Coronary heart disease occurs when atheroclerosis inhibits blood flow to the heart muscle in the coronary arteries. This disease is often the cause of human death. The method for diagnosing coronary heart disease that is often a doctor's referral is coronary angiography, but it is invasive, expensive, and high-risk. This study aims to analyze the effect of k-Fold Cross-Validation (CV) on the dataset to create features based on the rules used to diagnose coronary heart disease. This study uses the Cleveland heart disease dataset, where feature selection is performed using a medical expert-based method (MFS) and a computer-based method, Variable Precision Rough Set (VPRS). Evaluation of the classification performance using the k-fold 10-fold, 5-fold and 3-fold methods. The results showed the number of different attribute selection results in each fold, both for the VPRS and MFS methods, with the highest accuracy score in the VPRS method 76.34% with k = 5, while the MTF accuracy was 71.281% with k = 3.