Enhancing Integrity of Toll Gates:Fastag Fraud Detection

The FastTag Fraud Detection System employs a machine learning model to identify fraudulent activities in FastTag transactions. Key features such as 'Transaction_Amount,' 'Amount_paid,' 'Vehicle_Type,' 'Lane_Type,' and'Geographical_Location' are used to differentiate betweenlegitimate and potentially fraudulent transactions. The modelconsiders various classifiers including Stochastic Gradient Descent (SGD), K-Nearest Neighbors (KNN), XGBoost, Logistic Regression, and Support Vector Machines (SVM). The SGD Classifier emerges as the most effective, demonstrating high accuracy, perfect precision, and abalanced recall-precision ratio. The model building process involves encoding categorical features, splitting the dataset, and training and evaluating multiple classifiers. Test accuracyfor models like Logistic Regression was 94.5%; test accuracyfor SGD Classifier was 98.0%; test accuracy for Gradient Boosting Classifier was 97.2%; test accuracy for SVC was 97.0%; and test accuracy for KNeighbors Classifier was 97.9%. These evaluation results demonstrated the efficacy ofthe FastTag Fraud Detection System. Strong precision, recall,and F1- score metrics were displayed by these models, demonstrating their capacity to precisely identify fraudulent transactions and improve the security of electronic toll collection systems.