A Bayesian Segmented Linear Regression Model to Observe Breakfast Canyon Patterns in Honeybee Hive Weight Data
Open Access
- Author:
- Yaukey, Francesca
- Graduate Program:
- Statistics
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- November 09, 2023
- Committee Members:
- Ephraim Mont Hanks, Thesis Advisor/Co-Advisor
Stephen Berg, Committee Member
Bing Li, Professor in Charge/Director of Graduate Studies
Stephen Berg, Thesis Advisor/Co-Advisor - Keywords:
- Segmented Linear Regression
Bayesian Statistics
Honeybee Ecology
MCMC Algorithm
Time Series - Abstract:
- Honeybees are essential to ecosystem health and crop vitality, but unfortunately are threatened by large scale changes in land use, increased pesticide use, and industrialized agricultural practices. Better understanding of honeybee colony health and honeybee behavior can lead to better management of bee colonies, enabling bee researchers to help combat colony collapse disorder (CCD). Utilizing hive weight data, we wish to explore the existence of Breakfast Canyons, an ecological phenomena that can give insight into the foraging force of a hive. We create a Bayesian segmented linear regression model which allows for multiple breakpoints to indicate where the slope changes in the daily weight trends. We apply our model to individual continuous hive weight data collected from honeybee colonies in Michigan collected throughout 2018, and explore the existence of the Breakfast Canyon, following a formally established definition of a Breakfast Canyon. Further, we establish a new definition of a Breakfast Canyon based on relative slope patterns, and explore the existence of a Breakfast Canyon with the same data. To further establish the existence of the Breakfast Canyon, we apply our methods for Breakfast Canyon definition utilizing both Meikle's absolute slope definition and our new relative slope definition to simulated Brownian Motion data, and compare the results with the real Honeybee data results.