Spatial Statistical Models: an overview under the Bayesian Approach

TL;DR

This paper reviews Bayesian spatial statistical models over the past 20 years, highlighting their structure, challenges, and research gaps in the context of Big IoT Data and spatial analysis.

Contribution

It provides a systematic overview of Bayesian spatial models, discussing key elements and subclasses, and identifies research opportunities and gaps in the field.

Findings

Bayesian spatial models are underexplored compared to machine learning models.

The review covers elements like random fields, priors, and covariance functions.

Two subclasses, global and local spatial smoothing, are analyzed.

Abstract

Spatial documentation is exponentially increasing given the availability of Big IoT Data, enabled by the devices miniaturization and data storage capacity. Bayesian spatial statistics is a useful statistical tool to determine the dependence structure and hidden patterns over space through prior knowledge and data likelihood. Nevertheless, this modeling class is not well explored as the classification and regression machine learning models given their simplicity and often weak (data) independence supposition. In this manner, this systematic review aimed to unravel the main models presented in the literature in the past 20 years, identify gaps, and research opportunities. Elements such as random fields, spatial domains, prior specification, covariance function, and numerical approximations were discussed. This work explored the two subclasses of spatial smoothing global and local.