Insurance Fraud Prediction: An analysis Using Supervised and Unsupervised Machine Learning

The insurance sector is becoming increasingly concerned about insurance fraud, which causes losses totaling billions of dollars every year. One interesting way for identifying fraudulent activity has been suggested: the use of supervised and unsupervised machine learning techniques. In this research, we use various supervised and unsupervised machine learning techniques to give an extensive examination of insurance fraud prediction. Support vector machines, logistic regression, decision trees, Naïve Bayes, K-means clustering, hierarchical clustering, and anomaly detection are the techniques we use to present our study. We go over the rationale behind each method's choice, how it was applied to the dataset, and the results it produced in terms of accuracy and performance. We also share observations and suggestions based on our study. In conclusion, our findings demonstrate that the best strategies for fraud detection in insurance businesses are supervised machine learning techniques, particularly Support Vector Machines and Logistic Regression.