Create stars dataset

In sfclust: Bayesian Spatial Functional Clustering

knitr::opts_chunk$set(fig.align = "center", eval = TRUE)
knitr::opts_chunk$set(fig.height = 7, fig.width = 8, dpi = 150, out.width = '100%')
knitr::opts_chunk$set(comment = "#>")

Our package sfclust is designed to work with spatio-temporal data structured as a stars object. In this vignette, we demonstrate how to create and manipulate a basic stars object.

library(stars)
library(ggplot2)

Simulated data

To begin, we need data that varies across both regions and time. In this example, we simulate such data for illustrative purposes, but in practice, this would come from your study's specific regions and time periods.

set.seed(10)
space <- st_make_grid(cellsize = c(1, 1), offset = c(0, 0), n = c(3, 2))
time <- seq(as.Date("2025-01-01"), by = "1 month", length.out = 5)

Next, we simulate variables associated with each region and time point. In this case, we create values for cases, temperature, and precipitation. Each variable is stored in a matrix where rows correspond to regions and columns to time points.

cases <- matrix(rpois(30, 100), nrow = 6, ncol = 5)
temperature <- matrix(rnorm(30), nrow = 6, ncol = 5)
precipitation <- matrix(rnorm(30), nrow = 6, ncol = 5)

Creating a stars object

We now use the simulated data to construct a stars object. This object will contain spatial, temporal, and variable dimensions. While there are several ways to build such objects, the following method is commonly used:

stdata <- st_as_stars(
  cases = cases, temperature = temperature, precipitation = precipitation,
  dimensions = st_dimensions(geometry = space, time = time)
)
stdata

Manipulating a stars object

Once the stars object is created, you can use any of the methods provided by the stars package. For example, we can visualize the cases variable across time:

ggplot() +
    geom_stars(aes(fill = cases), data = stdata) +
    facet_wrap(~ time) +
    scale_fill_distiller(palette = "RdBu")

You can also add additional variables to the stars object. For instance, let’s add a population variable:

stdata["population"] <- rep(rpois(6, 1000), each = 6)
stdata

sfclust documentation built on June 8, 2025, 10:11 a.m.