Wei Zhang

University of Georgia

Optimal Designs for a Panel Mixed Logit Model

We discuss optimal designs for the panel mixed logit model. The panel mixed logit model is usually used for the analysis of discrete choice experiments. The information matrix used in design criteria does not have a closed form expression and it is computationally difficult to evaluate the information matrix numerically. We derive the information matrix and use the obtained form to propose three methods to approximate the information matrix.

Major Professor(s): 
Abhyuday Mandal, John Stufken
Friday, February 9, 2018 - 3:30pm
Brooks Hall Room 434

Ting Zhang

Boston University

Unsupervised Self-Normalized Change-Point Testing for Time

We propose a new self-normalized method for testing change points in the time series setting. Self-normalization has been celebrated for its ability to avoid direct estimation of the nuisance asymptotic variance and its flexibility of being generalized to handle quantities other than the mean. However, it was developed and mainly studied for constructing confidence intervals for quantities associated with a stationary time series, and its adaptation to change-point testing can be nontrivial as direct implementation can lead to tests with nonmonotonic power.

Thursday, March 8, 2018 - 3:30pm
Room 102, Caldwell Building

Hendrik Hamann


Big Geospatial-temporal Data and Analytics as-a-Service

The rapid growth of geospatial-temporal data sources from satellites, drones, weather modeling, IoT sensors etc., accumulating at a pace of petabytes to exabytes annually, opens unprecedented opportunities for industrial applications. However, the sheer size and complexity of the data also raises considerable challenges for conventional tools, relational geospatial databases, and cloud geospatial data services based on file systems (manifested as object stores or “cold” tape storages).

Thursday, March 22, 2018 - 3:30pm
Room 102, Caldwell Building

Yong Zeng

National Science Foundation, University of Missouri – Kansas City

Bayesian Inference via Filtering Equations for Financial Ultra-High Frequency Data

We propose a general partially-observed framework of Markov processes with marked point process observations for ultra-high frequency (UHF) transaction price data, allowing other observable economic or market factors. We develop the corresponding Bayesian inference via filtering equations to quantify parameter and model uncertainty.  Specifically, we derive filtering equations to characterize the evolution of the statistical foundation such as likelihoods, posteriors, Bayes factors and posterior model probabilities.

Thursday, October 12, 2017 - 3:30pm
Room 306, Statistics Building 1130

Tharuvai Sriram

University of Georgia

Leverage-based Sequential Sampling Method for Streaming Time Series Data

We consider a streaming time series data, which is assumed to come from a non-explosive p-th order autoregressive (AR(p)) model with p ≥ 1. Our goal is to estimate the parameters of this model using a subsample of random size drawn sequentially from the streaming data based on a stopping rule. Traditionally, sequential sampling is carried out after observing an initial sample of fixed size. However, our sampling starting point is chosen according to statistical leverage scores of the data and the subsample size is decided by a sequential sampling rule.

Thursday, September 21, 2017 - 3:30pm
Statistics Building Room 306


Subscribe to RSS - Colloquium