เทคนิคการพยากรณ์ปัจจัยที่ส่งผลต่อการพิจารณาอนุมัติข้อเสนอโครงการบริการวิชาการ

Abstract

The purpose of this research was to study the factors affecting the approval of academic service project proposals. Academic service project proposals supported the decision-making of the people involved, making it faster and more convenient. Data mining was used to analyze data in accordance with the CRISP-DM process and to obtain the most relevant factors for research purposes. Once the appropriate factors had been identified, the data processing was performed by using data analysis to find the optimal model, factor, and forecasting technique. In this research, three techniques were selected : Artificial Neural Networks (ANN), K-Nearest Neighbors (K-NN), and Support Vector Machine (SVM). The study develops 3 sets of factors. The first set of all factors (set 1) is a fundamental factor in the decision-making process. It consists of 16 variables: Location, Servicer, Personnel, Day, Budget, 5 variables of Suitability Assessment and 6 variables of Data Verification. The second set of factors (set 2) was developed by merging some factors of the first set of factors together. The Servicer with Personnel were merged to Partner, and Day with Location were merged to place. So the second set of factors had 14 variables. The third set of factors (set 3) was developed by adjusting the value from the factor (set 2) to minimize the number of variables, in order to increase processing efficiency. This was done by grouping all values stored in the Suitability Assessment into one variable. Among 3 sets of factors, it was found that the third set of factors had the highest accuracy of forecasting, and was the most appropriate form of reference for predicting the desired result. The results from this study revealed that the ANN was proved to be the most appropriate technique for providing the highest accuracy in prediction regarding the approval of the academic service proposal.