Fall 2021


Group meetings are for students and faculty at CMU and UPitt.
The STAMPS webinars are open to everyone.

These meetings are Fridays 1:30-2:30 PM. Our current plan is to hold these meetings in person in 232M (second floor of the Stat/DS Department in Baker Hall). If that plan is to change, we will send out an alert via the mailing list.

Public Webinars will also be at 1:30-2:30 PM.

September 3, 1:30PM EDT - Welcome Meeting #


September 10, 1:30PM EST - Public Webinar : Doug Nychka (Department of Applied Mathematics and Statistics, Colorado School of Mines)


September 17th, 1:30-2:30PM EDT - Coty Jen (CMU) #

Note: Advanced Data Analysis (ADA) related talk


September 24th, 1:30-2:30PM EDT - Hamish Gordon (CMU) #

Note: Advanced Data Analysis (ADA) related talk


October 8, 1:30PM EDT - Public Webinar : Yang Chen (Department of Statistics, University of Michigan)


October 15th, 1:30-2:30PM EDT - Mikael Kuusela (CMU) #

Title: Spatio-Temporal Analysis of Argo Profiling Float Data for Improved Understanding of Ocean Climate

Description: Argo floats measure seawater temperature and salinity in the upper 2,000 m of the global ocean. These data contain crucial information about ocean climate and dynamics but their statistical analysis is challenging due to the large size and complex structure of the data set. In this presentation, I will give an overview of our work on spatio-temporal statistics for Argo data, with a particular focus on the global ocean heat content in order to motivate a potential ADA project on this topic.

October 22nd, 1:30-2:30PM EDT - David Rounce (Civil and Environmental Engineering, CMU) #

Title: Global glacier evolution modeling: aggregating uncertainties across space and time

October 29th, 1:30-2:30PM EDT - Trey McNeely and Galen Vincent (CMU) #

Title: Detecting Distributional Differences in Labeled Sequence Data with Application to Convection in Tropical Cyclones

Description: Tropical cyclone (TC) intensity forecasts are issued by human forecasters who evaluate spatio-temporal observations (e.g., satellite imagery) and model output (e.g., numerical weather prediction, statistical models) to produce forecasts every 6 hours. Within these time constraints, it can be challenging to draw insight from such data. While high-capacity machine learning methods are well suited for prediction problems with complex sequence data, extracting interpretable scientific information with such methods is difficult. In order to detect distributional differences in these high-dimensional sequences, we convert a traditional two-sample testing problem into a prediction problem. In this framework, we leverage high-dimensional prediction algorithms to perform classical statistical inference and identify patterns in the evolution of TC convective structure leading up to rapid intensity changes, hence providing forecasters and scientists with key insight into TC behavior.

November 12, 1:30PM EST - Public Webinar : Glen Cowan (Department of Physics, Royal Holloway, University of London)


November 19, 1:30-2:30PM EST - Patrick Bryant #

Title: Quantifying Uncertainty in Data Driven Background Distributions

Description: In the absence of accurate background simulation, how can we quantify and assign uncertainties on data driven background model distributions? Such models often require extrapolation across high dimensional phase spaces from signal depleted to signal enriched regions with non-linear density estimation. I will present a tentative procedure to assign systematics in such a scenario via repeated estimates of synthetic data derived from real data. This allows direct estimation of the model variance as well as non-closure (bias) relative to the target synthetic data.

December 3, 1:30PM EST - Public Webinar : Elizabeth Barnes (Department of Atmospheric Science, Colorado State University)