Historical mine and mineral deposit datasets are routinely used to inform quantitative mineral assessment models, but they also can contain a wealth of supplementary qualitative information that is generally underutilized. We present a workflow that uses correspondence analysis, an exploratory tool commonly applied to multivariate abundance data, to better utilize qualitative data in these historical datasets. The workflow involves extraction of qualitative information on ore mineralogy from a mineral deposit database, attaches those data to a target geological feature, and analyzes the underlying data structure with correspondence analysis and hierarchical clustering. The output of correspondence analysis is inversely weighted to the relative frequency of ore minerals, and therefore rare mineral species (i.e., those with unusually low frequencies) can disproportionately contribute to the total variance of the dataset. We present a novel technique for aggregating frequencies of rare mineral species that minimizes this effect. We apply this workflow to evaluate how ore mineral assemblages in former and active mines vary in spatial relation to silicic calderas in the southwestern United States. The most common ore mineral associations observed spatially and genetically associated to calderas include those related to polymetallic, base metalrich systems and epithermal Au-Ag systems. Three other groups of mineralized calderas were identified, including: (1) Hg-Sb mineralized calderas in the northern Great Basin and western Nevada volcanic field; (2) calderas associated with elevated abundances of Mn oxides/hydroxides, fluorite, and Be-minerals, mostly in eastern Utah and New Mexico; and(3) calderas with numerous U ± F deposits, which are located in central Colorado, the eastern Great Basin and in northern Nevada. The latter three groups are associated with economically significant critical mineral resources, including the Li resources of the McDermitt complex and Be associated with the Spor Mountain on the margin of the Thomas caldera complex. We conclude that correspondence analysis is a promising technique that can enhance data exploration of the qualitative information held within mineral deposit datasets. Consequently, it could have numerous applications for mineral potential mapping, resource assessment projects, and characterization of mineral systems.