What's the best way to refine my wine intelligence? Since nothing immediately comes to mind, I'll take the fallback algorithmic approach and brute-force a solution: sampling lots and lots of wine! So, let's determine the amount of wine I need to drink to achieve the same culinary genius as Robin Williams in Neverland!
After googling around for a few minutes, I was able to find a wine training set composed of a dozen wine features (acidity, pH, alcohol, etc) and a ranked quality. I then proceeded to classify this wine quality (at different splits: good/bad, good/okay/bad, etc) with a fancy-pants neural network. The converged prediction iteration count, after normalizing by the amount of wine it takes to blindfoldedly distinguish between red and white, is shown in the chart below:
Now, I only need to convince my wife that purchasing 7 Balthazars of pink bubbly wine is intended for research and my life will be fantastic.