KLRfome - Kernel Logistic Regression on Focal Mean Embeddings

Release for completion of review for Journal of Open Source Software (JOSS). Updates included minor bug fixes, code standardization, major updates to documentation, addition of a vignette, and coverage of unit testing.

KLRfome is a package for predicting a geographic area's relative sensitivity for the presence of archaeological sites. This is achieved by fitting, predicting, and visualizing a Kernel Logistic Regression model on mean feature embeddings. This regression algorithm and package are created to improve upon the current methods in archaeological predictive modeling. These improvements include 1) modeling rich descriptions of archaeological land forms by mitigating undesirable spatial correlation between samples, 2) explicitly modeling the similarity between archaeological sites as characterized by rich features, 3) the ability to define research specific similarity measures, and 4) focal window prediction that can be modified based on theory or management goals.