Outliers Detection on Fisheries Commodity Transaction from Local Market in Tual City based on the x-means Clustering

Abstract

Correctly estimating factors that affect the income and profitability of coastal communities specifically catching fishermen is one way to increase the level of sustainable coastal community welfare. The purpose of this study is to detect outliers from fisheries commodity transactions in the fish market in Tual City, to have full view on which cases that may have certain traits that can improve the profitability of fishermen in the region. The research method used in this study is a survey method in which used as many as 1420 datasets obtained from interviews conducted starting from June 2017 to June 2018. The data mining approach is used in this study, in order to obtain the appropriate outliers then x-means clustering is used to cluster profitable transaction data, then data collected further being process with outlier detection. We found 25 high profitability transactions of 1420 datasets. The outlier analysis was carried out and found 1 outliers. We have figured that the species of fish and the location of the seller are the factors that influence the level of profit from the fishermen. This result is an input for policy makers and catch fishermen in conducting capture fisheries.