Stanford Professor Combines Machine Learning and Econometrics

As part of the Economics Department’s Seymour E. and Ruth B. Harris Lecture Series, Stanford economics professor Susan Athey lectured on new methods of including machine learning in econometrics late Tuesday afternoon.

Athey’s lecture focused on the wide variety of applications that machine learning—the use of algorithms to analyze and draw predictions from data—could have on everything from online analytics with search engines and auctions to isolating causal effects in medical trials.

Athey is the economics of technology professor at Stanford’s Graduate School of Business, where she currently researches the design of auction-based marketplaces and the economics of the internet. She was a professor of Economics at Harvard between 2006 and 2012.

In recent years, Athey has furthered her research by doing analytics and experiments with technology firms. She described her experience working with Microsoft as “like getting another PhD.”

During her time with various tech firms in applying machine learning to analyzing search engines and online auctions, she said she found a significant lack of fellow economists working in her research field.

However, Athey explained that she believes that economics and econometrics combined with machine learning could improve data analysis for these tech firms.

“Data is the domain of machine-learning people,” Athey said. Data that is analyzed without careful consideration of the methods used to calculate error could generate less accurate results, she added.

“It is really important to not just throw in this machine-learning method off the shelf,” Athey warned.

The Seymour E. and Ruth B. Harris Lecture Series is managed by the Economics Department and invites distinguished academics to lecture on topics that would interest professional economists, according to former department chair John Y. Campbell, who helps organize the series.