An Ensemble Methods for Medical Insurance Costs Prediction Task

The paper reports three new ensembles of supervised learning predictors for managing medical insurance costs. open dataset is used data analysis methods development. usage artificial intelligence in the management financial risks will facilitate economic wear time and money protect patients’ health. Machine associated with many expectations, but its quality determined by choosing a good algorithm proper steps to plan, develop, implement model. aims develop individual costs prediction provide high accuracy. Pierson coefficient Boruta are feature selection. boosting, stacking, bagging built. A comparison existing machine algorithms given. Boosting modes based on regression tree stochastic gradient descent Bagged CART Random Forest proposed. boosting stacking shown better accuracy than bagging. tuning parameters do not allow decrease RMSE too. So, shows weakness generalizing prediction. developed using K Nearest Neighbors (KNN), Support Vector (SVM), Regression Tree, Linear Regression, Stochastic Gradient Boosting. random forest (RF) combine predictions. One hundred trees built RF. Root Mean Square Error (RMSE) has lifted 3173.213 other predictors. ensemble Squared metric 1.47 best weak predictor (SVR).