Comparative Review of Credit Card Fraud Detection using Machine Learning and Concept Drift Techniques (original) (raw)
Related papers
Credit card fraud detection and concept-drift adaptation with delayed supervised information
2015 International Joint Conference on Neural Networks (IJCNN), 2015
Most fraud-detection systems (FDSs) monitor streams of credit card transactions by means of classifiers returning alerts for the riskiest payments. Fraud detection is notably a challenging problem because of concept drift (i.e. customers' habits evolve) and class unbalance (i.e. genuine transactions far outnumber frauds). Also, FDSs differ from conventional classification because, in a first phase, only a small set of supervised samples is provided by human investigators who have time to assess only a reduced number of alerts. Labels of the vast majority of transactions are made available only several days later, when customers have possibly reported unauthorized transactions. The delay in obtaining accurate labels and the interaction between alerts and supervised information have to be carefully taken into consideration when learning in a concept-drifting environment. In this paper we address a realistic fraud-detection setting and we show that investigator's feedbacks and delayed labels have to be handled separately. We design two FDSs on the basis of an ensemble and a sliding-window approach and we show that the winning strategy consists in training two separate classifiers (on feedbacks and delayed labels, respectively), and then aggregating the outcomes. Experiments on large dataset of real-world transactions show that the alert precision, which is the primary concern of investigators, can be substantially improved by the proposed approach.
Credit Card Fraud Detection: A Realistic Modeling and a Novel Learning Strategy
IEEE Transactions on Neural Networks and Learning Systems, 2017
Detecting frauds in credit card transactions is perhaps one of the best testbeds for computational intelligence algorithms. In fact, this problem involves a number of relevant challenges, namely: concept drift (customers habits evolve and fraudsters change their strategies over time), class imbalance (genuine transactions far outnumber frauds) and verification latency (only a small set of transactions are timely checked by investigators). However, the vast majority of learning algorithms that have been proposed for fraud detection, relies on assumptions that hardly hold in a real-world Fraud Detection System (FDS). This lack of realism concerns two main aspects: i) the way and timing with which supervised information is provided and ii) the measures used to assess fraud-detection performance. This paper has three major contributions. First we propose, with the help of our industrial partner, a formalization of the fraud-detection problem which realistically describes the operating conditions of FDSs that everyday analyze massive streams of credit card transactions. We also illustrate the most appropriate performance measures to be used for fraud-detection purposes. Second, we design and assess a novel learning strategy which effectively address class imbalance, concept drift and verification latency. Third, in our experiments we demonstrate the impact of class unbalance and concept drift in a real-world data stream containing more than 75 millions transactions, authorized over a time window of three years.
An adaptive profile based fraud detection framework for handling concept drift
2013
As e-commerce continues to grow, so does the opportunity for perpetrating online fraud. As a result many researches have been conducted to make online transactions possible in a risk free environment by proposing different fraud detection methods. Concept drift is an inherent feature in many data streams such as electronic financial transactions. Hence, many fraud detection techniques have tried to detect and preferably manage concept drift. In this paper, a new concept drift management framework has been proposed. In this framework a temporary profile has been introduced in order to retain new concepts in the incoming data stream independently from historical profile. When the historical profile reaches a different decision from the temporary profile this is an indication that most probably a concept drift has occurred. In this case, a window based method is applied as a strategy for managing concept drift. The ability to adapt normal profiles systematically makes this concept drift management framework applicable to any profile based fraud detection method. Simulation results indicate that the proposed scheme is able to reduce the false positives (FPs) of a typical fraud detection method to 4.3% on average in the presence of a wide variety of concept drifts in the incoming transactions. This is an average of 85.7% reduction in FPs for this fraud detection technique.
International Journal of Electrical and Computer Engineering (IJECE), 2023
In a streaming environment, data is continuously generated and processed in an ongoing manner, and it is necessary to detect fraudulent transactions quickly to prevent significant financial losses. Hence, this paper proposes a machine learning-based approach for detecting fraudulent transactions in a streaming environment, with a focus on addressing concept drift. The approach utilizes the extreme gradient boosting (XGBoost) algorithm. Additionally, the approach employs four algorithms for detecting continuous stream drift. To evaluate the effectiveness of the approach, two datasets are used: a credit card dataset and a Twitter dataset containing financial fraudrelated social media data. The approach is evaluated using cross-validation and the results demonstrate that it outperforms traditional machine learning models in terms of accuracy, precision, and recall, and is more robust to concept drift. The proposed approach can be utilized as a real-time fraud detection system in various industries, including finance, insurance, and e-commerce.
Credit Card Fraud Detection Using Machine Learning Algorithms
2021 Fifth International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud) (I-SMAC), 2021
Credit card frauds are easy and friendly targets. E-commerce and many other online sites have increased the online payment modes, increasing the risk for online frauds. Increase in fraud rates, researchers started using different machine learning methods to detect and analyse frauds in online transactions. The main aim of the paper is to design and develop a novel fraud detection method for Streaming Transaction Data, with an objective, to analyse the past transaction details of the customers and extract the behavioural patterns. Where cardholders are clustered into different groups based on their transaction amount. Then using sliding window strategy [1], to aggregate the transaction made by the cardholders from different groups so that the behavioural pattern of the groups can be extracted respectively. Later different classifiers [3],[5],[6],[8] are trained over the groups separately. And then the classifier with better rating score can be chosen to be one of the best methods to predict frauds. Thus, followed by a feedback mechanism to solve the problem of concept drift [1]. In this paper, we worked with European credit card fraud dataset.
