IMPACT OF FEATURE STANDARDIZATION ON HEART DISEASE PREDICTION: A COMPARATIVE ANALYSIS OF LOGISTIC REGRESSION AND SUPPORT VECTOR MACHINE MODELS

Cardiovascular diseases are among the leading causes of global mortality. Heart disease, in particular, remains a major contributor to this burden, highlighting the need for effective predictive models to enable early detection. This study investigates the impact of feature standardization using StandardScaler on the performance of two prominent machine learning models involving Logistic Regression (LR) and Support Vector Machine (SVM) for predicting heart disease. The research utilizes a dataset comprising demographic and clinical attributes of patients, focusing on the role of feature standardization in enhancing model performance. The study compares models trained on raw data and standardized data, applying performance metrics such as accuracy, precision, recall, and F1-score. Results indicate that feature standardization significantly improves the performance of both models. LR showed a clear enhancement in macro F1-score on the testing set, rising from 0.82 without standardization to 0.87 with standardization. SVM was slightly superior in its raw form but still improved after standardization, with the macro F1-score increasing from 0.85 to 0.86. These findings highlight the importance of data pre-processing and demonstrate how feature scaling can optimize machine learning models for heart disease prediction. This research contributes to the growing field of predictive healthcare, offering valuable insights for clinicians seeking reliable early detection tools for cardiovascular conditions.