Bayesian analyses of an exponential-Poisson and related zero augmented type models

Area of Science:

• Environmental Science
• Statistics
• Data Analysis

Background:

• The exponential-Poisson distribution is often used for environmental data, but struggles with frequent zero values.
• Zero-inflated data presents challenges for traditional continuous statistical models.

Purpose of the Study:

• To evaluate alternative statistical models for zero-inflated environmental data.
• To compare modified exponential-Poisson models with semi-continuous alternatives.
• To assess model performance on real-world precipitation datasets.

Main Methods:

• Development and application of modified exponential-Poisson and semi-continuous models.
• Bayesian analysis using Markov Chain Monte Carlo (MCMC) simulations.
• Evaluation of MCMC convergence and model selection criteria.

Main Results:

• Several alternative models were proposed to handle zero-inflated data.
• Models were successfully applied to precipitation datasets.
• Bayesian analyses provided insights into model performance and selection.

Conclusions:

• The proposed models offer viable alternatives for analyzing zero-inflated environmental data.
• MCMC methods are effective for Bayesian analysis and model comparison in this context.
• Careful consideration of model selection is crucial for accurate environmental data interpretation.