Analyzing Wine Spectator Reviews: Searching for Value

Posted on Jul 3, 2022

The skills the author demoed here can be learned through taking Data Science with Machine Learning bootcamp with NYC Data Science Academy.

Trying to purchase the right wine can be a daunting task, balancing price, quality, and suitability is intimidating to even seasoned oenophiles. Can we use data to make the task a little easier and more rewarding?

It is estimated the United States' wine market produces around 330 million cases of wine per year and continues to grow despite the pandemic and inflationary concerns. During the first six months of the pandemic retail wine sales grew by 19.3% by volume and 24.7% by value, trends indicate sales of premium wine (priced above $20 per bottle) will continue to rise over the coming years and vineyards are responding to this trend by producing an additional 30% by volume of more expensive bottles.

This leads an average consumer feeling lost among a constantly changing variety of products in an environment not friendly to the inexperienced. It has been shown most liquor stores heavily feature products with the highest profit margin and give very little credence to the quality of the wine they sell. A small amount of education into the economics of selecting a bottle of wine for your dinner table, can pay very large dividends.

'Wine Spectator' magazine is widely considered the foremost source of critical wine reviews in the United States, and arguably, the world. Reviewing over 15,000 wines per year in blind tastings from vineyards all over the globe, this mass collection of data on bottle price, varietal, location, and taste can allow us make smarter decisions on what bottle to bring home.

Wine Data Description

Each row in this dataset contains information on one bottle of wine including: price, review score (1-100), text description of flavor, varietal, year, and vineyard location. It will be useful to create price and review score 'buckets' to more easily parse our data:

Wine Price Buckets:

  • Cheap: 4-17($)
  • Inexpensive: 17-25($)
  • Moderate: 26-42($)
  • Pricey: 43-79($)
  • Expensive: 80-175($)
  • Outlandish: 175($) +

Distribution of Wine Price

Price Distribution

Distribution of Wine Review Scores:

  • 95-100 Classic: a great wine
  • 90-94 Outstanding: a wine of superior character and style
  • 85-89 Very good: a wine with special qualities
  • 80-84 Good: a solid, well-made wine
  • 75-79 Mediocre: a drinkable wine that may have minor flaws
  • 50-74 Not recommended

Distribution of Review Scores

We are primarily interested in the relationship between wine price and review score:

Review Score vs Price

A clear logarithmic relationship emerges demonstrating a positive correlation between price and review score

Let's explore this relationship in a little more detail:

Distribution of Wine Review Score vs Price Bucket

We can see more variation in review score in the lower price buckets and clear bias towards whole number reviews.

Review Scores by Country

The Terroir of wine can mean many things, it literally translates as 'a sense of place' but in general encompasses all of the factors that go into producing wine grapes in a vineyard, from the climate to the soil to the elevation. Every country has their own unique elements that contribute to a wine's flavor and quality, therefore it is useful to look at review scores by country.

Let's look at the review scores for the top 5 most reviewed countries from Wine Spectator:

Review Score vs Price for 5 Countries

Surprisingly, Italian wine (A country renowned for their wine) score noticeably lower by price against the other 4 sampled countries.

Let's see if there are particular price categories that are pulling down Italy's review scores.

Rating vs Price by Country

For the 'cheap' and 'Inexpensive' price buckets Italy's review scores seem to fall in line with the other sampled countries. However, once we approach the three most expensive categories we see a clear trend, A lower mean review score. Does this mean we should avoid purchasing more expensive Italian wine? To answer this question, we have to delve a little deeper.

Are you OK, Italy? - Pt. 1

Let's look more closely at how Italian wine is reviewed against the rest of the world.

Italian Wine Review Scores Compared to the Rest of the World

We see the same trend of lower mean review scores against the rest of the world. Are there any particular regions in Italy that are pulling the mean score down.

Are you OK, Italy? - Pt. 2

Italian Wine Regions with Lowest Mean Review Scores, Compared with the Rest of the World

The above plot shows which regions in each price bucket to avoid when purchasing Italian wine. Many regions are scoring a full point lower than the world average. In order to fully parse what this graph is showing us, it is important to understand the concept of P.D.O. (Protected Designation of Origin). PDO is a series of rules and regulations that dictate how certain food and beverage items can be labeled. In the instance of wine PDO means that a wine with this mark on the label has been produced in a specified area and has been aged and bottled in accordance with existing regulations and under strict control by government authorities.

