Students' data science model could help avoid costly natural gas compressor shutdowns

D2K Showcase winner uses data from compressors to improve maintenance

Students' data science model could help avoid costly natural gas compressor shutdowns

A student project to predict the need for maintenance in natural gas compressors and avoid unexpected shutdowns has won this year’s Data to Knowledge Lab Showcase, a competition to highlight semester-long capstone projects by interdisciplinary teams that use real-world data to solve problems.

The May 5th virtual event hosted by the Data to Knowledge (D2K) Lab involved students from 16 Rice majors including computer science. A dozen teams from the D2K capstone programs, some sponsored by affiliates, and five teams from the Introduction to Data Science course took part.

“You can’t learn data science without getting your hands messy on real datasets that are big and complex,” said Genevera Allen, founder and director of the D2K Lab and an associate professor of electrical and computer engineering, statistics and computer science. “Congratulations to all the student teams who rose above in an unusual academic year and did a fantastic job presenting at the D2K Showcase.”

CS I Predict, sponsored by CSI Compressco, a provider of natural gas compression services, created a program that uses compressor sensor data to predict shutdowns in advance.

The team members are master’s student Baekjun Kim in computer science as well as James McNaney in electrical and computer engineering. Undergraduates included Bradford Johnson in computational and applied mathematics, Artun Bayer in electrical and computer engineering and Jasper Huang in statistics. Each of them will receive $250.

Two teams received honorable mention:

  • Team Customer Support Chats created a model to predict the real-time probability that customers will have a positive experience during live support chats. Team members are master’s student Jacky Li in computer science and undergraduate statistics major Chris McCarthy and computer science majors Daniel Tang, Ray William and Berk Alp Yakici. sponsored the team.
  • Team Cardiac Signals made an autoencoder-based detector of cardiac signal instability to give early warning of events in babies with congenital heart defects. (The team also won the D2K Lab’s Fall 2020 Virtual Showcase.) The team members, all undergraduates, are physics majors Aneel Damaraju and Frank Yang and electrical and computer engineering majors Kunal Rai, Chiraag Kaushik, Tucker Reinhart and Andrew Pham. Texas Children’s Hospital sponsored the team.

Each honorable mention-winning member will receive $100.

The Team OpenStax OpenSyllabus project won the Audience Choice award for its natural language processing model to convert unstructured syllabus text into a standardized, structured format. Team members are undergraduates Nicholas Glaze in electrical and computer engineering, Steven Feng in statistics and Timothy Newton in computer science and master’s degree candidate Yikun Li in electrical engineering. Each of them will receive $50.