Modeling Code Analyzability at Method Level in J2EE Applications One of the main reasons for improving code structure is to make the cause of error code easily identified. Thus, developers need an analyzability prediction model to evaluate the analyzability of code in order to locate classes to be improved. To identify classes to be improved, developers must analyze all methods of classes for finding problem methods. Therefore, analyzability prediction model must be created for calculating analyzability level of method. Currently, J2EE applications are legacy systems and need to be continually maintained. Hence, code analyzability prediction at method level in J2EE application helps developers to know which method should be improved for increasing code understanding and reducing time for finding error causes. However, there is a lack of analyzability prediction model for J2EE application and existing research works on analyzability prediction model do not focus on method level. Therefore, this paper proposes how to create analyzability prediction model at method level for J2EE applications through ordinal logistic regression.