Published Paper Details:

The study mostly focuses on the use of machine learning techniques to find fraudulent behavior in financial facts. This is the main challenge in the financial sector, where it is important to recognize and prevent fraud. Images are hyperparameters of tuning class as a method for increasing fraud detection. These settings improve fraud detection system by helping the version of the extra precisely distinguish between real and fraudulent transactions. The work deliberately uses three machine learning techniques: XGBoost, LightGBM, and CatBoost. Every method has sure advantages; their combined use seeks to improve the general fraud detection method performance. The research includes deep learning algorithms to adapt to hyperparameters. This connection improves the effectiveness and adaptability of fraud detection systems, and increases the efficiency of identifying modified fraud strategies. The effort employs actual data to conduct comprehensive analyses. The findings indicate that Lightgbm and XGBOOST outperformed the contemporary method when assessing numerous factors. This suggests that, among other strategies, the suggested one is more a success in spotting fraudulent behaviour. It incorporates a Stacking Classifier, which combines with precise parameters Random forest and LightGBM classifier predictions. Through using the strengths of numerous models, this ensemble technique improves prediction accuracy by means of a GradientBoostingClassifier as the final estimator.