Deteksi anomali pada proses bisnis dengan Fuzzy multi criteria decision making topsis dan decision tree

Kunio, Uswatun Hasana (2015) Deteksi anomali pada proses bisnis dengan Fuzzy multi criteria decision making topsis dan decision tree. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[img] Text
5111100051-undergraduate thesis.pdf

Download (3MB)
[img] Text
5111100051-paperpdf.pdf

Download (673kB)
[img] Text
5111100051-presentationpdf.pdf

Download (1MB)

Abstract

Internal fraud atau kecurangan dalam perusahaan menjadi salah satu penyebab kerugian yang paling signifikan. Fakta ini mendorong perusahaan untuk memiliki kebijakan dalam hal keamanan dan sistem informasi yang kuat dalam deteksi fraud. Tugas akhir ini fokus kepada fraud berbasis proses. Fraud berbasis proses adalah suatu kecurangan yang terjadi dalam proses bisnis. Terdapat sepuluh tipe kriteria anomali terjadinya fraud yaitu Skip Sequence, Skip Decision, Throughput Time Maximum, Throughput Time Minimum, Wrong Pattern, Wrong Decision, Wrong Duty Decision, Wrong Duty Sequence, Wrong Duty Combine serta Wrong Resource. Untuk menangkap anomali ini dilakukan pemodelan ontologi terhadap Standard Operational Procedure dan event logs. Pembuatan rules pada model ontologi menggunakan SWRL rules digunakan untuk mendapatkan kasus yang terindikasi terjadi anomali. Fuzzy Multi Criteria Decision Making TOPSIS digunakan untuk perhitungan rating anomali. Dan Decision Tree digunakan untuk klasifikasi status anomali dari tiap kasus yang ada pada event logs. Dalam tugas akhir ini dilakukan uji coba terhadap studi kasus pengajuan kredit bank dan menghasilkan pendeteksian anomali yang akurat dengan akurasi sebesar 98%, specificity sebesar 1, dan sensitivity sebesar 0,71 untuk pelabelan rating dengan FMCDM TOPSIS. Sedangkan dengan MCDM TOPSIS menghasilkan akurasi sebesar 97,5% dengan sensitivity serta specificity masing-masing sebesar 1. ==================================================================================================== Internal fraud in a company has been one of the most significant cause of company loss. This fact pushed company to have regulacy about strong security and information system on fraud detection. This final project is focused on fraud process business based. There are ten type of anomaly, those are Skip Sequence, Skip Decision, Throughput Time Maximum, Throughput Time Minimum, Wrong Pattern, Wrong Decision, Wrong Duty Decision, Wrong Duty Sequence, Wrong Duty Combine and Wrong Resource. Ontology modelling on Standard Operational Procedure and event logs is used to get anomalous cases. The creation of SWRL rules is used to get the anomalous cases. Fuzzy Multi Criteria Decision Making TOPSIS is used to calculate the anomalous rate of case. And Decision Tree is used to classify the status of anomaly. The goal is to get the data of deviation behaviour or the violation with the confidence value. In this final project, credit application is used for testing purpose. The test resulting an acurate anomaly detection proven with 98% rate of accuracy, specificity rate of 1, and sensitivity rate of 0,71 for rating labelling with FMCDM TOPSIS. While with MCDM TOPSIS resulting 97,5% rate of accuracy, each sensitivity and specificity is 1.

Item Type: Thesis (Undergraduate)
Additional Information: RSIf 658.409 801 1 Kun d
Uncontrolled Keywords: anomali, Decision Tree, event logs, model ontologi, TOPSIS
Subjects: Q Science > QA Mathematics > QA9.64 Fuzzy logic
Divisions: Faculty of Information Technology > Informatics Engineering > (S1) Undergraduate Theses
Depositing User: - Taufiq Rahmanu
Date Deposited: 26 Nov 2019 07:43
Last Modified: 26 Nov 2019 07:43
URI: http://repository.its.ac.id/id/eprint/72054

Actions (login required)

View Item View Item