particular split.

Split fields. The fields designated in the modeling node as split fields, with their various possible values.

No. Records in Split. The number of records involved in this particular split.

No. Fields Used. Ranks split models based on the number of input fields used.

Overall Accuracy (%). The percentage of records that is correctly predicted by the split model relative to the total number of records in that split.

Split. The column heading shows the field(s) used to create splits, and the cells are the split values. Double-click any split to open a Model Viewer for the model built for that split.

Accuracy. Overall accuracy formatted as a percentage.

Model Size. Model size depends on the modeling method: for trees, it is the number of nodes in the tree; for linear models, it is the number of coefficients; for neural networks, it is the number of synapses.

Records. The weighted number of input records in the training sample.