Machine Learning Models for Salary Prediction in Peruvian Teachers of Regular Basic Education

This paper presents an analysis of machine learning (ML) models to predict the salaries of 108,317 appointed teachers in Ventanilla, Lima, Peru, using recent data. The focal point of the study is appointed teachers, deliberately excluding salaries of hired teachers’ from the analysis. A significant result of this research is the identification of a new ML model capable of predicting teacher salaries with considerable accuracy, based on regressor variables closely related to salary. This finding is noteworthy because it fills a gap in existing ML applications for salary prediction, indicating a promising direction for future research in this area. The methodology used to analyze the wage data, while comprehensive, does not account for gender differences, which may affect wage variation over the five-year period considered. This oversight suggests that future research should include a wider range of variables, including gender, to improve the accuracy and applicability of salary predictions for both appointed and contract faculty. Such an approach could provide more nuanced information on the factors influencing teacher salaries and help develop more equitable and effective salary models. One of the key contributions of the article is the detailed examination of the factors influencing salaries of appointed teachers, including age, educational level, length of service, teaching scale, and hours worked. The use of linear regression, Ridge, Lasso, and Elastic Net models yielded accurate metrics for choosing the best model for salary prediction. This research not only advances our understanding of the determinants of teacher salaries in Peru, but also provides a valuable framework for similar studies in other contexts. Comparison with other research highlights the robustness of the chosen ML models, underscoring the potential of ML in educational administration and policymaking.