Classification of anemia using data mining methods: An application

Nowadays, as a result of evolving information technology, amount of collected and stored data is increasing day by day. Even if data and information gain importance, increasing volumes of data makes impossible to analyze the data with human perception. Although accessing the desired data in huge databases is easy by classical polling methods, understanding of the importance hidden information in data makes data mining applications necessary. Data mining aims to gain useful knowledge and hidden patterns from large amount of data using computer programs. Data mining converts data into knowledge and useful information. Thus, it is possible to acquire present or prudential meaningful knowledge thanks to data mining applications. This knowledge that gained from data has an important role in decision making process. The demand for data mining applications is increasing due to easiness of collecting and storing data, technological development in database management system, and gaining useful information from huge and complex data. As a result of understanding the importance of the conversion data into knowledge, data mining applications used in various areas such as telecommunication, banking, insurance, medical, and etc. Using data mining applications in medical area is crucial because of the huge amount and vital importance of data. Data mining applications use for various purposes in medical area such as recognizing early symptoms of chronic diseases, optimizing process of medical treatment, planning patient flow. In this respect, gaining valuable hidden information with data mining applications encourages effective treatment. The main purpose of our study is to classify anemia with various data mining applications and compares them to obtain optimal classification method. Our study is aimed to create data set with examining complete blood count tests that belonging to different types of anemia patients and hematological healthy individuals within the scope of Uludağ University Internal Medicine Division of Hematology Department. Different test groups are planned to be created by changing parameters of generated data set. Classification of anemia diseases will be obtained by applying various data mining applications such as Decision Tree, Neural Networks, Bayesian classification and K-nearest Neighbor. In this context, comparisons between methods in test groups as well as parameters in methods will be attained by changing parameters in four different methods applied in our research. In this way, developing a system which assisting the physicians about classifying anemia disease is planned by applying data mining methods to complete blood test results.