Building a Prediction System of Influenza Epidemics with LASSO Regression Model and Baidu Search Query Data / 中国卫生统计

ABSTRACT

Objective To evaluate the performance of a prediction system built with LASSO regression model and Baidu search query data.Methods Based on a strategy using a combination of Bagging and multi-measure optimization method,this study proposed an ensemble LASSO regression model which had an obviously improved performance,and applied it to predict the epidemics of influenza in China.Results The results showed that the improved model had significantly smaller prediction error rates than that of the conventional LASSO regression model for influenza cases during the study period of 2011-2015.This study designed an open source R package,SparseLearner,which was conveniently used and further developed.Conclusion The combination of Bagging and multi-measure optimization method is an efficient strategy to improve the performance of LASSO regression model.The proposed ensemble LASSO regression model in this study can be applied for the prediction of infectious diseases epidemics.