Muhaimin, Amri (2018) Deteksi Anomali Pada Pemakaian Air Pelanggan PDAM Surya Sembada Kota Surabaya Menggunakan Kohonen SOM dan Local Outlier Factor. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
06211440000125-Undergraduated_Theses.pdf - Accepted Version Download (3MB) | Preview |
Abstract
Kehilangan air dalam distribusi merupakan masalah yang cukup serius, kerugian yang disebabkan oleh tingkat kehilangan PDAM Surya Sembada Kota Surabaya mencapai 2 Miliar Rupiah.Pada tahun 2011 terjadi pencurian air sebanyak 100 kasus. Deteksi anomali pada penelitian dilakukan menggunakan data selama Maret 2017 – Februari 2018. Data yang digunakan adalah pemakaian air kemudian didapatkan variabel-variabel rata-rata pemakaian air, maksimal pemakaian air, dan deviasi standar pemakaian air. Algoritma Kohonen-SOM mendapatkan 45 kelompok yang dianggap kelompok anomali dengan kriteria silhouette width kurang dari rata-rata silhouette width pada kelompok yang ter-bentuk. Terdapat 45 kelompok yang terduga anomali. Local Outlier Factor menghasilkan 1229 kejadian konsumsi yang tidak normal, 1229 kejadian tersebut terdiri dari 579 rumah tangga atau pelanggan. Perhitungan frekuensi yang dilakukan mendapatkan 42 pelanggan yang terduga anomali. Hasil deteksi anomali dengan metode PDAM dari 42 pelanggan te-rsebut hanya 16 yang terde-teksi. Hal tersebut dikarenakan metode PDAM gagal menangkap perilaku konsumsi yang aneh seperti konsumsi yang konstan setiap bulan. Karakteristik pelanggan yang terdeteksi anomali adalah mempunyai rata-rata pemakaian lebih dari rata-rata pemakaian golongan dan sub-zona.
============================================================
The loss of water in the distribution is a serious problem in PDAM Surabaya, the loss caused by loss rate of PDAM Surabaya reach 2 billion Rupiah. In 2011 there was a theft of 100 cases. Detection of anomalies in the study was conducted using data during March 2017 - February 2018. The data used is water consumption then obtained the variables likemean of water use, maximum water use, and standart deviation of water usage. The Kohonen-SOM algorithm obtained 45 groups considered anomalous group with silhouette width criteria less than the silhouette width average in the group formed. There are 45 groups of unexpected anomalies. Local Outlier Factor produced 1229 unusual consumption events, 1229 incidents consisting of 579 households or customers. The frequency calculation performed gets 42 suspected anomaly customers. The result of anomaly detection with PDAM method from 42 customers was only 16 detected. This is because the PDAM method fails to capture strange consumption behaviors such as cons-tant consumption every month. The characteristic of the customer detected by the anomaly is to have an average of more than average usage of classes and sub-zones.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Uncontrolled Keywords: | Anomali, Kohonen-SOM, Local Outlier Factor |
| Subjects: | Q Science > QA Mathematics > QA278.55 Cluster analysis Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) Q Science > QA Mathematics > QA76.9.D343 Data mining. Querying (Computer science) |
| Divisions: | Faculty of Mathematics, Computation, and Data Science > Statistics > 49201-(S1) Undergraduate Thesis |
| Depositing User: | Amri Muhaimin |
| Date Deposited: | 09 Jul 2021 10:13 |
| Last Modified: | 09 Jul 2021 10:13 |
| URI: | http://repository.its.ac.id/id/eprint/57048 |
Actions (login required)
 |
View Item |