<![CDATA[Cardiovascular disorders remain a major cause of death worldwide, emphasizing the importance of accurate and timely diagnostic support. This study presents a data-driven framework for predicting heart disease by integrating structured data cleaning procedures, exploratory data analysis (EDA), and supervised machine learning classification. The Heart Statlog (Cleveland-Hungary) dataset, which contains a range of clinical attributes associated with cardiac conditions, is used as the primary data source. To ensure data quality, several preprocessing steps are applied, including treatment of missing values, elimination of duplicate records, correction of inconsistent entries, and transformation of categorical variables into numerical representations. EDA is subsequently employed to explore feature distributions and inter-variable relationships using statistical measures and visualization techniques. Logistic Regression and Random Forest algorithms are implemented to construct predictive models and assess classification performance. The experimental results indicate that chest pain category, ST-segment slope, exercise-induced angina, and maximum heart rate are among the most influential predictors of heart disease. Furthermore, the Random Forest classifier achieves higher overall performance than Logistic Regression, suggesting its stronger capability in modeling complex clinical patterns. These findings confirm that rigorous preprocessing combined with appropriate machine learning methods can significantly enhance predictive accuracy and support the development of reliable clinical decision-support systems.]]>
Copyrights © 2026
