ABSTRACT
The last decade has seen a major shift in the education system with exponential growth in techniques for learning, knowledge dissemination, and sharing. The increased digitization of education has generated large volumes of structured, semi-structured, and unstructured data. Analysis of this data requires the use of various analytical techniques ranging from descriptive and diagnostic analytics to machine learning-based predictive models. It also creates possibilities for intelligent pattern detection models resulting in personalization of educational experiences, customization of content, and enhanced learning outcomes. This paper proposes a three-tiered framework for educational data analysis. The three layers of the framework map to analytical techniques for structured data, semi-structured/textual data, and unstructured multi-modal data, respectively. The first layer considers numerical/quantitative analysis of structured education data. The second layer proposes the use of Natural Language Processing (NLP) and text mining techniques for the analysis of semi-structured and textual data. The third layer examines unstructured/multimodal data and proposes the use of deep learning techniques to analyze images, audios, and videos. This three-tiered model detects patterns enabling personalization and customization of learning approaches. The relevance of this framework is that it provides direction to decision-makers for implementing an educational analytical model for better insights.
