Students from Rice University and across the globe came together for the 3rd annual Rice Datathon, a data science competition held virtually on February 5 – 6. Over 300 students registered for the event with 21% of them from outside Rice. The competition generated a variety of social impact and data science challenges faced by the Houston community, small businesses, the trading industry and telemedicine. Over $9,000 worth of prizes were awarded to students that participated in the competition.
A restaurant recommendation system to support small Houston businesses during the pandemic won the overall competition. Team food4thought topped 43 projects submitted by 159 final participants.
Team food4thought chose a topic from a predetermined list and were provided with related datasets to analyze. Team members, Rice Computer Science sophomores Sarah Han, Megan Xiao, Claire Xu and Andy Wang created an online platform that collects a user’s dining preferences and location information and delivers recommendations based on thousands of reviews of Houston restaurants. Each member of the team won $400.
“My biggest takeaway from the Rice Datathon is really seeing the potential of data, through other projects of the Datathon and our own,” Xiao said.
The next step for food4thought would be expanding it to other small businesses beyond restaurants and to other cities in addition to Houston. “COVID-19 has undoubtedly affected the entire world's local restaurants and small businesses, and we would love to use our platform to support them during these times,” Xu shared.
The team also won the Best Underclassmen award.
The second-place team, CovidCoughnet, created an algorithm that collects audio and recognizes the differences between COVID-19 coughs and non-COVID coughs, with an estimated accuracy of 90%. “We use data that is free and open source. Our approach is to use deep learning solution to diagnose different coughs,” team member Shryans Goyal explained.
Developed by Rice seniors Will Mundy majoring in Computer Science and Asian Studies, and Shryans Goyal in Computer Science, the program is not meant to replace contact testing but would give users a better sense of their possible infection from home, reducing potential transmission to others.
“The immense potential of deep learning in the medical space is something we're really excited about, and we hope to continue perfecting our diagnostic tool into a consumer-grade application that anyone could use,” Mundy said.
Team Titans - Telemedicine Trends won third place for its development of a means to measure how likely people in a census tract would be to use available telemedicine services. Team members were Rice admit Jacob Kasner, senior Franklin Briones (Bioengineering) and freshman Talia Frindell (Computer Science), and freshman William Hou of the University of California, Irvine.
“Our project succinctly identifies groups of high opportunity for telemedicine companies to focus on and could result in better healthcare access universally,” Kasner explained.
The team commented that this project was just a starting point. Moving forward, they look forward to sharing their work with a wider audience and a potential collaboration with 2nd.MD, who sponsored the challenge, on telemedicine.
“It would be fascinating to explore if/how the demographics and environment in those regions relate to telemedicine,” Hou added. “I could also see efforts building on the scoring system we threw together to construct a nation-wide model of telehealth trends that can be validated for accuracy.”
The team also won a track sponsored by 2nd.MD as well as the Best Visualization award.
Individual tracks are provided by company sponsors. The winners included:
Link, with an algorithm to recommend agencies and vendors for industry, sponsored by Bill.com;
Hotdog Sale Prediction, with a model that can forecast sales to avoid food waste due to the miscalculation of demand, sponsored by Chevron;
Make A Trade, a Bitcoin prediction and business logic program, sponsored by QuantLab;
Team Titans, Telemedicine Trends, sponsored by 2ndMD;
And social impact winner Cessation, which used machine learning to find correlations between smoke shops and poverty to propose policies that address tobacco addiction in Houston.
The article was originally posted on Rice News.