Machine-learning prediction of optical band gaps and photocatalytic performance in rare-earth-doped spinel ferrites
Year: 2026
Type: article
Abstract: Rare-earth substitution is widely used to regulate the lattice structure, defect chemistry, optical absorption, charge-carrier dynamics, and magnetic recoverability of spinel ferrites. Published photocatalytic results, however, are distributed across different hosts, dopants, pollutants, irradiation sources, catalyst dosages, and reaction times, which prevents direct comparison by degradation percentage alone. A leakage-controlled and reproducible small-data machine-learning workflow was therefore developed from 37 records compiled from published literature and arranged into 20 analytical groups, with closely related catalyst-condition records retained within the same validation group. Ridge regression, random forest, extremely randomized trees, and gradient boosting were evaluated using leave-one-group-out cross-validation. Random forest gave the lowest error among the four candidate regressors, with a mean absolute error of 21.43 percentage points (95% analytical-group bootstrap confidence interval: 12.88–28.10), a root mean square error of 27.11 percentage points (95% confidence interval: 14.67–34.19), and R 2 = − 0.56. A training-fold mean baseline gave a lower MAE of 19.96 percentage points, confirming limited generalization performance for unseen groups. Random row-wise five-fold validation was substantially more optimistic (R 2 = 0.36), illustrating the risk of protocol leakage in literature-derived datasets. Optical band gap remained an informative descriptor but was not sufficient to explain activity. The study provides a reproducible benchmark, uncertainty-aware interpretation, and a practical reporting checklist for future machine-learning-assisted studies of rare-earth-doped spinel ferrites.
Source: Scientific Reports
Language: English
Subfield: Materials Chemistry
Field: Materials Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $2,690
OpenAlex ID: W7220831424