Measures in the CrossValidation Report
Topic Status: Some information in this topic is preview and subject to change in future releases. Preview information describes new features or changes to existing features in Microsoft SQL Server 2016 Community Technology Preview 2 (CTP2).
During crossvalidation, Analysis Services divides the data in a mining structure into multiple crosssections and then iteratively tests the structure and any associated mining models. Based on this analysis, it outputs a set of standard accuracy measures for the structure and each model.
The report contains some basic information about the number of folds in the data and the amount of data in each fold, and a set of general metrics that describe data distribution. By comparing the general metrics for each crosssection, you can assess the reliability of the structure or model.
Analysis Services also displays a set of detailed measures for mining models. These measures depend on the model type and on the type of attribute that is being analyzed: for example, whether it is discrete or continuous.
This section provides a list of the measures that are contained in the CrossValidation report, and what they mean. For details on how each measure is calculated, see CrossValidation Formulas.
The following table lists the measures that appear in the crossvalidation report. The measures are grouped by test type, which is provided in the lefthand column of the following table. The righthand column lists the name of the measure as it appears in the report, and provides a brief explanation of what it means.
Test Type 
Measures and Descriptions 


Clustering 
Measures that apply to clustering models 




Classification 
Measures that apply to classification models 







Likelihood 
Likelihood measures apply to multiple model types. 




Estimation 
Measures that apply only to estimation models, which predict a continuous numeric attribute. 




Aggregates 
Aggregate measures provide an indication of the variance in the results for each partition. 


