An Overview of Bayesian Inference
Song S. Qian, Mark R. DuFour, Ibrahim Alameddine · 2022
The issue of model assessment and evaluation is addressed with predictive distributions of either future observations or predictive distributions of some known characteristics of the data. The problems of estimation and distribution are combined; instead of using the maximum likelihood estimator, parameter estimation is achieved by deriving the posterior distributions of unknown parameters. The uncertainty associated with such tests is often the source of confusion in practice. The errors associated with measurement uncertainty, lags in pollutant concentration changes and/or in biotic response to management actions, and lags in implementation of total maximum daily load plans often lead to conflicting results between water quality model predictions and statistical analysis of monitoring data. The Neuse River estuary is seen as threatened by eutrophication from rapid urban development in the upper Neuse River Basin and expanded agriculture activity in the middle and lower Neuse River Basin.