Seminar

Yes, you really can use this! - Applying Data Science to Real-World Problems

Thursday, March 2, 2017 - 5:00pm to 6:00pm
Schapiro Hall (CEPSR) Davis Auditorium
530 W 120th St.
New York, NY 10027
United States

Columbia Data Science Innovation Seminars

Ben Arancibia, Lead Data Scientist | Booz Allen Hamilton
Dan Liebermann, Lead Associate | Booz Allen Hamilton

Booz Allen Hamilton’s Dan Liebermann and Ben Arancibia will cover what it takes to get data science done in the real world. They will be sharing stories from the trenches – covering experiences and lessons learned from turning data science theory into reality when the problem (and the solution) are far from known. The talk will heavily engage the audience to hear their perspective, and cover the approach Booz Allen took to solve its clients’ problems. The goal is to get the audience thinking about what they would do in these situations and how they would apply their classroom experience.

Understanding What Sticks in the U.S. Presidential Election Race

Tuesday, November 1, 2016 - 1:00pm to 2:00pm
United States

Columbia Data Science Innovation Seminars

The US Presidential election is probably the most significant political event in 2016. The spectacular campaign period has featured many controversial news stories. Many analysts have given their views on how the public image of Clinton and Trump affect the candidates’ success; what kinds of stories stick with a candidate, who earns on a certain story, and what do the candidates’ constantly shifting public images mean for the outcome of the election?

Describing the landscape of political positions and measuring the effects of speeches and events is very difficult to do. This session will feature research from United Minds, a Weber Shandwick company, supported by text analytics technology firm Gavagai.

Estimating Causal Effect of Ads in a Real-Time Bidding Platform

Tuesday, April 26, 2016 - 5:00pm to 6:00pm
Davis Auditorium, 412 CEPSR, Schapiro Center
530 West 120th Street
New York, NY 10027
United States

Columbia Data Science Innovation Seminars

Prasad Chalasani
SVP, Data Science
MediaMath

A real-time bidding platform responds to incoming ad-opportunities ("bid requests") by deciding whether or not to submit a bid and how much to bid. If the submitted bid wins, the user is shown an ad. Advertisers hope that ad-exposure leads to an increased likelihood of a desired action, such as a click or conversion (purchase, etc). So an important quantity that advertisers want to measure is the causal effect of advertising, namely, what is the response probability of an exposed user, compared with the counterfactual (un-observable) response-rate of the user if they were not exposed to the ad. In an ideal randomized test, the user is randomly assigned to test or control AFTER the submitted bid is won, and test users are served the ad in the normal way, while control users are not. While this is ideal from a statistical perspective, in practice this approach has the drawback that money spent by advertisers is wasted when a user is assigned to control. At Media Math we have developed a methodology for causal effect measurement where users are assigned to test or control BEFORE bid submission. One challenge here is that not all test-group users are exposed to an ad; only a winning bid results in ad exposure, and the winning population can have a significant bias. This talk will describe our approach to handle this and other challenges to ad impact measurement in this setting, and how we use MCMC Gibbs sampling to arrive at confidence intervals for ad-impact.

Probabilistic modeling of big tables and sequences

Friday, October 23, 2015 - 9:00am to 10:00am
United States

Abstract: High-dimensional discrete data are collected in many application areas, but have seen limited consideration in the literature. We focus in particular on two related problems: (i) high-dimensional categorical data that can be orga-nized as a many way contingency table; and (ii) sequential categorical data having a complex dependence structure. The first problem arises in numerous applications ranging from survey research in social sciences and epidemiology to genomics and marketing.

Pages

Subscribe to RSS - Seminar
Back to Top