CLASSIFICATION OF BANDUNG REGENCY WELLS WATER BASED ON MINERAL ELEMENT USING CLUSTER ANALYSIS PARTITION AND HIERARCHY
Wells are one of the water sources used to satisfy daily needs. However, the quality of well water is determined by its mineral element. The purpose of this study was to identify the most optimal cluster analysis method and review the quality of well water based on drinking water quality standards....
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/83480 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | Wells are one of the water sources used to satisfy daily needs. However, the quality of well water is determined by its mineral element. The purpose of this study was to identify the most optimal cluster analysis method and review the quality of well water based on drinking water quality standards. Cluster analysis is used to group data by maximizing data similarity within a cluster and minimizing data similarity with other clusters. This study reviewed well water in 160 observation locations in Bandung district containing six minerals, namely As, Cd, Fe, Mn, Pb, and Zn. Observation locations were classified using Partitions (K-Means, K-Medians, K-Medoids) and Agglomerative Hierarchies (Single, Average, and Complete Linkage) with Euclid and Manhattan distance metrics. Each cluster analysis method was evaluated to determine the optimal number of clusters (k) with the Elbow method for Partitions, while Mojena for Hierarchies. Then, conduct a method selection to obtain the best cluster analysis method; (i) partition using Silhouette Coefficient (SC), Calinski-Harabasz Index (CHI), and Xie-Beni Index (XBI), obtained K-Means Euclid and K-Medians Manhattan, (ii) Agglomerative hierarchy only using Silhouette Coefficient (SC), obtained Average Linkage Euclid and Single Linkage Manhattan. Based on the results of cluster analysis, each member is mapped into many elements that satisfy drinking water quality standards, called well water quality levels. Each cluster analysis method has a range of well water quality levels one to four, the higher the level the better. Based on the analysis of regional characteristics, K-Medians Manhattan is the best cluster analysis method in the Partition method, while Average Linkage Euclid is the best in the Hierarchy method. However, K-Medians Manhattan better represents the distribution conditions of well water quality in Bandung Regency. The correlation of K-Means Euclid with quality level one is quite high, which is 0.75. The results of this study can be an important basis in water resource management and public health protection. |
|---|