Feature Selection Using Weight Methods in Cardiovascular Disease Dataset

This study investigates the significance of clinical features in predicting cardiovascular disease through feature weighting using six selection methods: Information Gain, Gain Ratio, By Rule, Gini Index, Support Vector Machine (SVM), and Principal Component Analysis (PCA). The analysis reveals notable variations in feature importance across methods, reflecting the strengths and limitations of each approach. Among the six evaluated features—age, body mass index (BMI), systolic blood pressure, diastolic blood pressure, cholesterol, and blood sugar—age consistently emerges as the most influential predictor, achieving the highest scores in SVM (1.33) and By Rule (0.91), and demonstrating strong relevance in Information Gain (0.49) and Gain Ratio (0.59). In contrast, BMI exhibits inconsistent importance, with moderate scores in By Rule (0.67) and Gain Ratio (0.28), but negligible or negative values in SVM (–0.11) and PCA (0.00), indicating method-dependent relevance. Systolic blood pressure shows a moderate and stable influence, while diastolic blood pressure contributes minimally across most methods. Cholesterol is particularly significant in PCA (0.94), suggesting its importance in multivariate contexts, despite low scores in other methods. Blood sugar demonstrates moderate relevance, with its highest scores in By Rule (0.67) and PCA (0.30). Overall, the results highlight age and cholesterol as the most consistent and influential features, while other attributes show varying levels of importance depending on the analytical technique applied.