The Old-age Care Demand Prediction Based On Random Forests

After China entered aging society,the growth of aging population is still accelerating.Old-age care demand prediction is urgently needed to realize refined management for the intelligent pension system.In order to predict old-age demand accurately,variance analysis,feature selection and classification model based on random forests methods are used to analysis nearly 20,000 old people's data in CHARLS questionnaire.After analyzing the significance of various factors on the elderly living preferences and establishing the prediction model for the elderly living preference,finally,the old-age care prediction system is realized.Besides,SSOOB algorithm based on stratified sampling is proposed to improve the accuracy and stability of random forests feature selection algorithm on multi-dimension and imbalanced datasets.Experimental results indicate that SSOOB algorithm can maintain the accuracy and stability of feature selection,and the elderly living preference can be predicted well through the classification model.Importantly,old-age care demand prediction system can provide multi-level prediction even on the community-level,which has the potential function to provide reference on accurate management for the elderly care related departments.