Frequentist, Bayesian, and Zero-One Inflated Beta Regression Models
Access changed 3/2/2017.
The main objective of this paper is to introduce readers to the beta regression. The beta regression is unique in its ability to adapt to many data trends despite skewness and other factors. The beta regression is also unique in its use of proportions and percentiles as its dependent variable. The paper will look at the beta regression from different perspectives, consisting of frequentist and Bayesian, as well as adjusting for zero-one inflation. Finally, the paper will show the utilization of the beta regression in applications such as experimental studies concerning BMI percentiles and operational data on crude oil proportions after distillation. Statistical programs such as R and OpenBUGS will be used in this paper to give readers the tools needed to fit beta regression and interpret the output.