Analyzing the Capabilities of Nature-inspired Feature Selection Algorithms in Predicting Student Performance
This work addresses the need for timely and effective predictions to support at-risk students in educational settings, representing an incremental improvement by combining existing methods.
The paper tackled the problem of predicting student performance by analyzing nature-inspired feature selection algorithms within ensemble methods, finding that this approach significantly increased predictive accuracy and reduced feature set size by up to 65% across three student datasets.
Predicting student performance is key in leveraging effective pre-failure interventions for at-risk students. As educational data grows larger, more effective means of analyzing student data in a timely manner are needed in order to provide useful predictions and interventions. In this paper, an analysis was conducted to determine the relative performance of a suite of nature-inspired algorithms in the feature-selection portion of ensemble algorithms used to predict student performance. A Swarm Intelligence ML engine (SIMLe) was developed to run this suite in tandem with a series of traditional ML classification algorithms to analyze three student datasets: instance-based clickstream data, hybrid single-course performance, and student meta-performance when taking multiple courses simultaneously. These results were then compared to previous predictive algorithms and, for all datasets analyzed, it was found that leveraging an ensemble approach using nature-inspired algorithms for feature selection and traditional ML algorithms for classification significantly increased predictive accuracy while also reducing feature set size by up to 65 percent.