Birds of a Feather Shop Together: Predicting Adoption with Social Networks - Sharad Goel - Yahoo! Research, USA
Abstract: Adoption is often predicted using individual-level attributes such as age, sex, and geographic location. The principle of homophily suggests that social data (e.g., the attributes of people with whom one is in contact) might also have predictive value, however a lack of such social network data has limited research into this question. To assess whether social data can improve predictive models in a variety of domains, we construct a network from email and instant message exchanges and apply it towards making individual-level predictions of retail spending, joining a recreational league, and reacting to online advertisements. In each instance, we find that network data help identify those individuals most likely to adopt, and moreover, that these data often improve upon traditional indicators. In predicting repeat adoption, we find social data generally diminish in utility, but can remain informative even in the presence of strong predictors. This is joint work with Daniel Goldstein.
Bio: Sharad Goel is a Senior Research Scientist in the Microeconomics and Social Systems group at Yahoo! Research. He is interested in empirical and theoretical problems at the intersection of computer science and the social sciences, particularly questions motivated by sociology and economics.