Machine Learning for Data Mining

Bagging

Let's go back into the Neural Net model and, this time, what we're going to do is bagging instead of boosting:

1. Go over to the Objectives tab and select Enhance model stability – it's sampling with replacement. Do not do bagging when you have small datasets or outliers. The main idea behind bagging is that new training datasets are generated that are of the same size as the original training dataset and this is done by using sampling with replacement. We're actually bootstrapping in this kind of situation.
2. Click on Run for the model.
3. Let's take a look at our generated model. This is the model that we ended up building:

Notice that ...