Deep learning neural network to predict beer recommendations

Analyzing hidden node input weights to find predictive correlations between different beers

Visualization of correlations between different beers, with highly correlated beers positioned close to each other, and negatively correlated beers positioned far from each other, based on analysis of which beers are predictive or inversely predictive of user ratings of the other beers.

Beer preference map with ABV % (alcohol content) overlaid in blue, and IBU (a measure of bitterness from hops) overlaid in green: http://beerchooser.com/beermap/beergmapabvibu.html

Beer preference map with beer color overlaid: http://beerchooser.com/beermap/beergmapcolor.html

Beer preference map with specific gravity overlaid (a beer measure affected by the ratio of dissolved sugars to alcohol content): http://beerchooser.com/beermap/beergmapgravity.html