A Method for Filling Missing Values of Asthma Based on Improved Random Forest
Asthma data contains a large number of missing values,making it difficult to accurately predict asthma.When the existing random forest algorithm is applied to fill in the missing data of asthma,the correlation between medical characteristics is ig-nored in the pre-filling process,and the data cannot be updated in time during the data filling process,so that the data cannot be up-to-date.Aiming at the above problems,an improved random forest algorithm is proposed.In the pre-filling stage,Pearson corre-lation analysis is used to construct a more accurate regression equation,change the pre-filling method in the random forest algo-rithm,and construct a pre-filling matrix to improve the filling efficiency of the algorithm.In the first stage,the random forest algo-rithm is used to fill columns by column,and a cyclic update mechanism based on local data is added.When a column is filled,the parameters of the regression equation are updated,and all parameters in the pre-filled matrix are further updated to ensure data syn-chronization.Experiments show that the improved random forest algorithm has better filling effect than other algorithms,and can ef-fectively improve the accuracy of asthma diagnosis.
asthma diseaserandom forest algorithmmissing value completionprefilled matrixcyclic update mechanism
万方数据
维普
