An Improved Estimation for Heterogeneous Datasets with Lower Detection Limits regarding Environmental Health

Area of Science:

• Environmental Science
• Statistical Modeling
• Bayesian Inference

Background:

• Analyzing environmental data with lower detection limits (LDL) is crucial.
• Classical estimation methods for mixture models have limitations with LDL data.
• Previous research primarily used classical estimation for mixture models.

Purpose of the Study:

• To propose a Bayesian analysis for environmental data with LDL using mixture models.
• To explore optimal mixture distributions and identify the best number of components.
• To investigate the sensitivity of Bayesian estimators to various factors.

Main Methods:

• Bayesian analysis utilizing mixture models.
• Exploration of optimal mixture distributions and component numbers.
• Sensitivity analysis of estimators concerning LDL, parameters, hyperparameters, and sample size.

Main Results:

• The proposed Bayesian estimators showed superior performance compared to existing classical estimators.
• Sensitivity analysis provided insights into estimator behavior under different conditions.
• Optimal mixture distributions and component numbers were identified for environmental data.

Conclusions:

• Bayesian analysis with mixture models provides a robust framework for environmental data with LDL.
• The proposed method offers more reliable and convincing results than traditional approaches.
• This study advances the statistical methodology for handling challenging environmental datasets.