Uncovering Patterns and Relationships: A Multivariate Data Analysis Case Study

Abstract

To distinguish between different cheese varieties, a total of 24 commercial cheeses were examined using three distinct methods to assess both general and thermophysical parameters. Rheological properties of different temperatures, stretch ability, melt ability and the formation of free oil were already present in the data. Box plots helped classifying textures of different cheese types on the basis of the general and thermophysical properties. Cluster analysis and multidimensional scaling helped classify the cheese types based on a variety of properties. Principal component analysis was applied, too, to the data set to receive a mapping of the cheese. A reduction of the analyzed parameters to three principal components accounting for 76.3% of the total variation was achieved, indicating that the methods are well suited to characterize the cheese samples. Mapping aids cheese producers to design new products with defined characteristics by affecting the manufacturing protocol. Also it helps cheese dealers to decide whether to buy a new variety of cheese produced in the market. Key Words: Multivariate data, Andrew’s curves, cluster analysis, mantel test, multidimensional scaling, principal component analysis, general properties, thermophysical properties of cheese, R programming.