Chemical zoning of minerals, which is commonly caused by incomplete chemical reaction, is often utilized to study magma cooling processes, and metamorphic rock reaction paths. Cr-Al chemical zoning of spinel has been reported as induced by deformation (lattice diffusion). However, there are no studies that address the correlates of chemical distribution (intensity) with deformation, and application methods of the Cr-Al chemical zoning. In this study, we observed differences for intensities of the Cr-Al chemical zoning with various geometrical properties of spinel grains within a dunite sample in the Transition Zone in the Horoman Peridotite Complex in Hokkaido, Japan. Using machine learning analysis, we present relationships between the intensities of the Cr-Al zoning and geometrical properties of spinel grains. We examine connections between the relationships and deformation mechanisms of spinel and estimate deformation temperature based on the results. As a consequence, the spinel grains are clustered into three groups based on the chemical zoning intensity. The intensity is more importantly affected by grain size than aspect ratio and is much greater with increasing grain size. These results suggest that lattice diffusion actively contributes more to total diffusion creep of spinel than grain boundary diffusion with increasing grain size. The deformation temperature of spinel is estimated as 1250°C–1100°C by comparing diffusion flux ratio (Rdiff) and the spinel grains.