Workflow for creating data visualizations/dashboards

Workflow for creating data visualizations/dashboards

August 16, 2019 DATAcated Challenge 0

 Workflow for creating data visualizations/dashboards 

In my opinion, a data science project has 4 important steps- Identify the problem (consumes 10% of the total project time), Prepare the data (70% of the total project time), Analyze the data (10%), Visualize insights (10%) and a bonus step of presenting the findings. However, the success of the entire project depends equally on each of these steps. Having said that, the success of an individual step depends on its preceding step i.e. analysis of data totally depends on how well and accurate the data is prepared. Likewise, the presentation of findings and convincing the audience (clients, senior leadership, etc.) to a great extent depends on how efficiently the insights are visualized.

Many people think that Data Visualization is unlike other projects, it does not require planning and we can jump right into it. But I beg to differ to this thinking. I believe, a little bit of up-front planning will save you hours of blood, sweat, and tears in the long run. While creating data visualizations or dashboards/storyboards, it is of utmost importance that you select the right charts or plots which help to showcase the analysis effectively and vividly. Attention to detail on the visualization quality that includes axis, labels, color schemes, scale, etc. works wonders during the presentation. If enough time is spent carefully considering these details, then the rest of the task will be like a breeze.

Hence, I like to abide by the following series of steps-

Step 1: Analyze your audience

Knowing your audience and their expectations, understanding the kind of decisions they make and what really matters to them, is very necessary before starting to create visualizations. Hence, this step involves answering a few fundamental but crucial questions like,

  • Who is your audience?
  • What is your audience’s data literacy?
  • What is your audience’s data visualization familiarity level?
  • How much time does the audience have?
  • How much precision is necessary?
  • How many decimal places are necessary?

 

Step 2: Choose the right chart

There are many methods to visualize data, new solutions, and chart types come out constantly, and each strives to create more attractive and informative charts than before. Try out different charts and see which one suits your need the best and helps you convey your finding with ease. Having fancy charts with convoluted conclusions helps no one. It is advisable to focus on the principle that visuals should clarify and summarize the key message to be clear and concise rather than confusing and overloading the reader with superfluous information. Also, if you are not sure how your chart will add value for the readers, don’t make one. Every chart needs a purpose.

 

Step 3: Improvise

This step is to get rid of the redundant information by decluttering the charts and visuals, and making them more engaging and pleasing for the audience. For example, add legends and labels, use relevant shapes and colors, use the right font and font size, add titles and emphasize the takeaway message. The aim is to convey maximum information in a minimalist possible way, standing true to the saying- A picture is worth a thousand words!

 

Step 4: Tell the story

In this step, I like to split my visualizations into components like time component and space components. All the line charts and time series analysis plots form the time component, and all the space or area related plots like the geographical maps form the space component. There could be other components as well based on the data. Putting each visual in a corresponding component helps to tell an impactful story.

Does your data tell a story? If not, you need to revisit the steps above and understand what went wrong.

 

I would like to end it with a meaningful quote-

The greatest value of a picture is when it forces us to notice what we never expected to see.

—John Tukey

 

Submitted by-

Dipti Bhattad

Leave a Reply

Your email address will not be published. Required fields are marked *