This document presents a heart disease prediction model using decision tree analysis. It selects 14 clinical features from patient data and develops prediction models using the J48 decision tree algorithm with unpruned, pruned, and pruned with reduced error pruning approaches. The results show that the pruned J48 decision tree with reduced error pruning has the highest accuracy at 75.73%, compared to 72.82% for unpruned and 73.79% for pruned. The reduced error pruning approach produces more compact decision rules with fewer extracted rules, improving predictive performance.