Funct. Mater. 2026; 33 (2): 353-361.

The article addresses the problem of predicting the solubility of impurities in copper-based solid solutions using multi-model techniques and reducing the dimensionality of predictor sets. Classical approaches (Darken–Gurry diagram, Hume–Rothery rules) and modern criteria based on atomic, electronic, and structural parameters are reviewed. The study proposes the Fuji model with the “distance to the copper point” metric in a multidimensional predictor space. Data processing involved normalization and principal component analysis (PCA), enabling dimensionality reduction and assessment of the relative contribution of individual predictors to the variation of the feature set. The results were visualized using box plots, biplots, and Pareto diagrams, highlighting the influence of specific properties on data variation. Correlation distances between the reference material (Cu) and impurities (Ti, Pd, Mo, Cr, V) showed qualitative agreement with the empirical solubility rankings. The findings indicate that the Fuji model combined with correlation distance is an effective tool for qualitative predicting solubility trends and can be further improved by integrating additional physical parameters. This work opens new prospects for the development of quantitative solubility assessment methods in materials science.