Classification of internship students On-demand schools with machine learning techniques.

Abstract

This research aims to develop a model for classifying student teachers according to the needs of network schools, and to determine the effectiveness of the model for classifying student teachers based on the needs of these schools. The sample consisted of schools within the responsibility area of the university, specifically in Bangkok. Five districts were selected through simple random sampling, and two schools from each district were chosen using quota sampling. Data was collected from teachers within the schools. Due to the varying sizes of the schools, the number of teachers ranged from approximately 10 to 30. Therefore, data was collected from 10 teachers in each school using convenience sampling, resulting in a total of 100 teachers across 10 schools. 

          When assessing the model’s effectiveness using the Confusion Matrix method for evaluating the classification of student teachers according to the needs of the network schools, the results were as follows: The accuracy score of the developed model was 92 percent. The precision score for most schools was 1.00, indicating perfect prediction accuracy, except for School 3, where the precision was 0.83 or 83 percent, and School 4, where the precision was 0.67 or 67 percent. The recall score for most schools was 1.00, except for School 1, where it was 0.80, and School 2, where it was 0.33. The F1-Score for most schools was 1.00, except for School 1, which scored 0.89, School 2 scored 0.50, School 3 scored 0.91, and School 4 scored 0.80.