Wine tasting is data science

A data science adventure into wine

Giuliano Manno
4 min readJul 3, 2019

“Sensory analysis is data science.”

That was the premise for a unique event earlier this month at our HQ: a wine-tasting led by sommelier and senior engineer Giuliano Manno as a means to creatively explore the premise of data science.

READ: not just an excuse to drink wine, but to genuinely and meaningfully explore data and its potential. The wine-tasting event was, according to Giuliano, a means “ to appreciate wine and the knowledge of wine and understand scientific approach to our senses”

Giuliano, AIS/WSA Level 3 sommelier and native Italian who comes from a long line of wine-makers and sommeliers, kicked off the event with a story about how has career has been intertwined with wine and computer science.

Wine tasting, similarly to data analysis is about reaching insights. In this initial part he explained the “Physiology of an Insight.”

“The first thing to understanding wine-tasting, is understanding data.”

How a single grape = insight

A wine grape is an insight of nine months-worth of data (that’s how long it takes for a grape to be harvested). The characteristics and complexities of a wine depend on a variety of factors from the grape’s development, from soil to climate (or terroir). Through wine-tasting, we can “understand the context with our senses,” Giuliano explained. “Sensory analysis is data science.”

Using only one of the five sense alone for wine-tasting isn’t sufficient to understanding that grape — much like only one data dimension does not create a sufficient data insight.

  • Sight: Is the first part of the analysis, which helps to assess the evolution and quality of the wine. Color gives important indicators about the age. A greenish-yellow indicates the presence chlorophyll, meaning a young grape, and a purple color in a red wine also means a young grape. Body and color intensity are also an indicator of a wine’s structure. The structure is the property that sustains the graceful evolution of the wine over time. Giuliano supported this section with the use of his educational app for detecting wine colors.
Open it on your mobile here.
Or scan the QR code with your mobile camera
  • Smell: While the scents perceived can vary by individual (there are few wrong answers, by the way!), the notes we pick up prepare us for the taste of a wine and provide fundamental information to appreciate it. (Find a bit more detail about that here.) The analysis of the olfactory descriptors is a special case of hierarchical clustering, the first passage evaluates the complexity (nodes) and the second one the sub-clusters and their associated intensity (weight), e.g. pinot noir: red fruits > berries > cranberries(strong). The evolution of the fruit descriptors (unripe, ripe, jam, in spirit) is an important insight to identify the quality, elegance and evolution of the wine.
  • Taste: A cleansed palate is key — Giuliano equipped us with non-flavored crackers and water to make sure we could properly taste each of the three wines. He was also hawkish about sipping (not chugging — this was a learning event, after all) and making sure we let the wine sit in our mouths for a bit, to fully pick up the flavors. You can learn more about flavor indicators here.

By iteratively focusing on these 3 areas (image below), we can correlate the collected data and have a better understanding (insights) of the qualities of the wine, as well as the trajectory of its evolution. Having better knowledge about a wine allows us to ponderate non trivial decisions, for example when it’s the best time to open another bottle, or what are the best food pairings to elevate the wine’s characteristics. Choosing the right wine is not an easy task.

The spiral to the insight


  • Wine-tasting, which leverages multiple senses and several layers of evaluation within each, was an effective metaphor of data science, which uses a multitude of data. Understanding the evolution of data over time (or the evolution of a wine) is a way to understand trends and support decisional processes.
  • It’s super handy to have a Level 3 sommelier in the office…. but he will publicly judge you if you don’t hold your wine glass by the stem. You’ve been warned.

Thanks to Giuliano for being patient with us and contributing to this article. We’re very fortunate to benefit from his dual passions of teaching wine and computer science!