A methodology to design, develop, and evaluate machine learning models for predicting dropout in school systems: the...

Area of Science:

• Educational Data Mining
• Machine Learning in Education
• Sociology of Education

Background:

• School dropout poses significant societal and economic challenges, exacerbated by the COVID-19 pandemic.
• Existing predictive models often lack a comprehensive view of individual student circumstances.

Purpose of the Study:

• To propose and evaluate a machine learning methodology for predicting individual student dropout risk.
• To develop a robust model that incorporates student, family, and school-level factors and their changes over time.

Main Methods:

• Development of a machine learning model using administrative educational data.
• Analysis of individual student trajectories, including the accumulation of events and changes in circumstances.
• Application of the methodology to the Chilean educational system.

Main Results:

• The developed model demonstrates a 20% improvement in predictive capability for actual dropout cases compared to previous research.
• Incorporating non-individual dimensions significantly enhances dropout prediction accuracy.
• The model successfully identifies individual student dropout risks.

Conclusions:

• The proposed methodology provides a robust framework for predicting school dropout.
• The model's enhanced predictive power supports targeted interventions and public policy.
• This approach offers valuable insights for educational systems seeking to reduce dropout rates.