MIT EECS Undergraduate Research and Innovation Scholar
Educational Data Mining to Analyze MOOCs
Massive Open Online Courses (MOOCs) forgo traditional methods of teaching in a classroom setting. Instead, all lectures, readings, problem sets, and projects are done online. Thus, there is a vast amount of behavioral data that could be modeled to explain how students study. I will be analyzing and developing new models using machine learning techniques to predict student behavior, such as dropout rate. My focus is in creating a latent variable model which measures a student’s motivation based on raw data collection. I will be collaborating with education scientists to help translate concrete data into latent variables
My interest in big data and data mining first started in my freshman year of IAP, where I competed in the Pokerbots competition. The competition provided log files of the thousands of hands played between your bot and all your opponents. After seeing how useful yet challenging the hand log files were to refining the strategy of our bot, I became interested in more real-world applications of data mining.