Predictive Analytics and Data Mining

Comparison of Data Mining Algorithms

Classification: Predicting a Categorical Target Variable

Algorithm	Description	Model	Input	Output	Pros	Cons	Use Cases
Decision Trees	Partitions the data into smaller subsets where each subset contains (mostly) responses of one class (either “yes” or “no”)	A set of rules to partition a data set based on the values of the different predictors.	No restrictions on variable type for predictors.	The label cannot be numeric. It must be categorical.	Intuitive to explain to nontechnical business users. Normalizing predictors is not necessary.	Tends to overfit the data. Small changes in input data can yield substantially different trees. Selecting the right parameters can be challenging.	Marketing segmentation, fraud detection.
Rule ...