Ανίχνευση και πρόβλεψη απάτης στο λιανικό εμπόριο μέσω της χρήσης αλγορίθμων μηχανικής και βαθιάς μάθησης με στόχο τη βελτίωση της εμπειρίας του πελάτη
Fraud detection in retail transactions through the use of machine and deep learning algorithms in order to improve the customer experience
View/
Keywords
Ανίχνευση - Πρόβλεψη απάτης ; Λιανικό εμπόριο ; Εξόρυξη δεδομένων ; Αλγόριθμοι μηχανικής μάθησης ; ΚατηγοριοποίησηAbstract
In recent years, due to the rapid development of modern technology and advanced communication
methods, fraud cases are constantly increasing at an alarming rate, resulting in huge financial
losses around the world. Fraud has become a global phenomenon of concern to the business sector
and affects all types of businesses regardless of size, profitability or industry. That's why modern
businesses are called upon to detect and anticipate the risk of fraud in any form in order to secure
their revenue and maintain their credibility. Detection of fraud is a process of detecting malicious
actions and practices.
Machine Learning and Data Mining techniques are crucial in detecting and anticipating fraud and
have been successfully applied to detect illegal activities such as credit card transactions, online
commerce, money laundering, and fraudulent activities in industries such as insurance sector, the
telecommunications sector as well as the medical and scientific sector. In particular, transaction
fraud detection is a process of analysis and processing of large volumes of data as well as the
creation of predictive models, through the application of Machine Learning algorithms, which
leads to the detection of suspicious transactions. This deep learning technology recognizes and
learns from complex patterns and combining important user transaction data from different sales
channels, has the ability to classify whether a transaction is illegal or legal.
The aim of this master thesis is to detect and predict possible fraud in transaction data originating
from the retail sector, through data mining techniques and the use of machine and deep learning
algorithms.