THE DEVELOPMENT OF DECISION SUPPORT SYSTEM FOR PREDICTION OF NEW UNDERGRADUATE STUDENTS INTAKE IN GOVERNMENT UNIVERSITIES BY MACHINE LEARNING AND ANALYTIC HIERARCHY PROCESS

Abstract

This research aims 1) to study the students’ decision data for studying in the government universities, 2) to develop the decision support system and prediction for studying in the government universities by integrating the machine learning and the Analytic Hierarchy Process (AHP), and 3) to assess the system performance.  This research was classified into 3 steps: 1) studying the decision criteria for studying in the government universities from related documents and research and in-depth interview with 30 guidance teachers in high schools in the areas of Phetchabun Province by purposive sampling for creating the AHP  decision model, 2) developing the system by  integrating the machine learning and the AHP for improving performance in creating decision alternatives and predicting the best results and reducing the training time use, and 3) assessing the system performance by comparing the accuracy of the multi perceptron neural network algorithm and the decision tree algorithm. This research found that the AHP decision structure consists of 10 criteria and 5 alternative subject majors. The algorithm of system processing consists of 10 steps and the decision tree with fold 5K, 7K, 8K and 10K which have highest accuracy in predicting subject major according to students’ attentions at 96.7%. The model training by the decision tree is faster than the MLP 1.33 seconds but the model testing by   MLP is faster than the decision tree.