Knowledge Discovery on the Data on Dissolution of Classes of the Ateneo de Davao University

Abstract

Dissolution of classes is a constant dilemma of the Ateneo de Davao University. Although, data on dissolved classes are not directly available, large and distributed data sets on similar and related context are present like data on student registration, and academic classes, enrolment logs, class schedules and effective curriculum has been analyzed to discover patterns that lead to the understanding of class dissolution. Data since 2004 was gathered, processed and analyzed using supervised and unsupervised data mining techniques and methods. The data revealed non-linearity and the nonlinear regression model built and cross-validated gave an R of 0.9929. Running A Priori association also resulted to rules (confidence => 99% and support => 35%) that gave insights to the class dissolution problem. Even with the general tendencies of the data towards non-linearity and dynamism, some order and pattern were derived allowing some control and predictability (using the M5-based Pruned Tree Model). With the knowledge derived from the class dissolution data, key themes of chaos theory were derived.