Pollution can cause the environment to not function properly and ultimately harm humans and other living things. Environmental pollution is a problem that needs to be resolved because it involves the safety, health, and survival of living things. Air pollution in Pekanbaru due to a long dry season has resulted in forest fires. Then, 70% of drinking water is contaminated by faecal waste. In addition, the contamination of the land by the Chevron company resulted in residents suing the company. Until now, there has been no research that has carried out a comparison between methods for grouping villages affected by environmental pollution at the provincial level in Indonesia, so it is important to select the best method for carrying out the grouping. The limitations of this research are the use of three methods for clustering: K-Medoids, Fuzzy C-Means, and DBSCAN.  Optimum cluster results with Fuzzy C- Means clustering there are 5 clusters, K-Medoids with 4 clusters, and DBSCAN method clustering has 2 clusters and 7 outlier provinces. The best method based on the lowest ICD rate value on the grouping of provinces in Indonesia on the number of villages affected by environmental pollution is using the Fuzzy C- Means method. 

The results showed that Fuzzy C-Means with five clusters have an optimal value compared to K-Medoids and DBSCAN with an ICD rate value of 0,351. Then continued with nonparametric ANOVA analysis. On the four pollution variables (water pollution, soil pollution, air pollution, and no affected by pollution), it can be concluded that the variables of the number of villages affected by pollution have an influence on formation of provincial groupings in Indonesia. This method can be used by the government to improve the quality of villages that are clean from pollution in Indonesia, monitoring and evaluation based on the clusters formed.

Author: Idrus Syahzaqi, S.Stat., M.Stat