Applying Supervised Machine Learning Algorithms for Fraud Detection in Anti-Money Laundering

Petcu Vlad-Gavril

Tircomnicu Vlad-Mihai

    Introduction

    Money laundering dates back to early twentieth century which is regarded a gangster era in American history. Gambling and alcohol sales were all increasing at the time, and people were earning a huge amount of cash that required to be laundered so the authorities would not know where their money came from.

    Money laundering may be broken into three parts, which may be referred to as placement, layering, and integration.

    Data Set

    Features are given as follows: CASH-IN, CASH-OUT, DEBIT, PAYMENT, and TRANSFER are the different types of transactions. 

    - amount - the transaction's value in local currency. 

    - nameOrigin - is the name of the client who initiated the transaction.

    - oldbalanceOrg - before to the transaction, the original balance 

    - newbalanceOrig - following the transaction, the new balance 

    - nameDest - the customer who is the transaction's beneficiary. 

    - oldbalanceDest - the receiver of the original balance before the transaction.

    - isFraud - This represents the fraudulent agents' transactions inside the simulation. The fraudulent activity of the agents in this dataset tries to profit by seizing control of clients' accounts and attempting to empty the money by transferring to another account and then cashing out of the system. 

    - isFlaggedFraud - The business model strives to regulate large transfers from one account to another and flags any efforts that seem to be fraudulent. An attempt to transfer more than 200.000 in a single transaction is considered unlawful in this dataset.

    Solution

    Machine learning technologies can be a helpful contribution to the existing challenges of money laundering detection. Fresh approaches can provide new insights and help detect questionable transactions more precisely.

    

    Conclusion

    To summarize this research, machine learning models might help with money laundering detection by effectively identifying money laundering transactions from routine transactions. Furthermore, this research findings suggest that Random Forest might help in real-world anti-money laundering detecting settings. Because of its accuracy and interpretability, it may be utilized in Anti-Money Laundering (AML) architecture in financial institutions.

    Bibliography

    1. Applying Supervised Machine Learning Algorithms for Fraud Detection in Anti-Money Laundering