By clicking “Accept All Cookies,” you agree to the storing of cookies on your device to enhance site navigation and analyze site usage.

Skip to main content

A-Lab Puts Analytics and Machine Learning into Action

January 30, 2017

A-Lab 2016_Photo 1

By Susan Young


At the same time that businesses are wrestling with the challenges and opportunities resulting from an explosion of big data and analytics, professionals are racing to acquire the skills necessary to thrive in the new, digital world.

MIT is doing its part to prepare students for this digital future.

Professors Erik Brynjolfsson and Sinan Aral launched the Analytics Lab (A-Lab) in 2014. The course, presented by the MIT Initiative on the Digital Economy, is one of the most popular among MIT students pursuing careers in data science, and is a part of MIT Sloan’s suite of Action Learning offerings. The program has worked with nearly 150 students from over a dozen departments within MIT, so far.

“MIT has become the epicenter of the big data revolution that is sweeping through business and the economy,” Brynjolfsson said. Analytics Lab “provides MIT students the opportunity to work with companies leading this transformation.”

During the course – which runs each fall semester – student-teams design and deliver a project using analytics, machine learning, or other digital technologies to solve business problems. Organizations from around the world, including IDE sponsors, provide their data, time, and insights at the start of the semester so that teams can deliver impactful findings in just three months.

Some projects are tightly focused on dilemmas companies currently face. That requires students to quickly understand particular business circumstances and domains before performing their descriptive, predictive, or causal analysis. Other projects are more open-ended, and students must think entrepreneurially about how to bring new value to existing data and suggest frontiers for future business opportunity.

Last fall, 21 real-world projects across a variety of problems were tackled – from determining theater box office performance based on U.S. Census data, to tracking economic activity and home values in a metropolitan area based on local crime reports. Other topics spanned the fields of finance, e-commerce, medical supply chains, workplace safety, and global health.

In December, the teams presented their findings to an audience of experts, entrepreneurs, and executives during an all-day session at the MIT Media Lab.

The “winning team,” as determined by a panel of guest judges – Analog Devices’ Sam Fuller, Dstillery’s Claudia Perlich, and MIT IDE’s Michael Schrage – analyzed mechanic-generated maintenance service sheets to determine the quality of locomotive repairs and to predict equipment failures. As companies move toward condition-based maintenance and repairs for efficiencies and cost savings, rather than traditional time-based service, the ability to anticipate the need of maintenance events is paramount to transportation companies and passengers.

The students used three years’ worth of unstructured data – free text entries with non-standardized spelling and abbreviations, and a diverse and technically specific vocabulary – which required them to employ multiple text-analytics methods. Using a Random Forest model, they were able to better predict equipment failures within the first 78,000 miles of locomotive travel than using regression analysis alone. Looking ahead, the student team suggested that more data inputs, such as sensor data and access to past performance of the maintenance warehouse or individual mechanic, would further improve their model’s predictive power – and the functioning of locomotives across the United States and beyond.

The Fall 2017 program kicks off in May with the release of requests for proposals to interested project sponsors. For more information, please contact Susan Young (