An 80-year-old Peruvian paint company turned out to be the ideal candidate for a technological refresh at the hands of an innovative MIT Analytics Lab (A-Lab) student team. In fact, four Masters of Business Analytics students — Nader Hoballah, An Luong, Antoine Roncoroni, and Alexandre Saillard — worked with Qroma to improve its marketing strategy with a customizable, image recognition tool to measure in-store brand presence.
At Schneider Electric, a leading energy management and automation company, the problem was to free up quotation specialists for customer-service tasks. Its A-Lab team — Rajlakshmi De, Zhechao Huang, Pierre Jockers, and Jixin Wang — was challenged to suggest most likely answers to the sales configurator questions using an iterative process where the recommendation system “became smarter” with each response.
For the past five years, graduate student teams like these have developed projects using analytics, machine learning, and other methods of analysis to solve real business issues and create new opportunities. The course, which runs each fall semester, is presented by the IDE and is part of MIT Sloan School of Management’s Action Learning offerings. It is led by IDE leaders, Erik Brynjolfsson and Sinan Aral. Presentations were made and judged in December, with Qroma named the winning team by a panel of “celebrity judges”.
The challenge at Qroma was to create a data product to measure brand presence at client stores that will help drive future marketing. To do this, the students built an image recognition tool to classify brands based on raw photos at the retail stores. Once they trained the model to recognize the images and created labels, classification and identification was 90% accurate, on average.
Projects ranged from predicting the future of jobs based on ad postings, to optimizing the efficiency of matching in a two-sided marketplace, segmenting and classifying brick and mortar store visitors using Wi-Fi data, and predicting unplanned locomotive maintenance. The runners up were the teams working with Schneider Electric and Falabella.
The Schneider team found that as more data was input and its four new algorithms were implemented, datasets improved and error rates were reduced. They also created a chatbot to shift work hours to customer-service tasks.Schneider team members answer questions from the judges.
The Falabella team, Jun Siong Ang, Samuel Lee, and Josh Lester, used machine learning to determine the relationship between purchasing data and loan potential. They found that spending does predict loan conversion which can help the bank identify potential customers based on spending patterns.
Admissions for Fall 2019’s Analytics Lab opens to MIT students in early May.
Read the full blog on Medium, here.