Modeling and Identification of Factors Affecting the Existence of Complications Caused by Diabetes with Data Mining Methods

In Iran, with the advancement of technology and the development of registration statistics, the need to use data mining methods has attracted more attention from researchers. Regression and classification tree is one of the important methods in Big data modeling, which has attracted the attention of many researchers for community control and prediction. The purpose of this study is to determine the influencing variables on the occurrence of complications caused by diabetes.

This paper is a cross sectional-analytical study. In this research, all diabetic patients covered by Mashhad University of Medical Sciences in 2017 were extracted from the SINA system. The number of diabetics with complications was 5016 and diabetics without complications were 53613. The method of fitting the regression tree model and classification and measurement criteria of the model is the coefficient of determination and the area of the Rock curve and the Lift diagram.

The rock curve for the fitted tree model is 73.8%, which shows the relatively high power of the model. Based on the Lift chart, the decision-making power of diabetes complications increases 3.5 times for the person who comes to visit.

The results of the regression model and tree classification showed that, in descending order, age, risk assessment factor, FBS, HbA1C, total activity time, cholesterol, FBS and HDL, cardiovascular disease, history of stroke, blood pressure, cholesterol Statin prescription, job with hard physical activity, living area, consumed oil, walking, consumption of vegetables and gender are more effective than other factors in the occurrence of diabetes complications.