Prediction of health insurance fraud by using the Adaboost algorithm over random forest classifier with upgraded accuracy

The purpose of this study is to provide evidence in favour of the concept of combining the cutting-edge Adaboost algorithm with the Random Forest classifier in order to better achieve the desired level of accuracy. Methodologies and Instruments for Research: From the two groups, a total of twenty samples were gathered for research purposes. An advanced version of the Adaboost Algorithm is provided to Group 1. The Random Forest Classifier Algorithm is one of the algorithms that are included in Group 2 of algorithms. According to the findings that were received from Clincalc.com, the sample size for Group 1 is equal to ten, and the same number of people is included in Group 2’s sample. Using an insurance data set that includes 40,000 customer claims, this study employs the Adaboost Algorithm and the Random Forest Classifier Algorithm in order to improve the accuracy of the prediction of health insurance fraud (32,000 for training and 8,000 for testing). When compared to the Random Forest Classifier technique, the innovative Adaboost methodology gets a higher degree of accuracy (91.66 percent). This is the outcome of the comparison (89.83 percent ). The hypothesis was validated by the findings of the independent sample T test, which showed that it was correct. With a significance threshold of 0.019 (p0.05) and a T-test power of around 85 percent (G power setting parameters: =0.05 and power=0.85), we are able to determine that there are differences between the two procedures that are statistically significant. A detection accuracy of +/-2 standard deviations was found to be the average, according to the findings. In the end, the one-of-a-kind Adaboost approach trumps the Random Forest Classifier methodology, which gets an accuracy of 88 percent. The Adaboost method obtains an accuracy of 91.66 percent.