The Dataset and code of study "Uncovering Dynamic and Nonlinear Driving Mechanisms of Production–Living–Ecological Space Change in Metropolitan Areas Using Interpretable Machine Learning"

This dataset accompanies the manuscript titled “Uncovering Dynamic and Nonlinear Driving Mechanisms of Production–Living–Ecological Space Change in Metropolitan Areas Using Interpretable Machine Learning”, submitted to the journal Sustainability. It is provided to support the reproducibility of the study’s results. The authors of the study are Jia Liao, Bin Quan, Kui Liu, and Zhiwei Deng. The dataset includes the following components: 1.“Intensity Analysis.xlsx”This file contains the results of the intensity analysis, including all three hierarchical levels of the analysis as well as the original transition matrix data. 2.“optimal hyperparameter and Validation.xlsx”This file provides the optimal hyperparameters of the XGBoost model and the corresponding validation results, including both the selected hyperparameters and the raw validation outputs generated by Python. 3.“XGBoost train and test data.csv”This file contains datasets for three time periods (2010–2015, 2015–2020, and 2020–2025). The tabular data were extracted from layers representing production–living–ecological space changes and their driving factors. These data are used as input for the XGBoost model, including both training and testing datasets. 4.“xgboost_train1/2/3.py”These scripts contain the code for training the XGBoost model. Users can utilize these scripts along with the dataset to reproduce the results of this study. 5.“SHAP1/2/3.py”These scripts are used to calculate SHAP values. Users can run these scripts with the provided dataset to reproduce the interpretability analysis results of this study.