| Title: | Sampling Effort Proxies for Species Distribution Models |
|---|---|
| Description: | Addresses geographic sampling bias in Species Distribution Models (SDMs) for data-poor regions. It generates sampling effort proxies using open-source infrastructure data, allowing researchers to sample bias-aware background points. This prevents models from learning sampling artifacts instead of true ecological niches. |
| Authors: | Aamna Naveed [aut, cre] |
| Maintainer: | Aamna Naveed <[email protected]> |
| License: | MIT + file LICENSE |
| Version: | 0.1.0 |
| Built: | 2026-11-18 14:28:12 UTC |
| Source: | https://github.com/cran/biasProxyR |
Plots occurrence points over a sampling effort proxy surface and calculates a "Bias Correlation Score" to quantify how much the data is skewed toward human infrastructure.
diagnose_sampling_bias(occurrences, effort_surface)diagnose_sampling_bias(occurrences, effort_surface)
occurrences |
An sf object containing occurrence points. |
effort_surface |
A SpatRaster object representing sampling effort. |
A list containing the bias score (percentage) and a diagnostic message.
if (requireNamespace("rnaturalearthdata", quietly = TRUE) && requireNamespace("osmdata", quietly = TRUE) && curl::has_internet()) { try({ effort <- generate_effort_surface("Luxembourg", resolution = 1000) bg_points <- sample_bias_aware_background(effort, n_points = 100) diagnose_sampling_bias(bg_points, effort) }, silent = TRUE) }if (requireNamespace("rnaturalearthdata", quietly = TRUE) && requireNamespace("osmdata", quietly = TRUE) && curl::has_internet()) { try({ effort <- generate_effort_surface("Luxembourg", resolution = 1000) bg_points <- sample_bias_aware_background(effort, n_points = 100) diagnose_sampling_bias(bg_points, effort) }, silent = TRUE) }
Downloads road data for a given country and calculates distance to roads to create a proxy for human sampling effort.
generate_effort_surface(country = "Pakistan", resolution = 10000)generate_effort_surface(country = "Pakistan", resolution = 10000)
country |
Character. Name of the country (e.g., "Pakistan"). |
resolution |
Numeric. Output raster resolution in meters (default = 10000). |
A SpatRaster object where higher values indicate higher probability of human sampling effort (closer to roads).
Generates background points for Species Distribution Models (SDMs) that are weighted by a sampling effort proxy surface. This cancels out geographic sampling bias in data-poor regions.
sample_bias_aware_background(effort_surface, n_points = 10000)sample_bias_aware_background(effort_surface, n_points = 10000)
effort_surface |
A SpatRaster object representing sampling effort (e.g., from generate_effort_surface). |
n_points |
Numeric. The number of background points to generate (default = 10000). |
An sf object containing the sampled background points.
if (requireNamespace("rnaturalearthdata", quietly = TRUE) && requireNamespace("osmdata", quietly = TRUE) && curl::has_internet()) { try({ effort <- generate_effort_surface("Luxembourg", resolution = 1000) bg_points <- sample_bias_aware_background(effort, n_points = 1000) }, silent = TRUE) }if (requireNamespace("rnaturalearthdata", quietly = TRUE) && requireNamespace("osmdata", quietly = TRUE) && curl::has_internet()) { try({ effort <- generate_effort_surface("Luxembourg", resolution = 1000) bg_points <- sample_bias_aware_background(effort, n_points = 1000) }, silent = TRUE) }