Machine Learning in Fine Wine Price Prediction

• DOI: 10.1017/jwe.2015.17
• Publication Date: 2015-08-03T05:33:59Z
• Authors: Michelle Yeo; Tristan Fletcher; John Shawe-Taylor
• Keywords: 0502 economics and business; 05 social sciences

Abstract Advanced machine learning techniques like Gaussian process regression and multi-task are novel in the area of wine price prediction; previous research this being restricted to parametric linear models when predicting prices. Using historical data 100 wines Liv-Ex index, main contributions paper field are, firstly, a clustering into two distinct clusters based on autocorrelation. Secondly, an implementation these with predictive accuracy surpassing both trivial simple ARMA GARCH time series prediction benchmarks. Lastly, algorithm which performs feature kernels returns as extension our optimal model. covariance kernel from regression, we achieve results comparable that regression. Altogether, suggests there is potential using advanced prediction. (JEL Classifications: C6, G12)