In Italy, PDO is often referred to by its native language equivalent, DOC (DENOMINAZIONE DI ORIGINE CONTROLLATA) For wine to be labeled as originating in a specific region it must meet the standards for that region. Restrictions do not just include grape origin and varietal but can include aging time, barrel sourcing requirements, alcohol content, residual sugar and a host of others.

The geographically names regions in the above chart, 'Northwestern', Northeastern,', Southern', 'Central', and also 'Other' do not appear on any DOC list, indicating the wines are not produced to any strict government or historical standards. The aforementioned regions account for approximately 60% of reviews which score below the world average. We can then conclude that the DOC label on Italian wine is of high importance, and to be wary of bottles not branded with a protected designation of origin seal.

Finding Value in Wine Pairings

There can be some anxiety when standing in the wine shop looking at options available. Imagine the disappointment when you uncork that bottle and take your first sip, and the first thing you experience is underwhelming. Nothing can prevent being let down by a bottle of wine, even seasoned professionals in the art of wine pairing and viticulture feel like they have wasted money from time to time.

Luckily Wine Spectator has provided us with a wealth of information which we can use to maximize our chances in selecting just the right bottle for our needs. With a little bit of knowledge of wine pairings and a lot of data we can engineer a feature looking for where we can find the best 'value' for our money.

Wine-Related Definitions

Wine Value

  • Value in wine will be defined by its review score divided by its price.
  • Value will be a figure of merit in review score per dollar spent
  • A higher relative number will therefore be a greater value wine

Wine Pairings

Using my own experience in oenology I will define 7 common entr├®e classifications and the varietals and blends that have the greatest chance to produce a good pairing

  • Fish : Sauvignon Blanc, Pinot Gris, White Blend, Chenin Blanc, Albari├▒o, Pinot Blanc
  • Red Meat : Cabernet Sauvignon, Bordeaux-style Red Blend, Nebbiolo, Rh├┤ne-style Red Blend, Cabernet Franc, Barbera, Verdejo, Petit Verdot
  • Salty : Ros├®, Sparkling Blend, Champagne Blend, Glera
  • Spicey : Syrah, Malbec, Tempranillo, Gamay, Shiraz, Tempranillo Blend, Grenache ,Petite Sirah, Garnacha
  • Rich : Pinot Noir, Merlot, Sangiovese, Zinfandel, Carmen├¿re, Torront├®s
  • Pork : Riesling, Gr├╝ner Veltliner, Gew├╝rztraminer, Blaufr├ñnkisch
  • Chicken : Chardonnay, Portuguese White, Viognier, Bordeaux-style White Blend, Rh├┤ne-style White Blend

Keep in mind, these pairing are a general guideline. For example, not all chicken preparations would be suitable for a Viognier or a Barbera for every steak diner, however; the above list will maximize our chances of selecting a good bottle. For more information on wine pairings, you can visit The Wine Cellar.

We have to be sure we have enough data for each of our pairings in each pairing and price bucket to create meaningful observations.

Data Analysis - Pt. 1

Percentage of Each Wine Pairing by Price Category

Expectedly, there are simply not enough reviews in the outlandish price bucket to generate any significant conclusions, so we will omit that price category from our consideration. The Pork and Salty categories have comparatively low percentages associated with them, however they still represent ~5,000 reviews each which is plenty for our purposes.

With the data at our disposal, it is trivial to construct a table that shows what region would produce wine of the highest value, the real challenge is trying to visualize all of the information in a compact format.

Value Wine Pairings by Price
Trouble Choosing the right wine?

The above table is faceted by wine price, each color represents a different pairing, and the size of each marker is relative value. For a practical demonstration of using this plot, imagine we are preparing a chicken dinner, and are willing to spend a moderate amount of money for a bottle of wine. We look at the blue marker, which corresponds to chicken and the central facet panel representing the price. The marker is at the intersection of Chardonnay on the x-axis and Spain, Cava on the Y-Axis. Concluding, the best chance of finding a high value wine for our chicken dinner at this price is a Chardonnay from the Cava region of Spain.

