Projects
Survey Designs for Distance Sampling: A Study of Zebra Mussels
First Place Winner of American Statistical Association and CAUSE Undergraduate Statistics Research Project competition
Awarded Distinction from Carleton College Statistics Department
The zebra mussel (Dreissena polymorpha), an invasive species from Europe, has spread through-out the Great Lakes and down the Mississippi River, infesting hundreds of lakes across North America, including Minnesota. Our group worked with researchers at the University of Minnesota to help develop the best lake sampling methods to accurately determine the abundance of zebra mussels in infested lakes. Using lake survey data from the University, we implemented the principles of line transect distance sampling to a model that describes the probability of detection for each zebra mussel in a given lake. Then, using the Horvitz-Thompson estimator, we made inferences about the population abundance of zebra mussels in the infested lake. Once we had a better understanding of estimating abundance in a single lake, we explored which survey design would most accurately predict the population abundance and reduce error. Because we only had data from one survey design, we took advantage of a package in R called DSsim, which allowed us to run simulations on generated populations. We constructed different survey designs under various conditions (including population size, number of transects, and hotspot presence) to investigate if and how the results would change. Finally, we conducted an experiment to analyze the relationship between time on a transect and detection. By evaluating the error, bias, and predictive statistics provided by the simulation and model fi tting, we are able to provide ecologists with a recommendation for the optimal zebra mussel sampling design.
Value of Emotion: Effects of Visceral Factors on Economic Behavior
The phenomenon of auction fever occurs when participants overbid for a certain good. Current theory suggests that there are two primary mechanisms that cause this: social influence and object driven effects. The environment created by those within an auction and the desire to own the good drives the bidding past the appraised value. These value-changing mechanisms have the potential to be further affected by emotional stimuli or visceral factors impacting the individuals participating. As it stands, there is limited research regarding the influence of these affective states and how feelings like sadness or happiness alters valuation and behavior directly. This study finds that an emotional stimulus may not cause a change in behavior on its own. Rather, these visceral states can only induce a change in behavior when working within a setting that already contains value-influencing mechanisms, like those present within an auction. As a result, I propose a change to current utility theory that accounts for the influence of emotion and other visceral states on behavior, and how it works in conjunction with market conditions. From here, further understanding of individual market behavior can be gained.
shiny & ggplot
Demonstration of various dynamic shiny plots and a random forest classification model. Project was completed in 2018 and demonstrates first project using such packages.