Relationships between intensity of deformation induced Cr-Al chemical zoning and geometrical properties of spinel: An approach applying machine learning analysis