A Data Mining based Target Regression-Oriented Approach to Modelling of Health Insurance Claims

Machine learning or Data mining algorithms can be used for predicting future management and thus treated as powerful tools. In recent days, data mining has become very important for gaining vital information in healthcare industries. The Health care insurance cost plays a vital role in developing medical facilities. To provide better medical facilities, it is very essential to forecast the cost of medical insurance which is one of the possibilities to enhance medical facilities. The paper deals with predicting the cost of the health insurance which has to be paid by the patient. Here various data mining regression algorithms such as decision tree, random forest, polynomial regression and linear regression are implemented to achieve the best prediction analysis. A comparison has been done between the actual and predicted expenses of the prediction premium and eventually, a graph has been plotted on this basis which will enlighten us to choose the best-suited regression algorithm for the insurance policy prediction. After the execution of these regression algorithms for prediction, correctness has been measured by the Coefficient of determination (r2_score), Root Mean Squared Error (RMSE) and Mean Squared Error (MSE) of each algorithm to check for the best-suited algorithm. Random Forest Regression was the best algorithm with an r2_score of 0.862533 which can be used in its best possible way for the correct prediction of the health insurance cost.