Decision tree theory
Decision or classification trees are extremely common predictive models that successively partition data based on the relationship between the target and predictor variables. Decision tree models indicate which predictors are most strongly related to the target. These methods are capable of combing through a large set of predictors by successively splitting a dataset into subgroups on the basis of the relationships between predictors and the target field. Decision tree models classify cases into groups and predict values of a target field based on the values of the predictor fields.
Decision trees have some very attractive features:
- They create segments that are mutually exclusive and exhaustive, meaning that they allow you to identify homogeneous groups
- They make it easy to construct rules for making predictions about individual cases
- They can handle a large number of predictors, so that the most important variables are at the top of the tree and the least important variables...
