Enhancing Orange Quality Classification via PCA-Based Feature Augmentation and Ensemble Learning

Accurate and efficient quality assurance is crucial for the orange industry's market competitiveness and customer satisfaction. This study aims to enhance the accuracy of automated orange quality assessment by integrating feature engineering techniques into machine learning pipelines. Instead of using Principal Component Analysis (PCA) for dimensionality reduction, we employ it to construct an alternative feature space and augment the original feature set. This enriched representation enhances the model's discriminative capability. Experimental results on a public dataset for orange quality evaluation demonstrate that the Extra Trees (ET) model achieved the highest accuracy of 75.51%, surpassing the previously best- performing K-Nearest Neighbors (KNN) model, which attained 69.38%. These findings highlight the effectiveness of incorporating additional representations derived from alternative feature spaces to improve classification performance. This work opens new directions for leveraging advanced machine learning techniques and feature enrichment strategies in quality control applications, particularly in agricultural product evaluation.