Why Data Visualisation and Effective Analytics Communication is so Important
In what I believe is the last part of my 12-week Mini assignment with CXL Institute, I decided to use this to learn more about data presentation and the art of visualisation.
I’ve always been found of using data well and presenting it effectively. I’ve been the Data Editor of many News sites in India where I played a key role in making charts, graphs, gifs, viz using tools like datawrapper, Infogram, Piktochart, tableau, etc
Tim Wilson is our instructor for this course and he has been in the analytics field for the last 20 years. He is a Senior Analytics Director at Search Discovery.
He tells us that there are 3 big sections
1.Communications Overview
2.Data Visualisation
3.Data storytelling & Presenting
The first part starts off with a Communications Overview.
We are told about that data viz is not just about making the data pretty and how ineffective communication of the analysis can negate the impact of the analysis done (I agree with this)
We learn about the lifecycle of an analysis. Tim tells us about the misrepresentation
1. Data speaks for it-self
2. It’s clear to me
3. I’m not an artist
Anyone who says the above needs to learn the curse of knowledge. Just because you know it does not mean someone else knows it.
>>>Is curse of Knowledge a hinderance to data vis>>>>>
Effective communication is about brain science.
What is the Brain Science of Communication
This is lesson 2. Here we are going to learn about three types of memory: iconic, short term and long term. And focus on understanding the limitations of short term memory as per Milers’ law and why it is critical to work within those constraints.
Effective analytics communication is not about making data pretty but about being understood.
Iconic Memory > Short term > Long term
Psychologist George Miler law says that our brains were able to hold 7+_ pieces of information in pour short term memory.
7 chunks of information at one time.
We need to reduce the cognitive load on our audience.
The example of the cognitive load where we are asked to find how many ‘5’
is very interesting and it is a good example to show when teaching data viz sessions.
Some very subtle changes were made to highlight the changes. And it is clear that the image on the right tells us in a jiffy how many 5’s are there.
Tim then tells us that our brains are terrible at comparing area and this is why piecharts are a problem. The below example gives us an idea about what he is saying. When we are asked to tell how large circle ‘A’ is compared to circle ‘B’ then we see that many will not be able to give a good enough accurate answer for the circles but may be closer when asked when shown the bar graph.
Do try this since you are here. The answer is at the bottom of this article.
It can be presented this way but this is a far better example.
The next line chart example is really interesting and tells us how to make a chart the correct way so we can let our brains relax.
As you can see from this above line chart, there are many areas the brain ends up ignoring when it see a chart. For some words like month and sessions, they are redundant and for others we have to tilt our head which is not a great experience to give to your viewers.
Gestalt Psychology: humans perceive patterns on configurations not merely ind components.
Two books he recommended are:
Brain Rules by John Medina ( see website here)
“Information Dashboard”designed by Stephen
From here, we move onto the infamous pie charts.
We are told that they make the brain do a lot more work or add cognitive load and this can be avoided with the use of other charts.
I like the idea of seeing charts thru a cognitive lens.
Tim shares examples of how pie charts can be wrongly used and he has a point with the examples he takes us through. I also looked around to see if I could find one example and this one on twitter has very high cognitive load.
Now we move onto learning about data visualisation.
Data pixel ratio? What’s that
We also learn how to apply that concept to a chart and table with some examples.
We are also given an assignment to do and I am going to do that as I have created a lot of charts, graphs (atlas a few thousand) in the past few years.
The assignment is
- Pick an existing visualisation
you created. - Update it by maximising the
data-pixel ratio. - Compare the original to the
new one.
Bonus: wait a week and repeat the exercise with the updated visualisation.
Using Colour Sparingly
Tim tells us that we are now in the meat of the data viz part of the course. Color is (of course) a really important part to keep in mind when making a data viz.
There is a tendency to sometimes use color a lot. It does not help and can be unnecessary.
But sometimes, it can be used to selectively point out a particular data point.
Interesting tips here on color and what to do and not to do as well.
We also have an assignment here as well so I am combining the two assignments into one which you see below.
- Pick an existing visualization you created that uses multiple colors.
- Update it to remove as much color as possible.
- Compare the original to the new one.
The chart I’ve used for the assignments was made in 2014 when I was experimenting with data viz. It has been made using Infogram.
As you can see in the first pie chart, it is tooooo colorful. Of course, this is not how a chart should be. Too much cognitive load and stress on the poor brain.
I’ve spent some time doing it up and you can now see the difference. The bar is certainly a much better version and doesn’t tax the brain so much.
(The answer for them is 16. Were you able to get a closer answer for the bar chart against the circle)
(This is a part of my series as a fellow with CXL institute where I am pursuing a mini degree in Analytics)
