Comparison of Novel Optimized Random Forest Technique and Logistic Regression for Credit Card Fraud Detection with Improved Precision
DOI:
https://doi.org/10.47750/pnr.2022.13.S04.082Keywords:
Machine Learning, Novel Optimized Random Forest, Fraud Detection, Logistic Regression, Credit bureau, Credit Card Holders.Abstract
Aim: Objective is to improve precision for credit card fraud detection by using Novel Optimized Random Forest Technique (NORFT) and comparison with Logistic Regression (LT).
Materials and Methods: In NORFT, it uses multiple Decision Trees to detect the credit card fraud by culminating the maximum attained probability values. The groups consist of NORFT and LR for comparison analysis. The sample size was estimated by using Clinicalc online tool, which is determined as N=2500 for each group with g-power value as 80% and datasets are collected from various web sources with recent study findings and threshold 0.05%, confidence interval 95% mean and standard deviation..
Results: The implementation resulted in precision as such NORFT (92.52%) and LR (71.60%). The statistical significance was performed using Independent Sample T-test between the groups, the study has a significance value of (p>0.05) i.e. p=0.649 and states does not have any difference in research.
Conclusion: Novel Optimized Random Forest Algorithm (NORFT) has significantly finer precision than Logistic Regression Algorithm (LR).
