News & Notes

Savvysherpa is hiring! Savvysherpa, Inc. is looking for Data Scientists!. Our Research Scientists participate in every stage of the research process, from data gathering to modeling and analysis to building data products. For more information see the company website.


SMMR Paper: A recent collaboration with the Mayo Clinic Department of Colorectal Surgery resulted in a paper illustrating the use of electronic health record data for prediction of post-operative surgical complications, which will appear in Statistical Methods in Medical Research.


Annual Reviews Paper: Bill Eddy and I wrote a paper for Annual Reviews that discusses multivariate ordering methodologies; in particular, we like the original idea of John Tukey. Our review is here (contact me directly for a pdf copy).


EPA FIFRA SAP Meeting: I participated in a scientific meeting of the EPA Scientific Advisory Panel as an ad hoc panel member reviewing EPA's approaches for prioritization of chemicals as possible endocrine disruptors. The meeting agenda and minutes are available here.

Improving the characterization of the distribution of extreme values is of paramount importance.
-- Alan Greenspan, former Fed chair

an imageMy primary personal research interests lie in the field of extreme value theory (EVT), the branch of probability and statistics which aims to describe and model tail events and their associated risks. While the probability theory underlying the study of extremes has been well-developed for decades, the suite of statistical methodology for the modeling of extreme values is currently growing rapidly. I have primarily worked on the modeling of multivariate extremes; to give an example, one may wish to estimate the probability of multiple securities in a financial portfolio experiencing tail loss events (of magnitude previously unseen) simultaneously. EVT provides tools to perform such an estimation given observed data. My entry into the field was through the study of climate extremes, and I published research in both the statistical and climate science literature on this topic. I have also worked on extreme value problems in financial risk management. an image

My work in applications of statistical methodology spans a wide range of fields, including climate science, astrophysics, finance, and environmental monitoring. In my current position, I work on applied problems in healthcare, engineering, and technology, using modern statistical and machine learning methods for dimension reduction, time series forecasting, and network analysis. I am also interested in the development of novel methodology for analysis of multivariate extreme events; as part of my dissertation, I proposed an estimation scheme for a phenomenon known as hidden regular variation.

Below is a partial list of publications and presentations. For a full list and more detail, please see my current CV.

Publications and Manuscripts

  • Weller G.B., Larson, D.W., Lovely, J., Earnshaw, B.A., and Huebner, M. (2017). Leveraging electronic health records for predictive modeling of post-surgical complications. Statistical Methods in Medical Research, in press.
  • Cisewski J., Weller G.B., Schafer, C., and Hogg, D.W. (2016+). Approximate Bayesian computation for the stellar initial mass function. In revision.
  • Tang M. and Weller G.B. (2016). Bivariate tail risk analysis for high-frequency returns via extreme value theory. Model Assisted Statistics and Applications, in press.  Supplementary file.
  • Weller G.B. and Eddy W.F. (2015). Multivariate order statistics: theory and application. Annual Reviews of Statistics and Its Application 2: 237-257.  link
  • Weller G.B. and Cooley D. (2014). A Sum characterization of hidden regular variation with likelihood inference via expectation-maximization. Biometrika 101(1): 17-36.    supplement (pdf)
  • Weller G.B., Cooley D., Sain S., Bukovsky M., Mearns L. (2013). Two case studies on NARCCAP precipitation extremes. Journal of Geophysical Research - Atmospheres, 118(18): 10,475-10,489.
  • Weller G.B. (2013). Joint Tail Modeling via Regular Variation with Applications in Climate and Environmental Studies. PhD Dissertation, Colorado State University, Fort Collins, CO. Dissertation (pdf)
  • Weller G.B. and Cooley D. (2012). An Alternative Characterization of Hidden Regular Variation in Joint Tail Modeling. CSU Department of Statistics Technical Report. PDF (24 Aug. 2012)    previous version (12 June 2012)
  • Weller G.B., Cooley D., and Sain S. (2012). An investigation of the Pineapple Express phenomenon via bivariate extreme value theory. Environmetrics 23(5): 420-439. Supplementary figures (pdf)

    Work in preparation

  • Huebner, M., Larson, D.W., Lovely, J., and Weller, G.B. Co-occurrence of complications after colorectal surgery.
  • Weller, G.B. Bayesian forecasting of pharmaceutical utilization with dynamic linear models.


  • Quantifying Tail Risk in Health Insurance Pools using Extreme Value Theory.
         Joint Statistical Meetings, Chicago, IL. August 1, 2016.
  • Leveraging Electronic Health Records for Predictive Modeling of Surgical Complications.
         International Society for Clinical Biostatistics 2015 Conference, Utrecht, Netherlands. August 26, 2015. slides (pdf)
  • Bayesian Forecasting of Generic Pharmaceutical Utilization via Dynamic Linear Models.
         Joint Statistical Meetings, Seattle, WA. August 13, 2015.
  • Modeling Multivariate Extreme Values via Regular Variation: an Application to High Frequency Financial Returns
         Department of Statistics Environmental Seminar, North Carolina State University (Invited), Raleigh, NC. March 3, 2015. slides (pdf)
  • Inference for Hidden Regular Variation in Multivariate Extremes
         Department of Statistics Seminar, Columbia University (Invited), New York, NY. April 14, 2014. slides (pdf)
         SIAM 2014 Conference on Uncertainty Quantification, Savannah, GA. March 31, 2014.
         Department of Statistics Seminar, University of Wisconsin-Madison, Madison, WI. September 23, 2013.
  • A Sum Characterization of Hidden Regular Variation in Joint Tail Modeling with Likelihood Inference via the MCEM Algorithm
         Joint Mathematics Meetings, Baltimore, MD. January 17, 2014.
         Eighth International Conference on Extreme Value Analysis (Invited), Shanghai, China. July 11, 2013.
         Department of Statistics Seminar, Carnegie Mellon University, Pittsburgh, PA. January 28, 2013.
         Department of Statistics Colloquium, University of Missouri, Columbia, MO. January 24, 2013.
  • Bivariate Extreme Value Analyses of NARCCAP Precipitation Extremes
         American Geophysical Union 2013 Fall Meeting, San Francisco, CA, December 12, 2013.
         Department of Meteorology Colloquium, Pennsylvania State University, State College, PA, October 2, 2013.
  • Tail Dependence and Its Applications
         Applied Probability and Statistics Seminar, University of St. Thomas, St. Paul, MN, December 6, 2013.
  • Flooding to Financial Disaster: An Introduction to Extreme Value Theory
         Concordia College Mae Anderson Alumni Lectures (Invited), Moorhead, MN. January 20, 2012. slides