Enhancing student retention: A predictive study on dropout rates and remedial strategies

Education is important for economic and social well-being where dropout is a concern for students and institutions worldwide. This thesis seeks to improve students' retention by predicting the students' dropouts and remedial strategies. Prediction algorithms can identify students with poor academic achievement, demographics, and social conduct who are most likely to drop out. In this thesis, the machine learning techniques are used to build a predictive model to analyze dropout or retention of the students at early stages. Key features like grades, age of enrollment, courses taken, courses approved, gender, tuition fees status, class duration etc. are evaluated and analyzed using Random Forest classifier and correlation analysis. The findings of the thesis show that institutional issues, socio-economic factors, academic challenges, personal and psychological factors have a significant impact on dropout rates. The output shows the students who are likely to be retained or dropped out from the institution with a data-driven predictive model that can be helpful to structure and increase strategies for retention. This thesis not only helps in shaping retention strategies but also can provide detailed analysis for advisors, administrations and policy makers to create a support mechanism by applying such educational structure that can be shaped for students having different economical, social, educational and psychological backgrounds.