Data and materials for "Validation practice in machine-learning estimation of the urban heat island: a systematic review of 222 studies"

Materials of a systematic review of validation practice in machine-learning estimation of urban air temperature, land surface temperature and urban heat island intensity (2 514 Web of Science records, 2015–2026; 222 studies analysed in full text with an 84-field grid). The deposit contains the review protocol with its dated amendments and the search expression; the title-and-abstract screening and full-text data-extraction instructions, including the 61 dated coding rules; the screening decision and code of each of the 2 514 records; the stratified blind verification sample of 150 records with the adjudication of disagreements; the 84-field extraction of the 222 analysed studies; Supplementary Table S1 and Supplementary List S2 of the article; and the Python scripts that reproduce the descriptive statistics, the meta-regression and the figures. See README.md for the structure and the coding conventions.