Analyzing Features and Model for Optimizing Accuracy of Vineyard Yield Prediction

Abstract

Vineyard yield prediction is important for optimizing vineyard management and decision-making. In this project, we developed accurate prediction models using climate data, including temperature, pressure, humidity, clouds, wind speed, wind direction, rain- fall, and historical harvest records. We evaluated the performance of different machine learning models for vineyard yield prediction.We trained and tested several widely-used machine learning models, including Random Forest Regression, MLPRegressor, Gradient Boosting, XGBoost, and Linear Regression. The models were trained on the provided dataset, and their performance was evaluated using various metrics, such as R-squared score, mean absolute error, explained variance score, and mean squared error.The best performing model was a random forest regression model, which was able to explain 58% of the variance in the data. The model had a mean absolute error of 0.18 and a mean squared error of 0.0705. These results suggest that random forest regression is a promis- ing model for vineyard yield prediction.