Description:

PARTICIPATION MAY REQUIRE REGISTERING WITH THE PSMG LISTSERV PRIOR TO THE SEMINARS IN ORDER TO RECEIVE PRESENATION MATERIALS.

TIME AND CALL-IN INFORMATION IS BELOW:

TIME: 10-11:30am PST / 11am-12:30pm MST / 12-1:30pm CST / 1-2:30pm EST
CALL: (855)747-8824 - Participant Code: 311-035-8507

For logistical or administrative concerns, please contact Katia Chernyshov at katia.chernyshov@northwestern.edu or 312-503-4334.

9/15/2015

Michelle C. Dunn
Senior Advisor for Data Science Training, Diversity, and Outreach Data Science @ NIH
National Institutes of Health

“Big Data Initiatives at NIH” (BD2K)

Despite the hope that data science promises, we are faced with a dearth of skilled researchers who can mine and extract meaningful information from the vast amount of heterogeneous biomedical data. The field needs creative statisticians who are facile with computational ideas and paradigms, to apply their skill set to the challenges of biomedical Big Data science. Although statistical challenges are plentiful in all areas of biomedical science, we highlight the particular challenges addressed by two complementary and large initiatives from the National Institutes of Health (NIH)—Big Data to Knowledge (BD2K) and the Brain Research through Advancing Innovative Neurotechnologies (BRAIN). I will talk about the opportunities available for statisticians to accelerate advances in these initiatives.
http://chance.amstat.org/2015/04/calling-all-statisticians/

9/22/2015

Martin Lindquist
Professor
Department of Biostatistics
Johns Hopkins University

“Analysis of Neuroimaging Data”

Understanding the brain is arguably among the most complex, important and challenging issues in science today. Neuroimaging is an umbrella term for an ever-increasing number of minimally invasive techniques designed to study the brain. These include a variety of rapidly evolving technologies for measuring brain properties, such as structure, function and disease pathophysiology. These technologies are currently being applied in a vast collection of medical and scientific areas of inquiry. This talk briefly discusses some of the critical issues involved in neuroimaging data analysis (NDA). This includes problems related to image reconstruction, registration, segmentation, and shape analysis. In addition, we will discuss various statistical models for conducting group analysis, connectivity analysis, brain decoding, and the analysis of multi-modal data. We conclude with discussing some open research problems in NDA.

9/29/2015

Jennifer Hill
Professor of Social Sciences
Deaprtment of Humanities & Social Sciences
New York University

“Causal Inference and Data Science”

Advances in the ability to flexibly model response surfaces have led to exploration of robust approaches to causal inference that do not require matching, weighting or subclassification on the propensity score. Tradeoffs exist between these methods however with regard to their performance and flexibility, particularly when the data are high-dimensional. This work examines the strengths and weaknesses of competing approaches to nonparametric modeling of the causal response surface including Bayesian Additive Regression Trees and Gaussian Processes. We also explore the ability of such methods to exploit information in the propensity score to achieve double robust (or approximate double robust) properties. The efficacy of these approaches to causal inference will be compared with respect to bias, statistical efficiency, computational efficiency, and the ability to identify neighborhoods of common causal support.

10/6/2015

Frauke Kreuter
Associate Professor
Joint Program in Survey Methodology
College of Behavioral and Social Sciences
University of Maryland

“Machine Learning: Introduction”

Abstract not available yet

10/13/2015

Trevor Hastie
Professor
Department of Statistics
Stanford University

“Sparse Linear Models”

Focusing on R package glmnet, I will talk about efficient procedures for fitting the entire lasso or elastic-net regularization path for linear regression, logistic and multinomial regression models, Poisson regression and the Cox model. The algorithm uses cyclical coordinate descent in a pathwise fashion. I will also touch on two recent additions, multiple-response Gaussian and grouped multinomial.

10/20/2015

Kosuke Imai
Professor
Department of Politics
Princeton University

“Causal Interaction in High Dimension”

Estimating causal interaction effects is essential for the exploration of heterogeneous treatment effects. In the presence of multiple treatment variables with each having several levels, researchers are often interested in identifying the combinations of treatments that induce large additional causal effects beyond the sum of separate effects attributable to each treatment. We show, however, the standard definition of causal interaction effect, typically estimated with the standard linear regression or ANOVA, suffers from the lack of invariance to the choice of baseline condition and the difficulty of interpretation beyond two-way interaction. We propose an alternative definition of causal interaction effect, called the marginal treatment interaction effect, whose relative magnitude does not depend on the choice of baseline condition while maintaining an intuitive interpretation even for higher-order interaction. The proposed approach enables researchers to effectively summarize the structure of causal interaction in high-dimension by decomposing the total effect of any treatment combination into the marginal effects and the interaction effects. We also establish the identification condition and develop an estimation strategy for the proposed marginal treatment interaction effects. Our motivating example is conjoint analysis where the existing literature largely assumes the absence of causal interaction. Given a large number of interaction effects, we apply a variable selection method to identify significant causal interaction. Our exploratory analysis of a survey experiment on immigration preferences reveals substantive insights the standard conjoint analysis fails to discover. (Last Revised May, 2015). http://imai.princeton.edu/research/int.html

10/27/2015

Mitsu Ogihara
Professor
Department of Computer Science
University of Miami

“Machine Learning: Overview”

Abstract not available yet

11/03/2015

Michael Sobel
Professor
Department of Statistics
Columbia University

“Causal Inference with fMRI Data”

Neuroscientists have identified a network (PPN) of brain regions involved in pain processing. An important question is if this is modulated by subjects' expectations, and if so, how this works. We analyze data from a study of 19 subjects observed on multiple trials. In each trial, a subject is told he/she will receive a high (H) or (L) low level of a pain inducing thermal stimulus. The H (L) cue is then followed by either an H or medium (M) (L or M) stimulus. The stimulus was then applied, after which the subject reported their level of pain. During the trial, the blood oxygenation level dependent (BOLD) responses from approximately 100,000 voxels are observed. We analyze these responses and compare the latent amplitudes under the H cue, M stimulus condition and the L cue, M stimulus condition, concluding that the cue modulates the response to the stimulus. We then construct a causal model for how the cue mediates the relationship between the amplitudes and the thermal stimulus.

11/10/2015

Robert Gibbons
Professor
Departments of Medicine & Public Health Sciences (Biostatistics)
University of Chicago

“Big Data for Small Minds”

With the promise of big data comes the responsibility of analyzing them wisely. Statisticians have struggled with the analysis of observational data for decades and too often this work has been ignored by data scientists. I present a series of examples of analytic work involving big data to illustrate good and bad analytic approaches.