Supporting of Waste Management in Indonesia Using Self Organizing Map for Clustering Analysis

Abstract

The amount of waste in Indonesia is continuously increasing, along with the increasing population and welfare. The government already takes various efforts to control the amount of garbage; for example, they prohibit using disposable plastic in any form. Mapping waste generation and waste composition hopefully will help the government (or local government) make a more precise policy. Therefore, this research employed a modern clustering analysis that was Self Organizing Maps (SOM). SOM is an Artificial Neural Network (ANN) that uses unsupervised learning to reduce data dimensions into two dimensions. SOM produces low-dimensional visualizations of high-dimensional data. The data of waste generation and composition were obtained from the National Waste Management Information System in 2017-2018. The variables processed were the daily average of waste per square kilometer (ton/km2), the daily average of waste per person (kg/person), and the daily average of waste per sub-district area (tons/sub-district). The researchers built the 3x3 hexagonal topology. The districts of Brebes, Buleleng, Cilacap, and Jepara were grouped into areas with high waste generation compared to other districts. Meanwhile, based on the composition of garbage, Morowali, Sinjai, and Palangkaraya districts were the districts where the composition of food waste, plastic waste, and textile waste was high. Tableau was used to visualize the result into map.