Inverse sampling intensity weighting (ISIW)

ISIW implements Inverse Sampling Intensity Weighting (ISIW) for adjusting geostatistical models under preferential sampling (PS).

The method is introduced in:

Hsiao, T. W. and Waller, L. A. (2025). Inverse sampling intensity weighting for preferential sampling adjustment. Online at https://doi.org/10.48550/arXiv.2503.05067

---

Overview

In spatial analyses, sampling locations are often treated as fixed and independent of the latent spatial process. However, when locations are preferentially sampled—that is, when the sampling intensity depends on the underlying spatial field—standard likelihood-based inference and kriging can be biased.

ISIW provides a practical, two-stage adjustment implemented as follows:

1. Estimate the sampling intensity at observed locations.
2. Construct inverse sampling intensity weights.
3. Fit a weighted geostatistical likelihood (with scalable Vecchia approximation).
4. Perform prediction using plug-in kriging under the adjusted model.

The method avoids specifying a joint latent model for the sampling process and spatial field, offering a computationally efficient alternative to shared-parameter approaches.

---

Installation

You can install the development version from GitHub:

# install.packages("pak")
pak::pak("tXiao95/ISIW")