Comparison Between Method Of SOM / Kohonen And Adaptive Resonance Theory 2 On Data Mining Retail Company

TESIS

Abstract

By : Anak Agung Gede Bagus Ariana

Email : gungariana86@gmail.com

Faculties : Fakultas Teknik

Department : S2 Teknik Elektro

Retail company is a company that specializes in buying and selling in small quantities, units, or retail. The retail business has an important role in the economic world, because without a retail business, a product of the manufacturer will not be delivery to the consumers. The object of research in this study is UD. Fenny, which is a retail company that sells materials and tools to make a cake. Problems faced by these companies is the difficulty to identify potential customers in order to better targeted marketing strategies. In order to perform the customer identification process, it is carried out the process of data mining in customer data clustering method. This research was conducted in two stages, the first stage is the preparation of data and the second stage is the process of clustering on the data that has been prepared. Clustering method used is the method of Self Organizing Map and Adaptive Resonance Theory 2. Both methods are methods that apply artificial neural network. Before the mining process, the data is formed into a datamart model or Recency Frequency Monetary RFM. Comparison of methods are performed by parameters for each method to obtain perfect cluster. Perfection cluster validation index is measured by using Davies-Bouldin, Dunn Index and Index Silhouette. Based on the results of study, the lowest Davies-Bouldin index value, highest Index Dunn, and Silhouette Index near by one is obtained from Self Organizing Map method. In this case, the method of Self Organizing Map can group the data better. Adaptive Resonance Theory method is able to provide more patterns in each parameter configuration changing. It can be seen from the cluster validation index values are more diverse in the method of Adaptive Resonance Theory 2 compared to Self Organizing Map method.

Keyword : Data Mining , Neural Networks , Self Organizing Map , Adaptive Resonance Theory 2

Download : File 1 File 2 File 3