Deep Learning Quick Reference

Defining our example problem

In our example problem, we will be using a wine quality dataset created by P. Cortez et al. (https://archive.ics.uci.edu/ml/datasets/wine+quality). We will be predicting the percentage of alcohol present in the white wine data, given the wine's 10 other chemical traits.

There are 4,898 total observations or elements in this dataset, which is perhaps large for a classic regression problem but it's quite small for a deep learning problem.

Some quick exploratory data analysis will tell us that the 10 chemical traits we'll be using to predict alcohol content are all continuous variables, on various scales.