Increasing healthcare costs have motivated researchers to seek ways to more efficiently use medical resources. The aim of our study was to adopt the explanatory data-mining approach to identify characteristics of emergency department (ED) visits for ED management. To that end, we adopted a behavior-based decision tree (DT) induction method that considers medical diagnoses and individual patients’ information, i.e., 11 input variables, in order to analyze characteristics of patients’ visits to EDs and predict the length of the stays. We interpreted the results based on the communicability and consistency of the DT, represented as a behavior-based DT profile in order to increase its explanatory power. Among the major preliminary findings, the DT with International Classification of Diseases diagnosis codes achieved better clinical values for explaining the characteristics of patients’ visits. Our results can serve as a reference for ED personnel to examine overcrowding conditions as part of medical management.