Holocene reef growth R project for Macedo and van Woesik

This code was used for the analysis for "Growth rates for coral reefs peaked at 25 °C through the Holocene" by Tonya Macedo and Robert van Woesik. <br>Overview:Main dataset: "Holocene_Reef_Growth_Final.csv" is the final dataset used for modeling, containing 1,890 samples. Main goals:Use a GLMM in INLA to determine fixed effects for coral growth rates during the Holocene. Use deep learning to evaluate non-linear relationships between variables of interest and coral reef growth rates through the Holocene. Reproduce figures from the publication or generate new figures with new data. Main scripts: Holocene_INLA_GLMM.RSets up model formulasFits modelsCompares and visualizes INLA model resultsHolocene_Deep_Learning.RSets up the H2O deep learning environmentFits 30 models using 6 different deep learning configurationsCompares and visualizes model resultsHolocene_Figures.qmdReproduces the figures used in the manuscriptUsage guidance:Each script is set up so that it can be run standalone. There is no order that the scripts need to be run. The project uses the here package for reproducible file paths, no need to set working directory.The repository is organized so that it can be cloned and executed as an R project.本代码用于Tonya Macedo与Robert van Woesik发表的题为《全新世珊瑚礁生长速率在25℃达到峰值》的研究的数据分析工作。 ## 概况 主数据集：Holocene_Reef_Growth_Final.csv 为本研究建模所用的最终数据集，共包含1890条样本。 ## 研究目标 一是借助集成嵌套拉普拉斯近似（Integrated Nested Laplace Approximation, INLA）中的广义线性混合模型（Generalized Linear Mixed Model, GLMM），确定全新世时期珊瑚生长速率的固定效应；二是采用深度学习方法，探究研究关注的变量与全新世珊瑚礁生长速率之间的非线性关系；此外可复现已发表论文中的图表，或基于新数据生成全新可视化成果。 ## 主要脚本 1. Holocene_INLA_GLMM.R：用于设置模型公式、拟合模型，以及对比并可视化INLA模型的结果。 2. Holocene_Deep_Learning.R：用于配置H2O深度学习环境，基于6种不同的深度学习配置拟合30个模型，并对比、可视化模型结果。 3. Holocene_Figures.qmd：用于复现论文手稿中所用的图表。 ## 使用说明 所有脚本均支持独立运行，无强制执行顺序。本项目借助here包实现可复现的文件路径管理，无需手动设置工作目录。该代码仓库可直接克隆后作为R项目执行。