Automated fraud detection model for electricity consumption

Other Title(s)

نموذج آلي لكشف الاحتيال في استهلاك الكهرباء

Dissertant

al-Nazli, Musab Bassam Musa

Thesis advisor

Ashur, Wisam Mahmud

University

Islamic University

Faculty

Faculty of Engineering

Department

Department of Computer Engineering

University Country

Palestine (Gaza Strip)

Degree

Master

Degree Date

2018

English Abstract

Financial losses due to financial frauds are mounting, recognizing the problem of losses and the area of suspicious behavior is the challenge of fraud detection. Applying data mining techniques on financial statements can help in pointing out the fraudulent usage.

It is important to understand the underlying business objectives to apply data mining objectives. Electricity consumer dishonesty is a problem faced by all power utilities that managed by a financial billing system worldwide.

Finding efficient measurements for detecting fraudulent electricity consumption has been an active research area in recent years.

This research presents a new model towards Non-Technical Loss (NTL) detection in electricity consumption utility using data mining techniques. Our Contribution proposed the genetic algorithm (GA) as an optimization technique to get the best KNN features.

The GA joint with KNN to build a hybrid GAKNN fraud detection model.

These frameworks improve the accuracy of the fraud detection system, then we compared the results with other hybrid Fraud detection models methods based on the decision tree, Naive Bayes and KNN. We implemented three fraud detection models, in the first, we implemented a KNN classification model based on the output of feature selection data by GA-KNN. In the second model, we implemented a binary classification decision tree based on the output of feature selection data by GA-KNN . In the third, model we implemented a multiclass Naive Bayes model, trained by classifier Y (the output of feature selection data by GA-KNN) . The KNN FDM has given the best accuracy in fraud detections with hit rate around 98%. This work applies a suitable data mining technique in this field based on the financial billing system for GEDCO in Gaza city.

The Selected techniques used in developing a fraud detection model.

The efficiency and accuracy of the model were tested and evaluated by one scientific method and reached one accepted technique. The intelligent model developed in this research predicts and selects suspicious customers to be inspected on-site. The model increases the detection hit rate of 20 % random manual detection to 98% intelligent detection.

Main Subjects

Information Technology and Computer Science

Topics

• Electric power
• Fraud
• Consumption

No. of Pages

51

Table of Contents

• APA
• MLA
• AMA

Language

English

Data Type

Arab Theses

Record ID

BIM-905173