Dissecting Wine Quality Factors from Wine Enthusiast's Ratings Database
Inspired by my long-time curiosity of how a particular bottle of wine was perceived in terms of its quality, I gathered a dataset of 150930 wines from Wine Enthusiast's ratings database. Variables used in the dataset included the wine's grade (out of 100), grape varietal, country, state or province, and sub-region for some. One caveat is the magazine only posts reviews on wines receiving a grade of 80 or more, so the wines included here have already been pre-filtered out of an even larger universe.
My Shiny app can be accessed here.
I started out finding the top producers by country, as well as the most popular grape varietals within each of them. The top ten wine producing countries in the dataset, by the number of observations, are:
1 US 62397
2 Italy 23478
3 France 21098
4 Spain 8268
5 Chile 5819
6 Argentina 5631
7 Portugal 5322
8 Australia 4957
9 New Zealand 3320
10 Austria 3057
My initial visualization ideas, not all of which were implemented, were based on the following questions:
1. Does price necessarily reflect a wine's quality, indicated by the points given?
2. What do the distributions of points by country, in the New and Old Worlds, look like?
3. Are there any countries that don't produce as much but offer high quality wines?
4. Where are the inexpensive but high quality wines?
Since there is a large number of wines graded in this set, not all countries received an in-depth examination A search for the countries not in the top 10 producers' list but that ranked high on the median point received gave me results of England, India, Germany, Slovenia, and Canada.
On the other hand, are all the largest ten producers also high quality producers? Finding overlapping countries in the two top lists -- both in terms of production volume (top 10) and quality (top 25) -- yielded a list of Austria, France, Australia, Italy, Portugal, US, and Spain.
I was able to find 23 wines that appeared in the top 15% of all wines in terms of quality with price tags in the bottom 15% of all bottles surveyed (the 15% came about after a few trial and errors, as 10% yielded nothing in intersection and 20% too many). The entire list can be found in the last tab in the Shiny app.
The conclusions I was able to draw from this research were the following:
- Overall, wine prices do tend to be positively correlated with perceived quality, at least among these bottles already pre-selected after passing a minimum threshold. However, the difference is not as great as many people might have imagined. In many instances, you could be paying the same price for bottles of wine that received ratings very far apart.
- There are several other countries that produce either high quality or large volumes of wine but don't appear on the other list. It means we should look beyond the usual choices in terms of geography and explore more offbeat places for refreshingly great wines or good deals.