Polygence mentor Ben introduces students to data science through project-based learning

Ben, a Stanford PhD candidate in Education, is the epitome of the two major qualities we seek out in our mentors: he’s an expert in his field of study — data science — and he’s a multifaceted, extensively experienced educator. Ben is not only working on his PhD in data science and education, but he has also been a high school math teacher through Teach for America. In between these experiences, he spent three years developing math curriculum for Khan Academy and two at the University of Chicago’s for a Data Science fellowship. For his PhD, Ben is now combining his two passions in the most perfect way. He is conducting research in education using data science, capitalizing on the abundance of data that is now available about how students learn after the explosion of online learning platforms in recent years. “We have unbelievable amounts of data within education today. My focus is educational data science, which I would broadly characterize as how to extract value out of that data. How can we give students better questions to help them learn more? How can we make smarter decisions about what material a student has mastered?”

As a Polygence mentor, Ben guides his students through their very own data science projects, in which they can wrangle any data set and tackle any research question of their choosing. “Polygence is project-based and one-on-one mentor based, which I think is ridiculously valuable in all of education, and especially for learning data science. Data science as a discipline and as an industry is project based... Those projects have a general sequence of steps that a student can get really good at, but the only way is by doing a project themselves. By far the best way to do a project is with the one-on-one support of a mentor.”

Ben is mentoring several students through sports analytics projects, tackling questions like — How can a football coach make the best decision after a touchdown: should the team go for a two point conversion or kick for a field goal? — and — How can we use machine learning to predict with high accuracy what kind of pitch a baseball pitcher will throw? Another of his students is studying the data of over a million people’s responses to the Big Five personality test, and investigating whether this data can be predictive of certain behaviors.

Ben cites University of Wisconsin Economics professor Ken West as a highly influential mentor in his life. Rather than teaching from a textbook, Dr. West guided his small seminar of students through their own individual economics data projects throughout the semester. “What I did with Ken West is similar in flavor to a Polygence project. It really was an inflection point in my development — being able to go from classes and individual skills, to going through a project with the support of Professor West.”

Polygence is such an exciting opportunity for Ben because it lets him put into practice something he continually reflects on as a data scientist, educator, and curriculum designer—how to best teach data science. “I've thought a lot about the pedagogy of how you help someone build the skill of taking a data set, writing code, figuring out what's going on in the data, creating graphs, building models, reporting those models and so on. How do you teach someone that process? This is essentially what I get to do with Polygence, so I have an absolute ball doing it. I deeply enjoy working with students who come up with such exciting ideas, who get so much joy out of learning the data science process.”

Ben’s students learn how to execute each step of a data science project—1) locate an interesting data set, 2) hone in on a research question, 3) analyze and model the data, and finally, 4) communicate the results. “The thing to highlight here is that the data science workflow is incredibly iterative — it's not this sequence of steps that you do in order. Rather it's a bunch of different things that you have to say—Okay, well, I made this graph. Okay, now I understand the data better, and now I can ask a slightly different question, but now I might need to make a new graph. We iterate through these steps in a certain way.”

Additionally, Ben’s students are learning what he terms the “hidden curriculum of data science education,” meaning all the little things which no one tells you explicitly how to do in a class, but rather that you pick up on through experience. “Data science is pieces, and if you get really good at those pieces—how to write code to analyze data, how to organize my workflow, what folder structure do I use, how connect my work to GitHub etc.—then you can be really organized as you're working. It synergizes well with being able to do the sort of actual data science of the four steps that I outlined.”

In Ben’s studies of education, he’s found that there’s no denying the superior value of one-on-one mentorship as opposed to a traditional classroom. “In the typical classroom, you're literally a passive consumer, and you have no influence over what's happening, which is totally the opposite of what happens when a Polygence mentor works with the students — the mentor is adapting and it's this beautiful relationship where both people are adapting to one another. You just get so much further, and this is an incredibly valuable and unique aspect of Polygence.”

Ben’s advice to incoming Polygence students is to visualize their project as their own work which they have all the creative control over. “As a student, it's often very easy to defer to the mentor... But I’d encourage you to think about it being you in the driver’s seat and your mentor in the passenger seat, not the other way around… Come up with your own ideas for your project, take inspiration from anything, and share these ideas with your mentor, even if — especially if — they're ridiculous. It'll help your mentor think of the direction in which to steer you, and it'll be a much, much more interesting project for you.”