Massive Open Online Courses have revolutionized global education by offering accessible and flexible learning opportunities. However, high attrition rates remain a persistent challenge, limiting the full potential of MOOCs. This study addresses this issue through the application of machine learning techniques to develop predictive models for identifying potential dropouts within the SWAYAM platform's MOOC offerings. Utilizing a real dataset from an ongoing MOOC, “Mathematical Economics” on SWAYAM, with over 10,000 student records, this research focuses on weekly assessment scores as the primary dataset feature. The study encompasses three distinct datasets, representing course progress at 4, 6, and 8 weeks, providing a comprehensive evaluation of predictive model performance across different timeframes. The study evaluated various algorithms, including k-Nearest Neighbors (kNN), Naive Bayes, Logistic Regression, Gradient Boosting, and AdaBoost, for their predictive capabilities. The results consistently highlight the superior performance of Naive Bayes in predicting dropout instances across all dataset lengths. Notably, Naive Bayes achieved an F1-Score of 0.999, 0.998, and 0.996 for the 4-week, 6-week, and 8-week datasets, respectively. These findings underscore its exceptional ability to balance precision and recall, maintaining high levels of accuracy and precision.