Evolutionary dimensionality reduction models for predictive analytics in education domain

Abstract

newline Current era of digitization has generated and currently generating large amount of data in every field. Data thus generated has been mined by researchers to receive useful insights in various domains. Education Data Mining (EDM) / Learning Analytics (LA) is an emerging discipline concerned with developing methods for exploring the unique types of data that come from educational systems or settings, and using those methods to better understand education domain s stakeholders such as students, educators and academic authorities for better outcome. EDM provides a comprehensive advantage to modern academia in comparison of traditional methods. Effective and efficient predictions of academic performance parameters using dimensionality reduction - is an important area to focus on, demanding novel methods to develop. Creating or preparing an education dataset is also a challenging task in EDM. This research integrates notion of evolutionary algorithm with prediction models in EDM to yield prolific predictive analysis. Evolutionary algorithms are applied for dimensionality reduction in predictive analysis which gives better interpretability concerning academic parameters, reduced error and reduced execution time with fewer dimensions. Proposed prediction models are investigated on two datasets out of which one dataset is prepared by following proposed dataset preparation framework. The education dataset generated from large repository in this study is donated in open repository. This research work by and large focuses and provides solutions to EDM challenges: Data collection or generation in EDM, Designing effective and efficient academic performance prediction models and effective feature engineering. It throws light on investigation of role of evolutionary concepts in EDM. This research study would be helpful to researchers who want to contribute in field of EDM to assist education domain for technologically empowered academia. newline