Abstract
To explore the correlation between the clinical indicators of patients with advanced lung cancer and lung infectious agent, 370 cases (including 222 patients with Candida, 148 patients with Klebsiella) which have 21 clinical indicators was subjected to data mining in this paper. We used random forest-recursive feature selection (RF-RFE) algorithm to select features, and Random Forest (RF), Support Vector Machine (SVM) and Decision Tree (DT) were used to classify the two infections respectively, and the corresponding classification error was estimated by 10 times 10 fold cross-validation. The result of the experiment shows that the classification model has high specificity with Klebsiella and RF has a better performance on classification classifying Candida and Klebsiella.