Hasta sağlığı veri tabanı kullanarak veri madenciliğine dayalı kalp hastalığı tahmini

Data mining (DM) is the process of finding or extracting knowledge from information on a huge piece data. DM uses intelligent methods to find patterns in the process of knowledge discovery (KD) in a database. The appearance field of DM promises to give a new technique and good tools. Also, DM can help the person to understand, solve big amounts of data remains on complex and unsolved problem. The wide functions in DM practice includes: classification, clustering, regression rule generation, sequence analysis and discovering association. The classification is one of the most important techniques of DM. As well as, many problems in various fields such as science, business, industry and medicine can be solved by using these approaches. Neural Networks (NN) have appeared as a good tool for classification. The study of Heart Diseases (HD) database is testing by using NN approach. HD diagnosis is not easy work which demands to a lot of experience and acquaintance. The common way for predicting HD is a doctor's checkup or different medical examination like ECG, Heart MRI Stress Test and etc. Nowadays, 'Artificial Neural Network' (ANN) has been commonly used to the technique for dissolving many problem clinical diagnoses. An ANN is the 'simulation of the human brain', it is a supervised learning. This research aims to optimize or reduce the number of biomedical test which asked from patients. Correspondingly to do a classification approach using NN technique and a Feature Subset Selection (FSS) algorithm. FSS is a pre-processing phase used to reduce number of attribute and remove irrelevant data. HD values are used and originally 13 attributes are involved to classify the HD. To reduce or optimize the number of attributes, different evaluators and search methods are determined. In this research the two studies are conducted on the STALOG data set. The first study used three algorithms: Naïve Bayes (NB) got on accuracy equal to 85.182. While J84 obtained on 91.4815 and NN algorithm got on 99.6296. However, in the second study the NB obtained on 85.925 and J84 got on 91.4815. Finally, NN obtained on 99.2593. Moreover, accuracy of ANN algorithm in the two studies got on the best result when compared with the results of the other algorithms.