Clustering of Thai Handcraft Customers using Combined SOM and K-means Algorithm

O. Chaimeun and A. Srivihok (Thailand)


Data mining, Neural Networks, SOM Algorithm, K means Algorithm, 2 Stage clustering


Entering the information era, information technology has been developed rapidly and has become significant for every business. Online business has been developed tremendously, however there are some pitfalls hidden since there is no face to face contact and customers are always long distance. It is difficult to understand the customer background and behaviors. It is worthwhile if the customer behaviors in product orders are more understood, this will enhance the strategic and marketing planning. The purpose of this research is to use the principles of the data mining to cluster Thai handicraft customers by using a hybrid algorithm, which is high performance for segmentation. The two stage clustering methods including: Self's Organizing Map (SOM) and K-means algorithm are used in this study. At the first stage, SOM algorithm is applied to calculate the optimum number of the clusters (k) that should be used for clustering. Output from first stage that has become input to the second stage, k value is used for k-means algorithm. The factor for segmentation data includes 6 factors: Zone address of customer, Date of order, payment type, product type, Quantity and business type of customer.

Important Links:

Go Back