Enhancing Online Learning Through Feedback Analytics Using Descriptive Analytics and Topic Modeling

Abstract

As online education continues to evolve, particularly within pioneering institutions offering fully online degree programs, the need for data-driven quality assurance becomes increasingly critical. From Academic Year 2018 to 2025, a substantial volume of student feedback was collected through Course and Mentor Evaluation (CME) and Voice of the Customer (VOC) surveys. The CME and VOC datasets offer greater insights into the learner experience. However, the amount and complexity of textual feedback present challenges for standard manual analysis. This study aims to extract meaningful insights from student feedback using a combination of descriptive analytics and Natural Language Processing (NLP) techniques—specifically, sentiment analysis and topic modeling. A total of 36,142 valid entries from the original dataset were kept after a rigorous data-cleaning procedure. To guarantee data quality and consistency, entries with missing, duplicate, or invalid responses were eliminated. Descriptive analytics were used to find recurring patterns and common problems, while topic modeling assisted in exposing underlying themes in the comments. A more detailed view of service gaps, student happiness, and instructional effectiveness is made possible by this dual approach. The study's findings aim to inform institutional enhancements in mentor interaction, course content, and the overall delivery of online learning. The research helps to continuously improve the quality of fully online higher education environments by creating a framework that is influenced by feedback.