Classification Model Comparison for Predicting Professional Fields

Currently, peoples who study at the Institute of Digital Arts and Science, Suranaree University of Technology, have to select one of the professional fields between Digital Technology and Digital Communication. The professional field was very important because it will have a direct effect on course modules to be studied and a future career path. The purpose of this research was to construct and compare the classification performance of professional fields prediction model by collecting and analyzing data from the student opinion questionnaire. The classification technique which is part of a data mining approach was applied with 5 algorithms, e.g. Decision Tree, Naïve Bayes, OneR, Support Vector Machines, and K-Nearest Neighbors. The performance of the classification model was evaluated and compared by Precision, Recall, Accuracy, and F-Measure with 10-folds, 20-folds, and 30-folds cross-validation. The evaluation results indicated that (1) the classification model obtained from the Naïve Bayes algorithm achieved the highest Accuracy at 89.6%, using 20-folds cross-validation; (2) the classification model derived from the Support Vector Machines algorithm achieved the highest Precision at 89.6%, using 30-folds cross-validation; and (3) the classification model obtained from the Decision Tree algorithm achieved the highest Recall and F-measure at 83.3% and 82.5%, respectively, using 10-folds cross-validation.