Ocena wpływu wybranych metod imputacji na wyniki klasyfikacji obiektów w modelach drzew klasyfikacyjnych
Assessment of the influence of selected imputation methods on the results of object classification using classification trees

Summary/Abstract: In contrast with most multivariate statistical analysis methods, classification tree is an example of the learning algorithm coping with missing values in special, original way. In the paper the influence of some selected missing data techniques on the results of object classification using CART and CRUISE classification trees was assessed. All the procedures were compared by artificially simulating different proportions and mechanisms of missing data using complete data sets mainly from the UCI repository of machine learning databases.

• Details
• Contents