Abstract:While log-Gaussian Cox process regression models are useful tools for modeling point patterns, they can be technically difficult to fit and require users to learn/adopt bespoke software. We show that, for suitably formatted data, we can actually fit these models using generalized additive model software, via a simple line of code, demonstrated on R by the popular mgcv package. We are able to do this because a common and computationally efficient way to fit a log-Gaussian Cox process model is to use a basis function expansion to approximate the Gaussian random field, as is provided by a generic bivariate smoother over geographic space. We further show that if basis functions are parameterized appropriately then we can estimate parameters in the spatial covariance function for the latent random field using a generalized additive model. We use simulation to show that this approach leads to model fits of comparable quality to state-of-the-art software, often more quickly. But we see the main advance from this work as lowering the technology barrier to spatial statistics for applied researchers, many of whom are already familiar with generalized additive model software.

What problem does this paper attempt to address?

The paper primarily addresses the issue of how to use Generalized Additive Model (GAM) software to fit Log-Gaussian Cox processes, thereby simplifying the technical threshold for spatial statistical analysis. Specifically, the paper addresses the following key issues: 1. **Simplifying Technical Difficulty**: Log-Gaussian Cox process regression models can be challenging to fit in practice, especially when the model includes latent fields. The paper proposes a method that uses Generalized Additive Model software to fit point event data in a straightforward manner, thus avoiding the need for specialized software. 2. **Fitting Models Using General Software**: The authors demonstrate a method that allows the fitting of Log-Gaussian Cox process models using the popular `mgcv` package in the R environment with just a few lines of code, thereby lowering the technical barriers for applied researchers to use such complex models. 3. **Estimating Spatial Covariance Function Parameters**: If the basis functions are appropriately parameterized, the Generalized Additive Model can be used to estimate the spatial covariance function parameters of the latent random field. 4. **Model Fitting Quality**: Simulation studies show that this method can achieve model fitting quality comparable to state-of-the-art software, and it is often faster. 5. **Lowering Technical Barriers**: The main advancement of this work is in lowering the technical threshold of spatial statistics for applied researchers, many of whom are already familiar with Generalized Additive Model software. In short, the paper proposes a novel method that makes it possible to fit complex Log-Gaussian Cox process models using common statistical software (such as the `mgcv` package in R), which is significant for promoting the application of such models across multiple disciplines.

Fitting Log-Gaussian Cox Processes Using Generalized Additive Model Software

Quick inference for log Gaussian Cox processes with non-stationary underlying random fields

Modelling Ride-Sourcing Matching and Pickup Processes Based on Additive Gaussian Process Models

lgcp An R Package for Inference with Spatio-Temporal Log-Gaussian Cox Processes

A two-stage Cox process model with spatial and nonspatial covariates

Inhomogeneous log-Gaussian Cox processes with piecewise constant covariates: a case study in modeling of COVID-19 transmission risk in East Java

Careful prior specification avoids incautious inference for log-Gaussian Cox point processes

Generalized additive models with flexible response functions

Bayesian Optimization through Gaussian Cox Process Models for Spatio-temporal Data

Mixture of Gaussian Processes and its Applications

Additive Kernels for Gaussian Process Modeling

Representing Additive Gaussian Processes by Sparse Matrices

Bayesian Variable Selection in Double Generalized Linear Tweedie Spatial Process Models

Bayesian model-based spatiotemporal survey design for log-Gaussian Cox process

Minibatch Markov chain Monte Carlo Algorithms for Fitting Gaussian Processes

Bayesian Non-linear Latent Variable Modeling via Random Fourier Features

Gaussian process regression with log-linear scaling for common non-stationary kernels

Laplace approximation for fast Bayesian inference in generalized additive models based on penalized regression splines

Flexible Basis Representations for Modeling Large Non-Gaussian Spatial Data

Gaussian Markov random field spatial models in GAMLSS

A Bivariate Copula Additive Model for Location, Scale and Shape