Student Performance Prediction in Engineering Design

• In Event: AERA Poster Session 14
  In Poster Session: Advanced Technology and Learning Poster Session

Abstract

Providing timely support to students’ learning of engineering design has been challenging using the traditional assessment methods. This study takes an initial step to employ learning analytics to build performance prediction models to identify struggling students. Specifically, to address the data sparsity and high dimensionality problems, a prediction workflow including a two-stage feature selection combined with an advanced supervised machine learning algorithm was designed and tested in a real engineering design class. Further, given that previous prediction research is usually implemented at arbitrary time points without consideration for the fluctuation of prediction model, this work proposed a brute force way to identify the best time point to run the prediction so that the performance is optimized and early enough.

Authors

• Wanli Xing, University of Florida

• bo pei, University of Florida

• Shan Li, McGill University

• Charles Xie, Concord Consortium