Fuji Apple Quality Analyzer: Predicting Good and Bad Quality of Malus Domestica (Fuji Apple) using Machine Learning Algorithm

Fuji apples are appreciated for their sweet, crisp texture and widely consumed. This study developed a machine learning model to accurately categorize Fuji apples as good or bad quality. Multiple models were trained and tested on a dataset of 4000 Fuji apples with features like sweetness, crunchiness, ripeness, size, weight, juiciness and acidity. These features underwent feature selection processes to test its significance on predicting the output such as Univariate Selection, Recursive Feature Elimination, Feature Importance and Multivariate Feature Selection. The study compares the performance of various classification algorithms such as the Support Vector Classifier model which performed the best, achieving 89% accuracy and other top performers were K-Nearest Neighbors (87.63% accuracy) and Random Forest Classifier (87.40% accuracy). This study developed a hardware device that can classify Fuji Apple as good and bad quality. With additional development, such models could enable objective apple quality assessment to benefit producers, distributors and consumers.