Optimization of Gradient Boosting Model for Wine Quality Evaluation

Abstract

In wine industry, evaluating wine quality and improving the prediction accuracy for quality are important. The aim of this study is to evaluate wine quality based on physical and chemical features. Gradient boosting model is applied on a dataset with both red wine and white wine samples to make predictions. An optimization of gradient boosting model is implemented by using a proposed tuning method for important parameters. Each parameter is tuned to search for its best value, and also multiple variables are adjusted at the same time to find out a combination of parameters that can make the accuracy of model the highest. By using the optimization method for gradient boosting model, the model can achieve a relatively high accuracy on predicting wine quality under the condition that no error is allowed.