Coffee Quality Prediction: A Comparative Analysis of Machine Learning Techniques Using CQI Data in Sensory Score Estimation

Coffee quality assessment plays a critical role in determining market value, directly influencing pricing and income for producers.However, traditional evaluation methods are often subjective and inconsistent.This study investigates the use of machine learning techniques to predict coffee quality scores represented as total cup points based on objective sensory and physical attributes such as flavor, acidity, aroma, aftertaste, balance, body, and overall impression.Using manually collected data from the Coffee Quality Institute (CQI), we conducted comprehensive preprocessing and exploratory data analysis to identify trends and relevant patterns.Feature importance analysis revealed that flavor was the most influential factor in predicting coffee quality, followed by category one defects and overall.Machine learning techniques including random forest, multiple linear regression (MLR), support vector machine (SVM), and decision tree were trained and evaluated using four performance metrics: mean absolute error (MAE), mean squared error (MSE), root mean squared error (RMSE), and R-squared (R).Random forest achieved the best performance with MAE of 0.1598 0.0712 and R of 0.8242 0.1907, followed by multiple linear regression (MAE: 0.1712 0.0522, R: 0.8149 0.1829), support vector machine (MAE: 0.1623 0.0776, R: 0.7951 0.2334), and decision tree (MAE: 0.2469 0.0745, R: 0.6944 0.1719).These findings demonstrate the effectiveness of machine learning in producing reliable, data-driven assessments of coffee quality.The implementation of such models can support more consistent grading practices, reduce human bias, and enhance transparency across the coffee supply chain particularly beneficial in markets where specialty coffee commands premium prices.