Selecting features/attributes using the low-level method, the filtering method, and the meta-classifier method
Feature selection is an important machine-learning process that identifies the most important attributes in a dataset from a set of attributes, so that if a classifier is generated based on the selected attributes, the classifier produces better results than the one with all the attributes.
In Weka, there are three ways of selecting attributes. This recipe will use all of the three ways of attribute selection techniques available in Weka: the low-level attribute selection method, attribute selection using a filter, and attribute selection using a meta-classifier.
Getting ready
The recipe will select important attributes of the iris dataset that can be found in the data directory of Weka's installed directory.
To perform attribute selection, two elements are required: a search method and an evaluation method. In our recipe, we will use Best First Search as our search method and a subset...
