1. K-State home
  2. »Arts and Sciences
  3. »Statistics
  4. »Academics
  5. »Undergraduate Research

Department of Statistics

Undergraduate Research

As K-State strives to attain the status of a top 50 public research university by the year 2025, both the University and the College of Arts & Sciences offer a variety of opportunities for undergraduate research.  

In addition to the experience, many of these programs also provide monetary awards to the recipients.  Two of our statistics undergraduates have received undergraduate research awards in the past two years.  Their accomplishments are highlighted below. 

Audrey Chang, 2016-2017

Audrey Chang has been awarded an Arts & Sciences Undergraduate Research Scholarship for 2016-1017.  She is working with Dr. Wei-wen Hsu to develop a zero-inflated Poisson model to analyze tornado data in the state of Kansas. 

Abstract

The purpose of this project is to use a zero-inflated Poisson model coupled with working independence assumptions to analyze the longitudinal tornado data in the state of Kansas from 1950 to 2015. The zero-inflated models are used often to accommodate excess zero counts in data. In the Kansas tornado data, the number of counties with no tornado touch downs during a certain period of time was often observed, therefore creating many zero counts. We started this project in August 2016, with Dr. Wei-Wen Hsu of the Statistics Department as my mentor. Through the proposed model, we will identify factors (e.g., location of county, season, temperature, etc.) that can be used to predict the frequency of tornado touch-downs in the future. We expect that the model can provide a new perspective about tornado monitoring and gain new understanding to Kansas’ tornado patterns.

Research Plan

This first semester, we have set aside two months reading and understanding of the data and related literature about the tornadoes and zero-inflated Poisson models, respectively. We are currently finishing up the data preparation (i.e., data merging and data cleaning) using statistical software SAS and will focus on the model development and data visualization for Kansas tornado data in the following few months. Starting February 2017, we will write a scientific paper and hope to publish our findings in a scientific journal afterwards as well as present our findings at a national or international conference.

Graham Seacat, 2016-2017

As a recipient of an Arts & Sciences Undergraduate Research Scholarship for 2016-1017, Graham is working with Dr. Trevor Hefley to develop a behavior model for mule deer.

Abstract

The purpose of this study is to implement statistical analysis in the field of ecology to learn more about the behavior of mule deer (Odocoileus hemionus). Animal behavior is the link between the biology of the animal and the ecosystem in which it lives. I plan to apply our movement modeling approach to a telemetry dataset, collected from GPS devices that measure location and heart rate, on mule deer from the Piceance basin in Colorado, USA. The data points typically are separated into two “behavioral states”, moving and sleeping. Unlike many telemetry data sets, the proposed dataset is ideal to test our methods because field based observations that provide visual proof of the animals’ behavior are also available.

Audrey Chang, 2016

Audrey Chang received an award for Undergraduate Research in the College of Arts & Sciences for Fall 2016.  She worked with Dr. Wei-Wen Hsu to develop a zero-inflated Poisson model for Kansas tornadoes.  

Abstract

The purpose of this project is to use a zero-inflated Poisson model coupled with working independence assumptions to analyze the longitudinal tornado data in the state of Kansas from 1950 to 2014. The zero-inflated models are used often to accommodate excess zero counts in data. In the Kansas tornado data, the number of counties with no tornado touch downs during a certain period of time was often observed, therefore creating many zero counts. Through the proposed model, we will identify factors (e.g., location of county, season, etc.) that can be used to predict the frequency of tornado touch downs in the future. We expect that the model can provide a new perspective about tornado monitoring and gain new understanding to Kansas’ tornado patterns.

Research Plan

The first two months will be set aside for reading and understanding of the data and related literature about the tornadoes and zero-inflated Poisson, respectively. The next three months we will conduct real data analysis using R and/or SAS as well as some data managements (i.e., data merging and data cleaning). The rest of the year will be used to write a scientific paper of the findings. We hope to publish our findings in a scientific paper afterwards.