While global machine learning models provide strong baseline predictions for retail pricing, they often fail to capture segment-specific demand patterns that influence optimal pricing. This research extends prior work by introducing segmentation using K-Means clustering, followed by segment-specific Random Forest models. Using a retail transaction dataset, three distinct market segments were identified: Premium (high-price, low-volume), Volume (low-price, high-volume), and Balanced (moderate price and demand). The segmented approach achieved improved predictive performance, with the Volume segment showing the highest accuracy gain compared to the global model. These findings demonstrate the effectiveness of segmentation-based modeling in enhancing pricing accuracy. The proposed approach enables retailers to implement personalized pricing strategies that maximize revenue while maintaining competitiveness across diverse customer behaviors.