Classification and Regression Trees(CART) Theory and Applications

This master thesis is devoted to Classification and Regression Trees (CART). CART is classification method which uses historical data to construct decision trees. Depending on available information about the dataset, classification tree or regression tree can be constructed. Constructed tree can be then used for classification of new observations. The first part of the thesis describes fundamental principles of tree construction, different splitting algorithms and pruning procedures. Second part of the paper answers the questions why should we use or should not use the CART method. Advantages and weaknesses of the method are discussed and tested in detail. In the last part, CART is applied to real data, using the statistical software XploRe. Here different statistical macros (quantlets), graphical and plotting tools are presented.