Applying Explainable Artificial Intelligence to Develop a Model for Predicting the Supply and Demand of Teachers by Region
- Teacher supply, Demand prediction, Artificial intelligence model development, XGBoost, XAI, SHAP.
Among various methods to improve educational conditions, efforts are being made to reduce the number of students per teacher. However, for policy decisions it is necessary to reflect multiple factors such as changes in the number of students over time and local requirements. Time-series analysis-based statistical models have been used as a method to inform policy decisions. However, the existing statistical models are linear and the accuracy of their predictions is inferior. Also, since there are both internal and external factors that influence the number of students and thus the prediction of the number of required teachers, it is necessary to develop a model that reflects this. Therefore, in this study, an artificial intelligence model based on machine learning was developed using the XGBoost technique, and feature importance, partial dependence plot, and Shap Value were used to increase the model's explanatory potential. The model showed a performance of less than 0.03 RMSE, and it was confirmed that among several factors the economically active population had the most significant effect on the number of teachers. Through this study, it was possible to examine the applicability of an artificial intelligence model with improved explanatory possibilities in predicting the number of teachers.