Dropout and transfer paths: What are the risky profiles when analyzing university persistence with machine learning...

Area of Science:

• Educational Psychology
• Data Science
• Sociology

Background:

• University dropout presents significant academic, social, and economic challenges.
• Previous studies faced limitations in data scope and analytical breadth.
• Analyzing student attrition requires a comprehensive perspective.

Purpose of the Study:

• To introduce a novel machine learning (ML) method for a holistic analysis of university dropout.
• To overcome limitations of previous studies by handling larger datasets and avoiding strong distribution assumptions.
• To enhance the interpretability of factors influencing student attrition.

Main Methods:

• Development and application of a machine learning model.
• Holistic analysis approach to examine university dropout.
• Utilizing a large dataset for comprehensive analysis.

Main Results:

• Model results align with prior research on personal and contextual variables, and first-year academic performance.
• Identified additional influential factors: dedication (part-time vs. full-time) and student age.
• Demonstrated the vulnerability of students based on age.

Conclusions:

• The proposed ML method offers advantages in handling large datasets and providing interpretable results for student attrition.
• Dedication and age are significant, previously underemphasized factors in university dropout.
• The model provides a more nuanced understanding of the complex phenomenon of university dropout.