Data and code to fit species distribution models using projection predictive variable selection

This repository contains data and code to replicate parts of the analysis of a paper currently under review in the journal Ecological Informatics and entitled Shrinkage-based Bayesian variable selection for species distribution modelling in complex environments: an application to urban biodiversity. The included data and code allow users to fit each of the two species distribution models (selective vs exhaustive models) with each of the two cross-validation approaches (regular stratified vs. environmentally stratified. In addition, code is provided to examine the sensitivity of the variable selection to changes in the parameterisation of the regularised horseshoe prior.本仓库收录了用于复现一篇已投稿至《生态信息学（Ecological Informatics）》期刊、目前处于审稿阶段的论文的部分分析工作的数据与代码，该论文题为《面向复杂环境下物种分布建模的基于收缩的贝叶斯变量选择方法：在城市生物多样性中的应用》。本次附带的数据与代码可支持用户结合两种交叉验证方法（常规分层交叉验证与环境分层交叉验证），分别拟合两类物种分布模型（选择性模型与穷尽式模型）。此外，本仓库还提供了用于探究变量选择对正则化马蹄先验（regularised horseshoe prior）参数化调整的敏感性的代码。