Samuel Levy
Assistant Professor of Business Administration
Marketing Area
University of Virginia Darden School of Business
​100 Darden Boulevard
Charlottesville, Virginia 22903 USA
Phone: 412-773-2627
Email: levys@darden.virginia.edu
Bio
I am an Assistant Professor of Business Administration in the Marketing Area at the University of Virginia Darden School of Business. My research investigates customer analytics in different contexts such as privacy, data fusion, pro-social behavior. I am also interested in economic models of choice with novel Bayesian statistical and machine learning methods in marketing. I received my PhD from Carnegie Mellon University.
My preferred methodological approaches are Bayesian econometrics, probabilistic machine learning, and Bayesian nonparametrics.
Interested in my research or in collaborating? Feel free to email me at levys@darden.virginia.edu. You can also find me on LinkedIn.
For more about my background, check out my CV.
Research
Digital Marketing Twins
with Longxiu Tian
A novel deep generative framework that extracts latent features from individual customer brand survey responses from competing firms, to optimize individualized marketing policies and flexibly uncover key drivers of brand affinity.
We propose a privacy preserving data fusion methodology - a way to combine multiple datasets - intended to preserve user anonymity while allowing for a robust and expressive data fusion process.
Relaxing Functional Form in Choice Models Through Gaussian Processes
with Alan Montgomery
We build a general utility model to detect and "learn" non-linear patterns of consumption from the data.
Understanding the Dynamics of Appeals Scales to Infer Potential to
Donate using Bayesian Nonparametrics
with Joy Lu and Alan Montgomery
​
We study the dynamic effect of suggested amounts, i.e. "appeals scales", on donation behavior combining a flexible nonparametric model and field experiments.
Multiview Topic Model For Purchase Prediction
with Dokyun Lee, Daniel McCarthy, and Alan Montgomery
​
A Bayesian framework that integrates transaction and clickstream data, providing accurate predictions of purchase timing for industrial clients in a non-contractual business-to-business setting, and reveals diverse customer and product clusters.