(An) application of discretization methods for graphical modelling

Abstract

One of the major problems in this context is combining nominal, discrete, and continuous variables in the same model. There are many studies to find an efficient algorithm for partitioning the range of a continuous variable to a discrete number of intervals. All these prompt researchers and practitioners to discretize continuous features before or during a machine learning or data mining task. Most methods used for discretizing a continuous variable use its relationship to another variable to determine the partitions. This is often found in classification procedures, such as decision trees and in naive Bayesian classifiers. There are numerous discretization methods available in the literature. Data can also be reduced and simplified through discretization. For both users and experts, discrete features are easier to understand, use, and explain. Therefore discrete values have important roles in data mining and knowledge discovery. Many studies show induction tasks can benefit from discretization: rules with discrete values are normally shorter and more understandable and discretization can lead to improved predictive accuracy. Widely used systems such as CART (Breiman et al., 1984) deploy various ways to avoid using continuous values directly. Discrete features are closer to a knowledge-level representation (Simon, 1981) than continuous ones. There are many other advantages of using discrete values over continuous ones. Our thesis aims at a systematic study of discretization methods with their history of development, effect on classification, and trade-off between speed and accuracy. Contributions of this thesis are an abstract description summarizing existing discretization methods, a hierarchical framework to categorize the existing methods and pave the way for further development, concise discussions of representative discretization methods, extensive experiments and their analysis. The method is demonstrated with data and these results show that with the tri...