Ohio University Lectures in the History and Philosophy of Science series presents Clark Glymour on “The Causal Revolution, Observational Science and Big Data” Friday, March 20, at 4 p.m. in Scripps Hall, 111.
The lecture series is presented by the Philosophy Department and the Spetnagel Fund.
Clark Glymour
Glymour is Alumni University Professor in the Department of Philosophy at Carnegie Mellon University.
Abstract: In recent decades, data for very large numbers of variables have been collected in genomics, brain imagining, climate science and other domains. These data are almost all “observational” in the sense that few or none of the variables have been randomized or otherwise experimentally controlled. In the same period, a variety of computerized methods for discovering causal relations from observational data have been developed. The use of observational data for causal inference is routinely challenged, and yet there seems no alternative if these very large data collections are to be of use for science, medicine and policy. Two problems are foremost: can computerized search for causal relations be “scaled up” to analyze data sets with hundreds of thousands of variables, and how accurate are the searches? This lecture describes the development of some of the central ideas in computerized causal search, their relations with conventional statistical inference, their applications in “small” and “large” problems, recent results on scaling up to 1 million variable problems, and the difficulties in assessing the accuracies of the procedures.
Comments