Abstract: According to the present disclosure, there is provided a fragrance property prediction system, including a data converter configured to analyze high-dimensional physicochemical property data matched to a plurality of fragrances through a principal component analysis technique and project the analyzed data into three-dimensional data on a three-dimensional physicochemical property coordinate system, a grouping unit configured to group three-dimensional coordinates for each fragrance in the three-dimensional physicochemical property coordinate system for each perceptual property category based on perceptual property category information previously matched for each fragrance and configured to calculate a centroid coordinate for each grouped perceptual property category, and a controller configured to convert a physicochemical property vector of unknown fragrance into the three-dimensional coordinate in the three-dimensional physio-chemical coordinate system, and predict a perceptual property category of the unkn