Improving On-Time Undergraduate Graduation Rate For Undergraduate Students Using Predictive Analytics

Overview

• The paper examines the issue of low on-time graduation rates among universities in Puerto Rico compared to the mainland United States.
• It aims to develop a predictive model that can accurately identify students at risk of not graduating on time.
• The researchers evaluated various predictive models using a dataset from the University of Puerto Rico at Mayaguez.
• The boosting model trained on an oversampled dataset performed best at predicting who would not graduate on time.

Plain English Explanation

The paper looks at a problem faced by universities in Puerto Rico - many students are not graduating on time. This is an important issue because it can have significant negative consequences for the students, the schools, and the local economy. The researchers wanted to create a way to predict which students are most likely to not graduate on time, so the universities can try to help those students and improve their graduation rates.

They tested out different machine learning models using data from the University of Puerto Rico at Mayaguez. The model that performed the best at predicting which students would not graduate on time was a type of model called a "boosting" model. This model was trained on a dataset that had been modified to have an equal number of students who graduated on time and those who did not.

The key idea is that by being able to accurately predict which students are at risk of not graduating on time, the universities can try to provide extra support and resources to help those students succeed and graduate on schedule. This could have big benefits for the students, the schools, and the local economy.

Technical Explanation

The researchers developed and evaluated various predictive models using a dataset containing information on 24,432 undergraduate students at the University of Puerto Rico at Mayaguez. They looked at two different scenarios:

1. Group I models used both first-year college factors and pre-college factors as inputs.
2. Group II models only used pre-college factors as inputs.

For both scenarios, the boosting model trained on an oversampled dataset performed the best at predicting which students would not graduate on time. Oversampling the dataset involved duplicating instances of the minority class (students who did not graduate on time) to create a more balanced training set.

The researchers also experimented with other models like logistic regression, decision trees, and neural networks. However, the boosting model consistently outperformed these other approaches in terms of accurately identifying at-risk students.

Critical Analysis

The paper provides a valuable contribution by demonstrating the potential of using machine learning to proactively identify students who may struggle to graduate on time. However, there are a few important caveats to consider:

1. The research was conducted using data from a single university, so the findings may not generalize to other institutions in Puerto Rico or beyond. Further studies across a broader range of universities would be helpful.
2. The paper does not delve into the specific factors or student characteristics that the models used to make their predictions. Understanding these "black box" decision-making processes could lead to more interpretable and actionable insights.
3. While the predictive performance of the models is promising, the paper does not explore how the universities could actually intervene to support the identified at-risk students. More research is needed on effective retention and support strategies.

Overall, this research represents an important step forward, but there are still opportunities to build on these findings and develop more comprehensive solutions to the problem of low on-time graduation rates.

Conclusion

This paper tackles the critical issue of low on-time graduation rates at universities in Puerto Rico. By developing a predictive model that can accurately identify students at risk of not graduating on time, the researchers have laid the foundation for universities to proactively intervene and provide the necessary support to help these students succeed.

The strong performance of the boosting model, especially when trained on an oversampled dataset, suggests that advanced machine learning techniques can be valuable tools for addressing this problem. If universities can use these insights to implement effective retention and support strategies, it could lead to substantial benefits for students, institutions, and the broader Puerto Rican economy.

Overall, this research represents an important step forward in using data-driven approaches to tackle complex challenges in higher education. As the field continues to evolve, we can expect to see more innovative solutions that leverage the power of predictive modeling to drive positive change.