Development of Graduation Prediction Model for Industrial Engineering Students Using Decision Tree

Abstract

Data mining is one of the rapidly growing data processing sciences. One application of data mining is to build a model of student graduation criteria. In this study, a prediction model for student graduation criteria was developed using a decision tree (DT). Several factors that influence the graduation criteria of students studied in this study include GPA, field of study, age, number of credits completed, and so on. The development of this model uses the open source Rapidminer software which is proven to have ease in processing but is very good at producing models. There are 3 prediction models produced, namely the DT model using the Gini Index method, Information gain and gain ratio. The resulting model has a fairly large root distribution in the predicate is very satisfactory. This means this predicate in the process does quite a lot of iterations. These three models can be used to predict student graduation because they have an accuracy and Kappa value greater than 80%. This shows that this model has a high level of confidence and can describe what is happening. The Gini Index model has the highest accuracy and kappa value compared to the information gain and gain ratio models with accuracy and kappa values of 0.963 and 0.932, respectively. This shows that the Gini Index model is superior for processing large data.

Keywords

• data mining
• prediction model
• student graduation criteria
• decision tree