Spatial Interpolation for data comprising hard and soft-interval forms

The Bayesian Maximum Entropy (BME) framework provides a flexible and principled approach to space-time data analysis by combining Bayesian inference with the maximum entropy principle. It supports optimal estimation using both precise (hard) and uncertain (soft) data, such as intervals or probability distributions—making it ideal for complex, real-world datasets. The BMEmapping R package implements core BME methods for spatial interpolation, enabling the integration of heterogeneous data, variogram-based modeling, and uncertainty quantification.

Installation

You can install the development version of BMEmapping from GitHub with:

# install.packages("devtools")
devtools::install_github("KinsprideDuah/BMEmapping")

Functions

• bme_map Constructs a BMEmapping object that encapsulates all required inputs (hard data, soft data, and spatial information) for performing BME interpolation.

• prob_zk Computes the posterior probability density of the variable of interest at a single unobserved location.

• q_prob_zk Computes the posterior probability density at a single unobserved location using the quantile-based (QBME) approach.

• bme_predict Estimates the posterior mean or mode, along with the associated variance, at an unobserved location.

• q_bme_predict Estimates the posterior mean or mode and associated variance at an unobserved location using the QBME approach.

• bme_cv Performs cross-validation on hard data to evaluate predictive performance of the BME model.

• q_bme_cv Performs cross-validation on hard data to evaluate predictive performance using the QBME approach.

Getting help

If you encounter a clear bug, please file an issue with a minimal reproducible example on GitHub.

Author

Kinspride Duah

License

MIT + file LICENSE