SFU Statistics and Actuarial Science Partners with MOJ.IO To Host Case Study
The Department of Statistics and Actuarial Science at Simon Fraser University held an industrial case study competition on April 24 (see http://people.stat.sfu.ca/~dac5/casestudy2017/casestudy2017/Case_Study_2017.html). Dave Campbell organized the competition, in partnership with Taran Gray from local Vancouver firm Moj.io. Moj.io kindly provided both a real data set and $2000 in prize money to reward students for their hard work.
Students were challenged with a data set derived from 383 cars with sensors that recorded 158 variables every 10 seconds from every car trip over several months. Altogether there were over 2.4 million rows of real industrial data. Students were asked to solve any or all of five potential problems: devising a driver quality score; determining if there are multiple drivers using a single car; inferring if a car was driving in the city or on the highway in a given trip; developing a wear-and-tear score on the vehicle; and estimating the time required to recharge a car battery. The Department offered two "getting-started" computer lab tutorials and supported the students with drop-in office hours. Throughout this time, they also had access to a Canvas page, which was monitored by Moj.io representatives as well, so they could "ask the experts" when certain questions arose.
Students had six weeks to answer any of the case study problems that they wished to pursue, culminating in a presentation event that took place three days after the Spring semester final-exam period. Despite having deliberately vague problems to tackle, and an industrial data set whose was realistic size created computational challenges larger than anything the students had seen in class, seven groups completed the challenge. Teams drew on their classroom skills and built on insights from their own driving experience to create models and evidence-backed statistical insights. The solutions were creative, diverse, innovative, and well executed. Six of the groups gave presentations to a panel of judges from industry and academia:
Kenneth Wong (Developer Evangelist & Technical Program Manager at Moj.io Inc.)
Lawrie Gaffney (Head of Customer Success at Mojio)
Soyean Kim (Lead, Research & Analytics at BC Safety Authority)
Dr. Tom Loughin (Professor and Chair, Department of Statistics and Actuarial Science)
Yonel Yonkov (Lead, Big Data & Analytics at Samsung Electronics)
The audience also included industry representatives from Moj.io and other local tech companies.
All of the presentations were excellent and demonstrated students' creativity by offering vastly differing approaches to solving the problems. The judges felt that all teams provided insight into the problems that would have offered value to a company in a real employment setting. Differences in quality were slight, but consensus was reached on a final ranking:
First place: " 'Mean' Girls": Brad Smallwood, Kristen Bystrom, Matthew Reyers
Second place: "Team YOLO": Charlie Haoxuan Zhou, Trevor Thomson, and Yi Xiong
Third place: "Sun and Pheonix": Anson Hsu & Pheonix Huangqi Hu
Fourth place tie: "Team Steve": Stephen Kane; "Vexen Software": Jordan Cho-Siksik and Michael Cline; and "Steven Segall Hard to Kill": Forrest Paton & Dani Chu
In addition, one team (Mike Grosskopf) completed the challenge but chose not to prepare a presentation.
The Department congratulates the participants and wishes to express deep thanks to Dave Campbell and to Moj.io. Dave worked many extra hours to ensure a high-quality, training-relevant event with incredible levels of industry engagement. The enthusiasm of the Moj.io team was evident in all interactions through Dave and directly with the teams.
Some photos from the event: