Analytics Lab

Changing industries with data

A-Lab projects are all about tackling business challenges using data and machine learning in a wide variety of industries, including healthcare, e-commerce, sports, and many others. Below is a sampling of some of the incredible ways that students have used analytics to address crucial issues that companies face.

A-Lab Impact Stories

Graphic of woman's hand touching medical plus sign icon with light blue background Action Learning The Power of Patients—and A-Lab Rental bikes and scooters parked on sidewalk with bus on the road in background Action Learning Rethinking urban mobility in Europe Blue graphic of globe and satellite with lines around them Action Learning Machine learning project supports UN Mine Action Service Two children kneeling over a wooden box of colorful chalk Action Learning Developing a web app dashboard to identify children's quotes for greeting cards

Past projects 

Challenge: Determine the effectiveness of a leading paint manufacturer’s marketing strategy across South America using thousands of raw unlabeled images of storefront displays.

  • SOLUTION: The student team built a ready-to-use, customizable, image recognition tool to measure in-store brand presence. The tool will inform how the company can better allocate its marketing budget and allow decision makers to further explore the effect of visual presence on B2B and B2C sales, product presence, and pricing.

Challenge: Given a medical diagnosis with many potential treatments, identify which patients may be "super-responders" to particular course of medication in order to provide patients with more personalized treatment plans.

  • SOLUTION: The student team leveraged the publicly-available MIMIC-IV Electronic Health Records (EHR) data set and a biomedical knowledge graph to create a predictive model that effectively compared a new patient to other patients in the system and predicted the best course of treatment given their medical history, demographics, and labwork. The student team provided recommendations to adjust randomized clinical trials to recruit and account for these so-called super responders.

Challenge: Programmatically measure the success of and improve a new customer service chatbot.

  • SOLUTION: Using natural language processing tools, the team automated the classification of successful and unsuccessful customer interactions and developed a method to identify specific areas where the bot failed. This allows the company to improve bot performance by creating the required volume of data and allowing better training on specific tasks.