Abstract

According to recent WHO assessments, it is imperative for some countries. Health care plans to develop precise tools for early diagnosis of cardiac illnesses through effective routine heart examinations. The quantity and size of medical datasets are both growing quickly, and cutting-edge data mining tools may be able to assist doctors in reaching useful conclusions. The selection of characteristics, the quantity of samples, the balance of the samples, the lack of magnitude for some features, etc. are the limitations of heart disease data. This study is primarily concerned with improving feature selection and reducing the amount of features. It was discovered that feature selection algorithms were highly good at detecting cardiac problems. In this study, we examine the effects of feature selection techniques, such as information gain and correlation features, on the performance of logistic regression. We also suggest a unique model known as Tuned Logistic Regression. The outcome of the Tuned-experimental Logistic Regression is compared and contrasted with two feature selection algorithms Correlation Feature Selection and Information Gain. Performance metrics are utilised to assess Tuned Logistic Regression's strong performance. While comparing the accuracy and overall outcome with existing algorithms, the proposed method is shown to be superior to other strategies currently used in the industry