From a LinkedIn wine discussion group:
We are currently developing a predictive model that can identify the overall quality of wine based on input variables such as: fixed acidity, volatile acidity, citric acid, residual sugar, chlorides, free sulfur dioxide, total sulfur dioxide, density, pH, alcohol content etc. The models submitted so far already have a prediction accuracy much higher than expectations…This accuracy is going to improve further. Wine producers, wine sourcing specialists and testers could directly benefit from this algorithm, to improve the quality of wine they deal with.
You knew, eventually, it was all going to come down to a spreadsheet, didn’t you?