The Unicask Project: A Blend of Whisky and Data

Picture this: you are at a whisky tasting, surrounded by great drams, and someone asks, “So what do you actually like?”

With over 100 distilleries in Scotland alone, I did not have a great answer. I had notes, opinions, and favourites, but nothing structured enough to support it.

That is where The Unicask Project came from.

Why This Project?

I wanted to make my whisky tastings more consistent and actually useful over time. Instead of scattered notes, I started treating each tasting as a data point.

This project let me combine something I enjoy with the skills I am building in data analysis. The goal was not to remove the subjectivity, but to structure it enough that I could start answering simple questions properly, such as:

• Which regions do I actually rate highest?

• Are there distilleries I consistently enjoy?

• How do my preferences change over time?

At its core, this is about taking unstructured, subjective data and making it analysable.

Whisky tasting is inconsistent by nature. Palates vary. I might pick up toasted marshmallow in a dram, while someone else might think it just tastes like burning. Both are valid, but difficult to compare.

By standardising how I capture each tasting, I have built a dataset that allows for consistent comparison across regions, distilleries, and time.

What You Will Gain:

By reading this article, you will gain insights into how personal whisky tasting data can be analyzed to identify preference patterns, compare distilleries, and discover new favorites. You will also see how data can enrich even the most personal of hobbies.

Key Takeaways:

• Many of my highest-rated whiskies come from Islay and Highlands.

• My favorite dram is Smokehead, a Special Malt with an ABV of 43.0%.

• Generally, Single Malts outshine Blended Malts in my ratings.

• Most whiskies I tasted ranked from “Solid” to “Pleasant,” with few reaching “Excellent.”

• Surprisingly, some renowned blended malts did not rate as highly as lesser-known single malts.

Dataset Details:

The dataset is self-generated and built from individual tastings. Each row represents a single dram and includes:

• Distillery

• Region/Country

• ABV

• Rating

• Latitude & Longitude of Distillery

The focus is on keeping it simple and consistent so it scales over time and remains usable for analysis.

Analysis Process:

I used Excel to structure and clean the data, then built the visual layer in Power BI.

The aim with the visuals was straightforward. I wanted to make the data easy to explore. Not just dashboards for the sake of it, but something that reflects how you would naturally think about whisky by place, by distillery, and by overall performance.

Visuals and Insights:

The table shows each dram with key details such as rating, ABV, region, and distillery. Colour-coded markers make it easy to scan performance, and tooltips provide additional context such as minimum and maximum ratings and count of ratings. It is the most direct way to compare individual whiskies.

The maps bring in the geographic side. At a glance, you can see how ratings are distributed globally, then drill into regions such as Scotland or Japan for more detail.

Markers are colour-coded by rating, and tooltips show distillery-level detail, including all expressions and the average rating. Each location reflects an overall score based on those averages, making it easy to compare distilleries and spot patterns or outliers.

Main Takeaways:

A few things stood out quickly:

• Some regions consistently trend higher in my ratings

• There is still a lot of variation within those regions

• My scoring became more consistent over time

• Visualising the data made standout drams and patterns much easier to spot

It also challenged a few assumptions. There are distilleries and regions I expected to enjoy more that did not always deliver, and others that surprised me.

Conclusion and Personal Reflections:

This project is really about process.

It shows how you can take something messy and subjective and turn it into a dataset that supports analysis. That includes:

• Structuring and standardising inputs

• Designing a simple, scalable data model

• Building visuals that make the data easy to explore and understand

This is the same approach you would apply to customer feedback, reviews, or any other unstructured data.

As the dataset grows, the analysis becomes more reliable. The next step is to continue refining both the structure and the visuals, and to explore deeper comparisons over time.

Call To Action:

This started as a way to organise my own notes, but it has turned into something more useful.

Now, if someone asks me which distilleries I like, I have an answer and the data to support it.

If you found this article interesting or know someone looking to hire a data analyst, let’s connect...send me an e-mail or find me on LinkedIn! I’d love to hear your thoughts or answer any questions you might have.