Dataset Shift Quantification for Credit Card Fraud Detection
2019 IEEE Second International Conference on Artificial Intelligence and Knowledge Engineering (AIKE)
Machine learning and data mining techniques have been used extensively in order to detect credit card frauds. However purchase behaviour and fraudster strategies may change over time. This phenomenon is named dataset shift [1] or concept drift in the domain of fraud detection [2]. In this paper, we present a method to quantify day-by-day the dataset shift in our face-to-face credit card transactions dataset (card holder located in the shop). In practice, we classify the days against each other and measure the efficiency of the classification. The more efficient the classification, the more different the buying behaviour between two days, and vice versa. Therefore, we obtain a distance matrix characterizing the dataset shift. After an agglomerative clustering of the distance matrix, we observe that the dataset shift pattern matches the calendar events for this time period (holidays, weekends , etc). We then incorporate this dataset shift knowledge in the credit card fraud detection task as a new feature. This leads to a small improvement of the detection.
A Learning Based Approach for Credit Card Fraud Detection
In the recent years, Finance fraud is a serious problem with far consequences in the financial industry. Data mining is widely used and applied to finance databases to automate analysis of huge volumes of complex data. Fraud detection in credit card is also a data mining problem. It becomes challenging due to two major reasons-first, the profiles of normal and fraudulent behavior change frequently and secondly due to reason that credit card fraud data sets are highly skewed. Here we investigate and check the performance of Naïve Bayes and KNN on highly skewed credit card fraud data. Credit card transactions data is collected from European cardholders consists of 284,786 transactions. These machine learning techniques are applied on the raw and pre-processed data. The performances of the techniques are evaluated based on accuracy, sensitivity, specificity and precision. The results indicates the optimal accuracy.
A comparative study of credit card fraud detection using machine learning
AIP Conference Proceedings
Credit card fraud is one of the worldwide problems which a ects everyone. Credit card fraud detection is a distribution issue with the aim of automatically and adaptively categorizing genuine and fraudulent transactions. Any malicious behavior causing financial loss to the other party is classifying as fraud. For example, in poor nations, the use of digital currency, or even plastic money, is on the rise it has a history of defrauding people. They have a track record of scamming others. In recent years, credit card fraud has increased. Customers and institutions all over the world are paying billions of dollars. Fraudsters continue to thrive despite the multiple fraud-prevention devices in place. In this study report, we're seeking to come up with new ways to swindle people. As a result, combatting these scams demands the implementation of a sophisticated fraud detection system. Fraud is not only detected, but also prevented by the system. The systems must be able to learn from past fraud schemes and adapt to new ones. The notion of credit card fraud, as well as the many types of fraud, were discussed in this study. Several fraud detection methods, such as logistic regression, decision trees, and random forests, were studied. Existing and proposed theories for credit card fraud are carefully scrutinized, and these strategies are tested using quantitative metrics such as accuracy and discovery rate. The system showed a high level of fraud detection, equal classification, a high Matthews correlation coe cient, and a false alarm level. The crime of stealing sensitive information, supplanting, grazing or stealing data on the part of the merchant, lost or stolen cards, producing fake or counterfeit cards, making a real site, and removing or replacing a magnetic line on the card keeping user information are all examples of credit card fraud. The study came to the conclusion that existing models have flaws and proposes a new technique for fixing them. Obstacles to fraud detection are expected to change and metamorphosize into hidden impediments in the future, based on how fraudsters do these illicit behaviors.
Al-Kindi Center for Research and Development, 2024
Credit card fraud detection remains a significant challenge for financial institutions and consumers globally, prompting the adoption of advanced data analytics and machine learning techniques. In this study, we investigate the methodology and performance evaluation of various machine learning algorithms for credit card fraud detection, emphasizing data preprocessing techniques and model effectiveness. Through thorough dataset analysis and experimentation using cross-validation approaches, we assess the performance of logistic regression, decision trees, random forest classifiers, Naïve Bayes classifiers, K-nearest neighbors (KNN), and artificial neural networks (ANN-DL). Key performance metrics such as accuracy, sensitivity, specificity, and F1-score are compared to identify the most effective models for detecting fraudulent transactions. Additionally, we explore the impact of different folds in cross-validation on model performance, providing insights into the classifiers' robustness and stability. Our findings contribute to the ongoing efforts to develop efficient fraud detection systems, offering valuable insights for financial institutions and researchers striving to combat credit card fraud effectively.
MACHINE LEARNING BASED CREDIT CARD FRAUD ANALYSIS, MODELING, DETECTION AND DEPLOYMENT.
Credit card fraud is critical business risk that every stakeholder of financial system including issuer, acquirer etc. has to manage tightly to ensure business continuity and credibility of payment system. As the popularity of the credit card payment as one of the online payment mode is increasing more and more due to the revolution that has taken place in E-commerce sector. Traditional fraud models designed years back deliver near about 70% accuracy and were meeting business needs till this time. However fraudsters are increasing gaming the system to create new types of frauds which has resulted in consistent decrease in model accuracy. The fraudulent transactions and real transactions are scattered all around and there is very little difference to distinguish between them. Many techniques based on Artificial Intelligence, Data mining, Fuzzy logic, Sequence Alignment, Genetic Programming, Machine learning has evolved in detecting various credit card fraudulent transactions. This paper represents how the combinations of different clustering and machine learning algorithm which can best adapt to the changing scenarios taking place can be used and deployed on a very large scale to detect the fraudulent transactions and use to ensure the credibility of the payment system.