Regression algorithms for prediction of vine yield and assessment of wine quality from vineyard of rural Karnataka

Quality evaluation and yield prediction play critical roles in the wine industry's increasing adoption of technology to boost sales and productivity. The primary objective of this research is to explore and compare the efficacy of various regression algorithms in predicting wine yield and assessing wine quality. The application of various Machine Learning (ML) models would aid the vine growers in saving time and improvising the quality of final product at the initial stages of growth itself. Utilizing real-time analytical data from Grover Zampa Vineyards, this study aims to demonstrate a practical application of ML models in the wine industry. Systematic approach was employed to analyze wine yield data using 11 multiple regression algorithms. User-friendly interface was developed using Streamlit software to allow users to input specific parameters and receive predictions on wine quality. Gradient Boosting Regressor achieved an MSE of 1.69, an R-squared value of 0.75, and MAE of 0.97. Random Forest algorithm achieved an MSE of 1.94, an R-squared value of 0.71, and an MAE of 1.11. These findings highlight the potential of ensemble learning techniques in accurately predicting yield and assessing wine quality, thereby providing a robust tool for the wine industry to enhance product quality and market competitiveness.