Machine Learning for Shield-Scale Soil Health Mapping: A Systematic Review

This dataset supports the systematic review article "Machine Learning for Field-Scale Soil Health Mapping: A Systematic Review", which synthesises evidence from 46 empirical studies on the application of machine learning algorithms for predicting soil chemical properties at field-scale agricultural systems (<100 Ha). The review covers supervised machine learning approaches including Random Forest, Support Vector Machines, XGBoost, Artificial Neural Networks, Convolutional Neural Networks, and ensemble methods, applied across diverse geographic contexts including China, the United States, Peru, Turkey, Spain, Canada, the Netherlands, India, Germany, Russia, Belgium, and Iran. The dataset includes the following supplementary files: Supplementary Document 1: PRISMA 2020 Checklist (complete 27-item checklist with page number references) Supplementary Document 2: Detailed Quality Assessment Results (JBI Critical Appraisal Checklist scores for all 46 included studies) Supplementary Document 3: Full Data Extraction (complete dataset including authors, year, country, objectives, ML algorithms, data sources, target variables, sample sizes, validation methods, performance metrics, strengths, and limitations) Supplementary Document 4: Hyperparameter Settings (detailed hyperparameter configurations for each ML algorithm reported in the included studies) Supplementary Document 5: Spatial Validation Methods (summary of spatial cross-validation approaches, block sizes, and validation strategies employed across studies) Supplementary Document 6: Feature Selection and Spectral Transformation Techniques (methods used for feature engineering, band selection, and spectral indices) Supplementary Document 7: Uncertainty Quantification Methods (approaches for reporting prediction intervals, standard errors, and confidence intervals) Supplementary Document 8: R² Performance Data for Boxplot (extracted R² values for ensemble and single classifier models used to generate Figure 3) Supplementary Document 9: Geographic Data (country-level data with ISO codes, study counts, and author lists used to generate Figure 4)