ngme2 Documentation
In ngme2: Linear Latent Non-Gaussian Models with Flexible Distributions

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

The full documentation for ngme2, including worked examples and rendered output, is available on the package website:

Getting Started

Overview of ngme2: Introduction to the package and its main features
Installation and configuration (OpenMP): How to install ngme2 and enable parallel computation

AR(1) model: Autoregressive order-1 models with non-Gaussian noise
Random walk model: RW1 and RW2 models
Ornstein-Uhlenbeck process: Continuous-time OU models
SPDE Matérn model: Spatial models based on the SPDE approach
Separable space-time (tensor product) model: Space-time models via tensor products
Non-separable space-time model (advection-diffusion): Anisotropic and advective space-time models
Matérn SPDE model on metric graph: Spatial models on networks and metric graphs
Bivariate type-G models: Joint models for two response variables
Using the generic model: How to define custom latent models

Combined Gaussian and NIG noise models: Mixed noise specifications
Model estimation and prediction: How estimation and prediction work in ngme2
Models with replicates: Handling replicated observations
Cross-validation: Time series and spatial cross-validation tools
Prior templates: Stationary and non-stationary prior specifications
SGLD posterior-like sampling: Uncertainty quantification via stochastic gradient Langevin dynamics