The value-based wine selection chart is a novel way of showing the highest value wine pairing for price and entr├®e selection, but it is useful to generalize a little further and look at which larger terroirs produce the highest value wine.

Data Analysis - Pt. 2

Let's look at how often countries show up in our value calculations for the top 3 highest value wines for each entr├®e pairing.

The above plot shows that most of our high value wines are coming from the United States and France. Lacking any other information and you are searching for wine with high value, buy American or French.

Future Work on Wine

This dataset is surprisingly robust, I could easily spend dozens of more hours picking out fascinating insights and facts about wine, some of the highlight may include:

  • Explore Bias in wine taster's reviews against certain countries or varietals
  • Find some correlation to some of the adjectives in the taste column with review score or price, perhaps even create some predictive model to determine a wines price or review score based on its description.
  • Exploration into some of the smaller country's wine production. For example, how does Greece's wine compare with the rest of the world.
  • Create a recommendation algorithm, which can take in certain key words and generate a wine region recommendation.

About Author

[email protected]

Education: Stevens Institute of Technology Bachelors of Engineering: Engineering Physics, Solid-State and Optical Engineering Bachelors of Science: Applied Mathematics Associates: Applied Physics Experience: 10+ years Quality Control Manager: Testing and Characterization of Solid State Frequency multiplied Diode Pumped...
View all posts by [email protected] >

Leave a Comment

No comments found.

View Posts by Categories


Our Recent Popular Posts


View Posts by Tags

#python #trainwithnycdsa 2019 2020 Revenue 3-points agriculture air quality airbnb airline alcohol Alex Baransky algorithm alumni Alumni Interview Alumni Reviews Alumni Spotlight alumni story Alumnus ames dataset ames housing dataset apartment rent API Application artist aws bank loans beautiful soup Best Bootcamp Best Data Science 2019 Best Data Science Bootcamp Best Data Science Bootcamp 2020 Best Ranked Big Data Book Launch Book-Signing bootcamp Bootcamp Alumni Bootcamp Prep boston safety Bundles cake recipe California Cancer Research capstone car price Career Career Day citibike classic cars classpass clustering Coding Course Demo Course Report covid 19 credit credit card crime frequency crops D3.js data data analysis Data Analyst data analytics data for tripadvisor reviews data science Data Science Academy Data Science Bootcamp Data science jobs Data Science Reviews Data Scientist Data Scientist Jobs data visualization database Deep Learning Demo Day Discount disney dplyr drug data e-commerce economy employee employee burnout employer networking environment feature engineering Finance Financial Data Science fitness studio Flask flight delay gbm Get Hired ggplot2 googleVis H20 Hadoop hallmark holiday movie happiness healthcare frauds higgs boson Hiring hiring partner events Hiring Partners hotels housing housing data housing predictions housing price hy-vee Income Industry Experts Injuries Instructor Blog Instructor Interview insurance italki Job Job Placement Jobs Jon Krohn JP Morgan Chase Kaggle Kickstarter las vegas airport lasso regression Lead Data Scienctist Lead Data Scientist leaflet league linear regression Logistic Regression machine learning Maps market matplotlib Medical Research Meet the team meetup methal health miami beach movie music Napoli NBA netflix Networking neural network Neural networks New Courses NHL nlp NYC NYC Data Science nyc data science academy NYC Open Data nyc property NYCDSA NYCDSA Alumni Online Online Bootcamp Online Training Open Data painter pandas Part-time performance phoenix pollutants Portfolio Development precision measurement prediction Prework Programming public safety PwC python Python Data Analysis python machine learning python scrapy python web scraping python webscraping Python Workshop R R Data Analysis R language R Programming R Shiny r studio R Visualization R Workshop R-bloggers random forest Ranking recommendation recommendation system regression Remote remote data science bootcamp Scrapy scrapy visualization seaborn seafood type Selenium sentiment analysis sentiment classification Shiny Shiny Dashboard Spark Special Special Summer Sports statistics streaming Student Interview Student Showcase SVM Switchup Tableau teachers team team performance TensorFlow Testimonial tf-idf Top Data Science Bootcamp Top manufacturing companies Transfers tweets twitter videos visualization wallstreet wallstreetbets web scraping Weekend Course What to expect whiskey whiskeyadvocate wildfire word cloud word2vec XGBoost yelp youtube trending ZORI