Next Meetup

Evaluating stochastic seeding strategies in networks
Our October meeting will feature a talk from Dean Eckles, assistant professor at MIT, about inference methods inside networks. Join us to learn more about this interesting topic and share your story with fellow bayesians. Title: Counterfactual policy evaluation with combinatorial treatments and stochastic policies (or, Evaluating stochastic seeding strategies in networks) Abstract: We often want to evaluate or optimize a policy that assigns units to different treatments, including assignments based on their observable characteristics and even their own past response to treatment. These problems have been addressed in the dynamic treatment regime and contextual bandit literatures, though many relevant ideas are familiar from causal inference more generally or Monte Carlo methods, such as importance sampling. In this talk, I will consider applications where we are interested in strategies for selecting "seeds" in a network, where these seeds are assigned to some costly treatment with the aim of causing information or behavior to spread (or, conversely, we might vaccinate these seeds). In particular, we focus on strategies that are stochastic because they make use of only local information. One such stochastic seeding strategy is to select random network neighbors of random individuals, thus exploiting a version of the friendship paradox, whereby the friend of a random individual is expected to have more friends than a random individual. Here we show both how stochastic seeding strategies can be evaluated using existing data arising from randomized experiments in networks designed for other purposes and how to design much more efficient experiments for this specific evaluation. In particular, we consider contrasts between two common stochastic seeding strategies and analyze nonparametric estimators adapted from policy evaluation or importance sampling. Using simulations on real networks, we show that the proposed estimators and designs can dramatically increase precision while yielding valid inference. We apply our proposed estimators to a field experiment that randomly assigned households to an intensive marketing intervention and a field experiment that randomly assigned students to an anti-bullying intervention. Joint work with Alex Chin & Johan Ugander. Speaker Bio: Dean Eckles is a social scientist and statistician. Dean is the KDD Career Development Professor in Communications and Technology at Massachusetts Institute of Technology (MIT), an assistant professor in the MIT Sloan School of Management, and affiliated faculty at the MIT Institute for Data, Systems & Society. He was previously a member of the Core Data Science team at Facebook. Much of his research examines how interactive technologies affect human behavior by mediating, amplifying, and directing social influence — and statistical methods to study these processes. Dean’s empirical work uses large field experiments and observational studies. His research appears in the Proceedings of the National Academy of Sciences and other peer-reviewed journals and proceedings in statistics, computer science, and marketing. Dean holds degrees from Stanford University in philosophy (BA), symbolic systems (BS, MS), statistics (MS), and communication (PhD). Agenda: 6:30pm: Meet and greet. Networking 7:00pm: Talk by Dean Eckles + Q&A 8:00pm: Networking 8:45pm: End of the event Sponsors: This event is sponsored by QuantumBlack, a McKinsey Company.

280 Congress St

280 Congress Street · Boston

    Past Meetups (26)

    What we're about

    Welcome to the Boston Bayesians Meetup group.

    We are a small group of data scientists, statisticians, engineers and researchers interested in the practical applications of Bayesian statistics. Join us for friendly academic briefings, stories from real-world projects, and open discussion of Bayesian inference, tools, techniques and case studies.

    We follow a conventional format of 1 or 2 presentations from volunteers in the group and/or invited experts, and general conversation and socialising afterwards. Think of us as your Bayesian self-help group in the Boston Area.

    Some topics of interest are:

    - Bayesian Data Analysis.
    - Probabilistic programming with PyMC3, STAN, others.
    - Model checking and comparison techniques.
    - Computational methods for approximate Inference.
    - Real-world examples of hierarchical models.
    - Bayesian methods for Machine Learning and deep learning.
    - Bayesian methods for Reinforcement Learning.
    - Non-linear and non-parametric models.

    Call for presentations:

    We provide a supporting environment to share your ideas and get feedback on your work. If you are interested in presenting, please submit presentation proposals at

    Follow us on Twitter: @BostonBayesians

    Members (1,575)

    Photos (49)

    Find us also at