An Assessment of YouTube Educational Video Quality Through Machine Learning

Abstract

This paper presents a machine learning-based approach to assessing the quality of educational videos on YouTube. With the growing number of educational videos available on the platform, verifying efficacy and credibility has become increasingly important. To address this issue, we propose a framework that allows for automated evaluation of video content based on analytic parameters provided by the YouTube Data API. We look at features such as view count, comment polarity, channel subscriber count, like/dislike ratio, and comment count to capture various dimensions of video quality and engagement. We introduce our target variable as the "Educational Quality Score" (EQS), which measures a video's impact on education using the RACED model - Relevance, Accuracy, Clarity, Engagement, and Depth. Our method offers educators and content creators a tool to enhance their selection of quality educational YouTube videos.