Integration K-Means clustering and AHP for recommendations batik MSMEs

Program study of Information System, Department of Engineering, Universitas Trunojoyo Madura, Bangkalan, Indonesia
Program study of Informatic Engineering, Department of Engineering, Universitas Trunojoyo Madura, Bangkalan, Indonesia

^* Corresponding author: ykustiyahningsih@trunojoyo.ac.id

Abstract

Batik MSME industry is a creative industry sector in Indonesia which contributes quite a lot to Gross Domestic Product. Batik products have been recognized worldwide as one of creative products from Indonesia by UNESCO which confirmed batik as an intangible Cultural Heritage of Humanity. There are around 250 batik makers in Madura Indonesia. The problem is that the large number of batik craftsmen makes it difficult for cooperatives to determine MSME priorities and the Cooperative Work Program. Some batik indicator data is not all filled and there is still categorical and numerical data. The aim is to group batik based on the number of workers, number of products, age, education, business license, turnover, and number of batik motifs. The method used is data preprocessing using Min-Max normalization to convert categorical data into numerical and averages to overcome imputation of empty data. The data grouping method uses K-Means Clustering. AHP is used to determine indicators that have most influence on clustering and ranking of Batik MSMEs. The research contribution is integration of K-Means with AHP and preprocessing techniques. The most optimal cluster evaluation technique uses SSE. Based on the test results, the optimal cluster is K=3, with an SSE value = 0.287, Cluster 1 (Low) = 28%, Cluster 2 (medium) = 33%, and cluster 3 (High) = 39%. The results of recommendations for four highest weighting criteria using AHP are number of customers 24%, employee training 18.8%, product branding 17%, market place 16.3%.