Suggestion of mixed classification algorithm in diagnosis of diabetes with data mining and developing a decision support system

Diabetes has been causing many damages in the human body in the long term. It is predicted that 463 million people in the world have diabetes. 50,1% of the patients with diabetes continue their lives unaware of their disease. Considering that there is no treatment that can completely cure diabetic patients and there are many people unaware of the disease, it can be said that early diagnosis is quite important. In this study classification algorithms which is one of the data mining techniques were used for the diagnosis of disease. Feature selection was made in order to select the variables that were best representative the data set. The success rates of classification algorithms with the selected features were tested in the WEKA. A mixed method was proposed by using Naive Bayes, artificial neural network and decision tree together, which are among the algorithms that give the best results. It was observed that the classification success of the mixed method, which was created by considering the estimation results of the majority of the classification algorithms, was higher than the individual classification success of the algorithms. A visual interface was designed in Visual Studio to facilitate data entry with C# programming language. With the interface created, the user physicians were enabled to see the results they wanted to learn without any complexity.