Featured Student Stories

Whether you know it or not and whether you like it or not, data about you is constantly being collected. From your current location, to your advertisement engagement, and even your step-count, your behaviors are converted to 0s and 1s and sent to companies to crunch your numbers. While many for-profit companies will take your information and manipulate it for their own monetary gain, Timothy NeCamp, a fifth year Statistics PhD student, wants to use that data for good. With the objective of improving mental health, Tim uses data taken from mobile devices to track the habits and moods of people in stressful environments, like medical interns. Smartphones and wearable sensors provide real-time data streams to monitor physical and mental health and deliver interventions precisely when they are needed in a process termed mobile health. Point-of-care and diagnostics become immediately available and literally at hand through these devices. For example, devices that monitor heart rate and other symptoms of stress could send push notifications providing real-time advice on how to mitigate stress.

Just after an early morning run, Tim finds that his best research hours are in the morning where he’ll read papers, analyze data, and implement new algorithms to improve mobile health technology. Instead of physicians monitoring data to inform patients, the analysis is instead conducted computationally. Using artificial neural networks, a computational framework that mimics the brain’s neurons, vast amounts of biometric inputs like step count and heart rate can be used to diagnose patients and predict patient outcomes. The algorithms are also able to synthesize a large amount of information. One such analysis, called principal component analysis, is able to take many variables and identify the most relevant ones for understanding patient health. But programing isn’t everything; many times he turns to filling pages upon pages with derived mathematics. “Writing out the math,” he says “gives [him] a better insight into the core ideas behind the algorithm” and helps him come up with new ideas and strategies.

Outside his research, Tim uses his statistical skills to help non-profits perform data analysis and visualization. Leading a student group called STATCOM, he has had the chance to work closely with groups like Food Gatherers, a group that redistributes food around Washtenaw to combat hunger, and local school districts. Working with the Ann Arbor Area Community Foundation, a broad organization focused on improving the quality of life for the local community, has had a lasting impact on Tim, and he often has follow-up meetings about his work to allocate services for the aging population of Washtenaw County. Partnering and meeting with local organizations, Tim and other STATCOM volunteers work with the “messiness of real-world data” in a way that has a real impact on people’s lives. Even after some projects are completed, he still meets with the partner organizations to follow-up and implement improvements.

Tim’s attention to the community and its improvement makes even more sense given his teaching background. Through the program Teach for America, he taught 8th grade math for three years before coming to graduate school. While the kids’ enthusiasm was his favorite part, he found himself wanting to teach at a higher level and be more involved with scientific research. Still, he says, “Going from socializing and talking six hours a day to just being on a computer was a big contrast.” To socialize and decompress from work, he’ll go out with friends outside of his department to get fresh perspectives on life or go rock climbing. Ultimately, Tim hopes to be able to stay in academia, perhaps one day running his own research group to continue to use statistics to improve the world around him.

Imagine that you want to learn and understand a new mathematical concept. Do you opt for a visual demonstration of the concept? Or are you more comfortable with a verbal explanation?

Jarvis Miller is a graduate student in the Statistics department, though he spends most of his time working with researchers in the School of Information. He studies the ways students learn, particularly when it comes to learning university-level statistics. To do this, he uses data about 100,000 students from Massive Open Online Courses, or MOOC (www.mooc.org). Specifically, he’s interested in the types of questions students were asked and which questions the students answered incorrectly. He’s even interested in the different ways they answered the questions incorrectly, such as whether they missed a procedural step or completely confused concepts. With these features in mind, Jarvis wants to group students together and design experimental lessons to assess the effectiveness of different teaching methods (e.g. visualizations or verbal explanation) on the different groups of students.

How does one land in such a research topic? Jarvis described his frustration when tutoring students in statistics and his students’ own frustration toward not understanding the concepts in their statistics courses. “They would say ‘This doesn’t make sense.’ I started asking ‘Why doesn’t this make sense?’” He began to realize that his students were not taught to develop an intuition for what many statistics concepts actually mean or what to do with a dataset that isn’t as clear-cut as in a textbook. “Statistics is one of those courses...that tend to be necessary for every major. If something is so broadly taught, it should be broadly understood.”

As a high school student Jarvis disliked school, until he took physics with a teacher who engaged students by requiring them to rederive common equations of physics through interactive demonstrations, lab activities, and group discussions. He never envisioned himself going to college until that high school physics class. Jarvis also didn’t know what graduate school was until an undergraduate professor at Rice University learned of his passion for improving education and recommended pursuing a graduate degree. So he took the GRE just 13 days before application deadlines while in the midst of final exams, and he now gets to spend his time improving statistics education.

Though he’s only just begun his research project, Jarvis’s day-to-day life is far from boring. Armed with the sample of MOOC students, he develops algorithms to group students together, meets with his research advisor to brainstorm new ideas or discuss a recent paper, or interacts with students to learn more about different ways of learning. When he’s not analyzing data or devising clever ways to group the students in MOOC, he’s likely on a run, tutoring students in statistics, or grabbing a drink with friends.

When all is said and done, Jarvis hopes he will impact the way statistics is taught in universities. He plans to document what he learns about the different ways students learn and create an “instruction manual for instructors” to use for teaching statistics. He also recognizes that smaller colleges might not have access to the same resources as larger four-year universities and likely won’t have the same class sizes or demographics. Additionally, professors might not want to spend time developing their own interactive visualizations. So he wants to gather all of his visual demonstrations into an easy-to-use software package that will be openly available to anyone who wants to use it. Hopefully by the end of his career, Jarvis won’t only have improved statistics education, but he’ll have made learning statistics easy too.