Evaluation of hybridized homogenous supervised learning schemes in creditcard fraud detection. Oladimeji Ismaila W., Adeleye Falohun S, Oluyinka Omotosho I. Department of Computer Science, LadokeAkintola University of Technology, Ogbomoso, Nigeria Online published on 26 August, 2023. Abstract Fraud detection is a way of finding patterns in data that do not conform to expected behaviour. Fraud detection finds extensive use in a wide variety of applications such as fraud detection for credit cards, intrusion detection for cyber-security, military surveillance for enemy activities. As credit card being the primary method of payment in online transactions, credit card frauds have also been observed to surge as the number of online transactions have increased. The credit card industry has studied computing models for automated detection systems which have now been the subjects of academic research. This paper evaluates an ensemble homogenous supervised learning system (EHSLS) to detect fraud in credit cards online transactions. Random Forest is a type of supervised machine learning algorithm based on ensemble learning which is used for regression and classification tasks and well suited when the class distribution is unbalanced. The work flow of the proposed fraud detection system includes data preparation phase, implementation phase and evaluation phase. Cross validation technique was used for training and testing. The results showed the EHSLS produced 89.47%, 88.83%, 96.80%; CD-CPNN produced 93.80%, 91.70%, 95.13%; RBFN-PSO generated, 93.9%, 95.1% 91.7% while CPNN-GA gave 96.89%, 93.75%, 97.30% for recall , precision accuracy, respectively. However, the system developed produced the least f-measure value of 89.15%. Top Keywords Fraud detection, Credit cards, Cross validation, Random forest, F-measure. Top |