This article lists the documentations available on Ensemble modeling with Bagging/Voting, Stacking and Boosting methods.

Available Resources:

• Explanation of Data mining concepts (including Bagging, Stacking, Boosting)
• Example on Predictive data mining and deployment for a continuous output variable
• Example on Predictive Data Mining for Categorical Output Variable (Classification)
• Concepts on Stacking
• Concepts on Boosting
• Concepts on Bagging

Notes:

• Statisitca don’t have specific workspace nodes available for general Stacking or Boosting functionality.  
• Model bagging/voting functionality is available via "Rapid Deployment node" by default.
• Stacking could be achieved with some manual work. There isn’t a workspace node in Statistica that will do this automatically, but a user could easily take predictions from one model and use them as inputs into another model.  For example, it is straightforward to output the terminal node ids from a decision tree analysis and then use them as inputs into a neural network.
• Boosting would be more difficult and may be easiest accomplished via scripting. Boosting would be more difficult to accomplish since it requires you to define a weight and/or probability of selection based on accuracy of the current model’s prediction. Admittedly, this would be difficult but possible for a determined analyst. Boosted trees is the best example of boosting that is available in the software already but is specific to decision trees. Note also that Random Forests is an ensemble method available in the software and the SANN module can easily create ensemble networks.