All Zicklin News

Zicklin Data Scientists Storm Pitney Bowes HQ to Crush Baruch College Data Challenge

January 2, 2024

Zicklin graduate students Connor Dooling (MS Statistics, ’23), Fengchu (Andrea) Lai (MS Business Analytics, ’24), and Hrithik Shukla (MS Business Analytics, ’23) are the winners of the Baruch College Data Challenge with Pitney Bowes.  

From left: Lai, Shukla, Dooling

Now in its fifth academic year, the Data Challenge—a collaboration between Pitney Bowes and the Zicklin School of Business that began in 2019—offers students an opportunity to bridge the gap between data science and a business problem by presenting technical findings to a non-technical audience; to succeed, they must draw on their skills in storytelling, summarization, and presentation.  

This year for the first time, the competition was held on site at Pitney Bowes headquarters in Stamford, CT. Five finalist teams (winnowed down from the original 25 that entered) each gave a 10-minute presentation to a panel of Pitney Bowes judges, with five additional minutes allotted for questions and feedback. Finalists also created technical posters of their presentations to discuss with other students and Pitney Bowes stakeholders during a networking event.  

The competition was based on a real-world data challenge that Pitney Bowes’ own data scientists face: how to forecast enterprise clients’ throughput at scale and draw inferences from the forecast and throughput characteristics.  

Student teams worked on the business problem for three weeks. Chaoqun Deng and Chung Eun Lee, faculty members in the Paul H. Chook Department of Information Systems and Statistics, assisted the teams by providing resource guides and a workshop on time series, and students could also check in with the Pitney Bowes team during office hours.  

“Through this challenge, I was able to grapple with the complex needs of various business partners and develop machine-learning models to aid in forecasting revenue,” said Connor Dooling. Hrithik Shukla said the “standout moment” of the challenge was “not only meeting but exceeding the accuracy standards set by Pitney Bowes, and achieving the lowest MAPE [Mean Absolute Percentage Error] score with our time-series forecasts.”  

Added Andrea Lai, “We learned invaluable lessons on leveraging machine learning to develop predictive models for revenue forecasting and delivering impactful business insights through hands-on, real-world projects.”  

“It was great to meet the finalist student teams and Baruch staff in person to discuss their solutions and ideas. The students did a tremendous job on incorporating quantitative data science methods and business understanding into their final read-outs,” said Christian Bernards, senior manager of data science for SendTech Solutions, a business segment of Pitney Bowes. “It’s been a pleasure to collaborate with the Zicklin School at Baruch College for the last four years, and we are glad that we took it one step further this year by adding the in-person finals,” he added.  


Categories: , ,