Event Date:
Monday, February 7, 2022 - 3:30pm to 4:30pm
Event Location:
- Phelps 1160
Title: Life is a Mixture of Opportunities: Mixture Distributions, Inferences, and Applications
Abstract: Undergraduate statistics curriculums often limit their students’ to consider “textbook” datasets, that is, normally distributed unimodal datasets. While such data is suitable for understanding basic statistical concepts, they restrict students’ abilities to think critically about real-world applications. Generally, real-world datasets break the normality assumptions, contain multimodal distributions, and/or involve many parameters. For instance, health care professionals do not treat every patient with the same medical treatment. Thus, we need our students to learn statistical models that reflect the nature of our healthcare system, which breaks our population into various sub-populations.
Furthermore, non- “textbook” datasets contain anomalous values for various reasons such as imperfect data collection, faulty equipment, or mis-recorded data. Our students need to learn techniques and processes to assess the validity of abnormal values and statistical methods that alleviate the impact of these anomalies. Mixture models enable us to (1) characterize the sub-populations within our entire population and (2) provide model flexibility to alleviate the normality assumptions. This presentation introduces mixture models by demonstrating the data generation process, discussing inferential techniques, and illustrating the flexibility of mixture models with an application in the Virginia Tech Stability Wind.
January 25, 2022 - 8:35am