Next level Data Visualization: The World in Tune


Next level Data Visualization: The World in Tune

The world according to musical preferences

The challenge in consultancy is to gather all individual skills of a divers team in a client project. Our ambition is to develop and implement more advanced visualizations at clients. But reality is that at the end of the day we are building dashboards and less advanced visualizations. Visualizations that provide unknown & valuable insights.

Frans Geurts | September 24, 2019

Undoubtedly extremely important, but it does not really raise the bar for data visualization within Deloitte. Yet, the ‘Information Is Beautiful Awards’ gave our team a big opportunity to step up. We made a scroll story in which we rearranged 60 countries of the world in to new continents, based on the music preferences of their habitants. We did so using the latest techniques of data visualization.

Check out the World in Tune


The project consisted of five phases:

1.       Brainstorming

2.       Gathering the data

3.       Data discovery

4.       Sketching

5.       Programming


Phase 1: Brainstorming

Of course it is important to let all your ideas run freely during a brainstorm. But it is as equally important to know what you are looking for. That is why we defined some criteria about the topic:

1.       The topic is big enough;

2.       Open data which could be gathered relatively easy and is GDPR complient;

3.       The topic is recognizable to most people.

Pretty fast after that ‘music’ came across. But we immediately identified one major downside: ‘everybody’ is building Data Visualizations around this theme at the moment. The IronViz (Tableau) theme was indeed: music. That wasn’t really in our advantage. But, we came up with a really nice hypothesis and gave it a try.

Our hypothesis:

Do people who live in countries which are close together (e.g. same continent) have similar music preferences?

And if not -what we expect-

How would the world look like if we would arrange the world based on la-ti-da instead of latitude?

Phase 2: Gathering the data

Music streaming service Spotify has currently over 252 million users. Every day, Spotify tracks the 50 most popular songs for 60 countries. Besides, Spotify provides 13 different features for every track on their platform. With their API we were able to gather all these songs and features. Within a week we had the data set we needed.

Phase 3: Data discovery

As said before, Spotify provides 13 audio features for every track. We did a deep dive into these features. We quickly decided to focus on three features, which are the most independent. First of all: Danceability. Danceability describes how suitable a track is for dancing. This based on a combination of musical elements including tempo, rhythm stability, beat strength and overall regularity.

Then there is Valence. Valence is a measure for describing the musical positiveness conveyed by a track. High valence tracks sounds more positive (happy, cheerful) while low valence tracks sound more sad, depressed.

And the last feature is Tempo. Tempo measures overall estimated tempo of track in beats per minute. It’s the speed or pace of a give piece.

By combining these three features, we were able to create a unique DNA per country. This is were the math comes in. Because different features have different scales. Tempo is measured in BPM (on a scale from 50 to roughly 220) while Valence has a (calculated) scale from 0 to 10.

Phase 4: Sketching

Parallel to phase 3 we started sketching. Sketching is a really powerful approach in data visualization. It is a really quick, iterative, activity. It shows really fast whether a given type of visualization works (for the given data) or not. After a lot of -rather bad- sketches, we came up with a first circular view. In this view, the 60 countries would be arranged based on continents. At this phase, our working title was ‘beat of the world’. From a visual point of view, we would like to stay close to ‘music’. But also to ‘beat’. We agreed that the metaphor of a ‘equalizer’ fitted perfectly right here.

We also agreed that it would be a ‘scroll story’. At least for the first part. It gives us the possibility to combine an explanatory visual with an exploratory visualization. In the first part, we determine what the viewer sees (and what not). We guide him or her through our story. Through the insights we gathered in the data discovery phase. To ensure that nothing will be missed and that the story has a logic and decent build up.

From a design point of view, we committed to the style and colors of the Deloitte brand. And although a lot of these visual assumptions are written down in a style guide, they still provide some flexibility. Within these boundaries, our visual designer managed to come with a clear, bright and tight design for the visualizations and the UX/UI.

Based on these principles and conditions, we were ready to build.

Phase 5: Programming

We built an equalizer that ‘revealed’ the DNA per country. As the equalizer builds up per feature, we also provide the two songs at both ends of the given spectrum. With this we add an extra layer of information encoding. Not only a visualization and a corresponding text, but also the sound of Danceability, Valence and Tempo. We really want the viewer to ‘ feel the data’.

In the meantime we already discovered that our hypothesis was right. Countries really differ a lot. To test this hypothesis we built an algorithm that logically clusters the 60 countries, based on the given features. It really was the most important step in the process. After all, it is the backbone of our story.

The next step was to rearrange countries to a ‘new’ continent. We really struggled on this part of the visualization. We couldn’t just simply give countries a new spot on the map. Especially since we had data for only 60 countries. Eventually, we came up with the idea to lay these countries flat. It showed that it’s even more obvious that countries that are close together are not always similar. But since we we’re rearranging countries from existing continents to ‘new’ continents, we felt that it was necessary to show a map in between. To show that some clusters of countries are really remarkable. Japan, for instance, is a cluster on its own. To emphasize the transition, we used thin white lines that fade out after the animation. All these animations (and those appointed before) were supported by the phrases of our teammate who focuses on storytelling. He came up with names for new continents like Sad East Asia and The Kanye West.

At this point our explanatory story ends. We showed how the world would look like when it was arranged  on music. In other words: “The World in Tune”.

The last step we would like to add, is to make it personal. What’s in it for me? The story is initially on a pretty high level of countries and continents. But an important question that viewers could have is: where do I belong (based on my music preferences).

Therefore we added three DJ sliders, corresponding with the three features. This allows viewers to customize their own music preference. In return, they get back the continent where they should live, based on their musical DNA and a track list tailored to that preferences.

Check out the end result here.

Finally: The chase is better than the catch…

It’s a quote I shared with my team before. Of course, our main goal is winning. And we’re a pretty competitive team. But the biggest profit at this stage is teaming up with a bunch of really talented people with different skills. Don’t forget: the entire visualization was build outside office hours. But a lot of good food, club mate and a good challenge can get a lot going on. It provides expansion of our Advanced Data Visualization portfolio and will pay back in the future. When we deliver these kind of solutions for our clients, even more.

We confirmed our hypothesis that inhabitants of neighboring countries do not listen to the same music. We delivered a complete story for the viewer to read including something for them, to see with which new continent they match. Last but not least we are proud of the way we visualized the data so that it provided new insights that otherwise would have remain hidden.

Our team

Lianne Duinkerken

Frans Geurts

Lisa Kroes

Pim Peeters

Maarten Snijders

Niels Tammes

Joost de Theije

Bas Wagenmaker

Interested in an advanced visualization that tells a compelling story with your data or reveals insights you we’re not aware of? Get in touch below or through our LinkedIn profiles. 

Did you find this useful?