Abstract:

When a class starts, students are getting ready for the class. Depending on how they prepare for the journey and situations that might affect their focus on the class, some of them may not fully be engaged in the learning. To help students onboard into classes, early intervention on those students who might be challenging in a class is necessary. Professors may raise concerns on students’ progress in classes so all other resources can be involved to provide help. Early progress survey of classes is an easy way to raise such concerns based on students’ performance in the first weeks.

For many classes, although class activities in early weeks are relatively light, student performance can indicate whether a student is fully engaged in the class and ready to perform required learning on advanced topics. Such early prevention approaches will help students, as well as all education supporting resources, to understand the issues the students are encountering.

Early prevention and identifying students who might need help is a critical process of many universities. The early prevention approach can be data driven through a predictive model. Data from previous classes can be collected and analyzed to identify individuals who might need help. The model can help understand how the activity performance in early weeks may impact overall student success in a class.

The model implemented in this paper provides an assistant way that is driven by data simulating student performance in the early weeks. The goal is to identify accurately students who might be failing in the class. The paper discusses the process of data collection and preparation for a classification model before the model is fitted on the data. The data can be collected from previous classes that reflect students’ performance. The data may cover student activities in early weeks and final class performance. We would like to predict student’s overall performance in a class based on data from early weeks. The predictive power of the model is evaluated and interpreted.

Keywords:

Data analytics, predictive model, classification, student success.