Machine Learning Techniques For The Ordinary National Educational Test (O-Net) Prediction: Case Of Small Sized Schools In Nakhon Sawan Province, Thailand

Abstract

The objectives of this research were to analyze and predict the outcomes of Ordinary National Educational Testing (O-NET) and to compare the results of Machine Learning techniques with 4 data analysis techniques, namely K- Nearest Neighbor technique, Logistic Regression technique, Decision Tree technique, and Support Vector Machine technique. They also use Microsoft Excel 2013 and Jupyter. The data used to forecast trends of O-NET scores were scores for each subject of schools in KrokPhra District, Nakhon Sawan Province, namely Thai, math, science and English by showing counting data for the quality level of the data used for the forecast analysis. The trend of O-NET scores showed that most of the students' data were at a fair level. The average scores in each subject showed that the average scores for most Thai subjects were at good level. The average scores in most math subjects were good. The average scores in most science subjects were good. And the average scores of most English subjects were at a good level, respectively. And comparing the results of the machine learning method, it was found that the Support Vector Machine technique was the best predictor. It has 65.75% accuracy and the model has more efficiency than other techniques.