Analysis of the impact of Feature Selection methods in predicting cardiovascular disease

Heart disease is one of the most common diseases and the number of people with heart disease is currently increasing. However, if the patient is not cared for in a timely manner, it can lead to the patient's death. Therefore, accurate diagnosis at the initial examination stage with appropriate treatment can lead to preventing from an increase in mortality due to heart disease .To achieve this, the existing techniques in the field of data mining can be used. Data mining extracts useful data from existing datasets that lead to the prediction or categorization of information through clustering, classification, or the discovery of hidden patterns. In this study, we intend to present a feature-based approach and cumulative learning to classify heart disease data. The proposed solution has ۳ main steps. In this issue, each of the selection methods of the selected feature, which include Pearson correlation coefficient, Information Gain and PCA, will identify its effective features and based on these features, the prediction model will be obtained using the decision tree. In the results obtained using Pearson method, ۹۰% accuracy is obtained, which is the highest accuracy compared to other methods