Final Year Project
Thesis Title: Suspicious Online Car Advertisement Detector using Principal Component Analysis (PCA)
E-commerce is an online site that acts as a medium for sellers and buyers exchanging goods and services. Despite of the goods of online site, there are also some risks and disadvantage.
Some sellers take buyers’ trust for granted (Hoon, 2018) and cheat by using fake product specification and/or advertise items they do not own (Sankhwar & Pandey, 2017). Thus, sellers could get money without giving item to the buyers (Covallaro et al., 2017)
Problem Statement
Problems (cyber crimes, scams in online shopping)
Currently, Machine Learning and Data Mining techniques have essentially been used for detecting other cybercrimes such as phishing websites or emails and identity deception
Tool (Machine Learning & Data Mining)
Example real case study that use Machine Learning and Data Mining:
Phishing websites or emails (Falcon et al., 2016 ;Gokulan et al., 2017; Babagoli et al., 2018)
Identity deception (Mohamed et al., 2017; Bhattacharyya et al., 2018; Chandralekha & Shanthi, 2018; van Der Walt & Eloff, 2018)
Credit Card Fraud (Kumar Jayasingh & Kumar Swain, 2011)
Solution (Principal Component Analysis)
Example real case study (field) that use Principal Component Analysis (PCA):
Engineering (Magd et al., 2016)
Health (C. Brenneman & Maly, 2017; Allgar et al., 2018; Kumar & Bhadauria, 2018; Morais & Lima, 2018)
Geological and earth sciences (Ghanem et al., 2017; Ceferino et al., 2018)
Industry (Cheng et al., 2016; Coussement et al., 2017; Cizek et al., 2018; Gajjar et al., 2018)
Hypothesis
The PCA could be used to detect any outlier from two dimensional spaces plotted in a scatter plot.
The outliers could be scammer advertisements because their abnormal distance from other values.
Objective
To develop a web-based application for detecting suspicious car advertisements in online shopping site.
To analyse and prove the suspicious car advertisements using Principal Component Analysis (PCA) and cluster analysis.
Project Scope
Shopping website: mudah.my
Advertisement: Vehicles (car)
Limitation: Data scraped may not meet requirements of the study
Algorithm: Principal Component Analysis (PCA), Cluster Analysis
System: Developed by using Dash framework
Project Significance
For Buyer: As a guideline to buy in mudah.my
For Seller: Will be more aware and honest in advertise and sell their products
