README.md

In GWlasso: Geographically Weighted Lasso

GWlasso

The goal of GWlasso is to provides a set of functions to perform Geographically weighted lasso. It was originally thought to be used in palaeoecological settings but can be used to other extents.

The package has been submitted to CRAN and is awaiting evaluation

Installation

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

# install.packages("devtools")
devtools::install_github("nibortolum/GWlasso")

Example

This is a basic example on how to run a GWlasso pipeline:

library(GWlasso)

## compute a distance matrix from a set of coordinates
distance_matrix <- compute_distance_matrix <- function(coords, method = "euclidean", add.noise = FALSE)

## compute the optimal bandwidth 
  myst.est <- gwl_bw_estimation(x.var = predictors_df, 
                              y.var = y_vector,
                              dist.mat = distance_matrix,
                              adaptive = TRUE,
                              adptbwd.thresh = 0.1,
                              kernel = "bisquare",
                              alpha = 1,
                              progress = TRUE,
                              n=40,
                              nfolds = 5)

## Compute the optimal model
my.gwl.fit <- gwl_fit(myst.est$bw,
                      x.var = data.sample[,-1], 
                      y.var = data.sample$WTD,
                      kernel = "bisquare",
                      dist.mat = distance_matrix, 
                      alpha = 1, 
                      adaptive = TRUE, progress = T)

## make predictions 

predicted_values <- predict(my.gwl.fit, newdata = new_data, newcoords = new_coords)

GWlasso documentation built on April 3, 2025, 7:08 p.m.