Endicott and Net Atlantic Partner on North Shore Data Modeling Competition
The quickly-growing field of predictive data modeling -- using collected data to make accurate predictions -- is in use all around us: Netflix or Amazon customer suggestions, future business cost planning, advertising targeting, and more. For students, learning predictive modeling and data analytics gives them marketable, highly-sought-after skills to put to use in the job market after graduation. To help better prepare students for professional growth and to introduce them to the world of data analytics, Endicott College and Salem-based email marketing company Net Atlantic partnered to create the North Shore Data Modeling Competition.
Winners, announced during the May 3rd awards reception, are teams from Gordon College and North Shore Community College. In first place was Matt Versland from Gordon College, who received a $1,200 cash prize. Second place and an $800 cash prize went to Matt Carleton, Tyler Carey, and Chrissy Pace of North Shore Community College. Participants included nine teams: four from Endicott College, two from Gordon College, one from Salem State University, one from North Shore Community College, and one from Beverly High School.
“Predictive modeling is growing in popularity, inspired by the success of companies like Facebook, Google, and Netflix who use predictive model data to help inform their business decisions and improve their service or product,” said Phil Lombardo, associate professor, mathematics at Endicott College, who ran the competition with Endicott’s Hank Feild, assistant professor, computer science. “Many other businesses are following suit. Evidence for this trend is a growing demand for data analysts, statisticians, and data scientists.”
The competing students tackled an original email marketing predictive modelling problem using real, anonymized data provided by Net Atlantic. The goal of this predictive model is to forecast whether an email newsletter will have high success, medium success, or low success based on the provided features in the training data. This predictive model would help create an effective subject line for a marketing email before sending it.
“A good predictive model can inform business decisions,” continued Lombardo. “For example, if I can predict the month where gasoline prices are cheapest, I may decide as a business owner to ship the majority of my goods to local warehouses at during that time. If I can use a predictive model to provide customers with reliable recommendations (think Netflix), I can improve their user experience and drive more customers to use my products.”