SOM 혼성군집 모형을 이용한 쌀 시장세분화

Combined SOM Model for Rice Market Segmentation

Multidimensional Scaling( MDS), Self-Organizing Map(SOM) and K-means method are common tools for clustering analysis in market segmentation. This study investigate validity analysis of alternative clustering results obtained using the algorithm named MSK(MDS+SOM+K-means) composite model . Thus, the MSK uses the combined MDS and SOM model to determine an initial solution, and the napplies K-means algorithm to find the final solution for the rice market segmentation. In order to verify the MSK composite model , the subjects selected for the analysis were 284 housewives living in Seoul. The reported results show that the MSK model is significantly better than the MDS, SOM, K-means, and MDS+SOM model with respect to mean within cluster variations(MWCV). As a result, the MSK model showed the optimal segments number and the rice market in Korea was divided in to 9 segments. Each segment was identified by distinctive characteristics such as consumer behavior, demographic characteristics and purchasing attitudes. Therefore, there are consumer groups with various shopping orientations and purchasing behavior s in the rice market.