What Joe Strummer can teach Data Scientists

December 14,2016 , Eleni Kotzampasaki
View Image : 2.png

At Aquila Insight we work closely with clients to understand their business questions, use advanced analytics and technology to answer these issues through data, and communicate back how the resulting insights can be transformed into action.

As a data scientist, I am passionate about unravelling the interesting stories hidden behind data. During my first attempt at pulling a deck of slides together for a client presentation, I focused more on what was important to me as a data scientist rather than what the senior executives in the audience would find useful from a business perspective. I included a set of tables as evidence of a sound statistical model and even the SAS code used to perform the analysis. After running through the slides with a colleague and being told that the deck would mean nothing to the majority of people in the meeting, I was crestfallen.

Due to the nature of our job, data scientists spend a significant amount of time cleaning up datasets and exploring what the best algorithm is for each specific business question. We get very excited and proud of our analysis, sometimes forgetting that data science adds no business value unless transformed into actionable insight.

The feedback on my first deck of slides was a eureka moment. Performing the best analysis isn’t helpful unless it is beautifully presented and embedded in a story that the audience finds engaging.

The question I now ask myself before putting any graphs into a slide is: what do I want my audience to know?


Adding visuals is one way to engage an audience and share with them data insights. Producing beautiful graphs can draw the viewer’s eye to the main facts, aided by the use of text, colour, comparisons and pivotal points of data.

A common misconception is that data speaks for itself and that no text is needed when communicating with charts. I find that text gives context to the audience and makes sharing information effortless. For example, eligible legends, spelled out acronyms and active slide titles summarising what is happening put me in a great position for successful communication.

Another important decision to make is the choice of colour when communicating insight with graphs. A good practice is to shift everything to the background with the use of grey tones, using colour sparingly and strategically to draw attention to the main point of the graph.

Comparisons can also provide useful insight regarding performance and tend to appear in almost every data analysis deck. However, comparison charts can be noisy as lines are plotted over each other making the graph challenging to read. It is more efficient to isolate points and create more charts on new slides, play with chart orientation, use colour and text to draw definite conclusions out of comparisons.

Making the data the pivotal point in the narrative makes the communication of insight, and its relevance to the audience, successful.

My advice when communicating data is to think like a film director and:

  1. set the plot, what the audience needs to know before they see the data.
  2. unravel the twists, what’s interesting about the data.
  3. guide the audience to a concise ending which is the take home message.

To paraphrase the words of Joe Strummer, ‘every chart has a story to tell’ and it is our job as data scientists to make that story as compelling and relevant to our audiences’ business objectives as we possibly can.