Material Selection: Material Perception Data Analysis Using Clustering Analysis and Association Rule Analysis of Data Mining

Abstract

To select materials suitable for products, material perception, which is the feeling consumers have about materials, has been studied. Material perception data were obtained through surveys using digital logic for bipolar adjective pairs. The material perception data were analyzed through unsupervised learning of data mining. Prior to data analysis, to increase the reliability of the data, the homogeneity of the data between surveys was tested using clustering analysis, correlation analysis and chi-squared test. After checking the homogeneity of the data between surveys, the data were merged. The merged material perception data were analyzed using relative frequencies, hierarchical clustering, and association rules. The relative frequencies obtained from survey participants' selections were used to determine the prevailing perceptions of each material and as basic data for other analyses. In the hierarchical clustering analysis, hierarchy was identified using distances within clusters and distances between clusters. Through association rule analysis, the consumer's simultaneous perceptions of the material can be known, so not only the individual characteristics of the material but also the relational characteristics can be considered when selecting materials based on consumer's perception. The analyzed characteristics were designed into a material perception map, and this material perception map will be a powerful tool to help product designers make better choices that match consumers' perception and experience when selecting materials.