Data Visualisation – How to design charts and graphs

With size of our data increasing exponentially in today’s world, it only is of use to us if we can interpret it right. Data visualisation provides us with the ability to examine, interpret and communicate the underlying data in an intuitive manner. Data viz, when done right, could prove hugely beneficial in improving our operations, and thus improving our business.
Data visualization is only as useful as the data itself, but it is crucial for us to not to misrepresent the data. The data visualisation, when created must fulfil the audience it has been catered for. Only then, can we construe the insights that could be gained from it. Therefore, we need to have some guidelines while designing the charts for it to be effective.

1) Choosing the right chart/graph for the data
Data can either be numerical or categorical. But the representation of it might depend on the context in which the data is looked at. Some examples are given below. A combination of given graphs and charts can also be used for effective representation.
a) Data over time
Data that needs to be looked at over a period of time can be done on the following charts. Sales over month would be a good example
• Line chart
• Area chart
• Bubble chart

b) Data compared over categories
Data that needs to be compared across two or more categories can be done by the following charts. Sales across different verticals would be a good example.
• Bar graphs
• Boxplot
• Stacked or clustered column charts

c) Data in proportions
Data that needs to be represented as proportion of a whole. Total sales of all the products as a proportion would be a good example
• Tree map
• Donut/Pie chart
• Funnel chart

d) Data as a distribution

Data that needs to be distributed to be observed as itself or across another numerical variable. Sales of houses with respect to size of the house would be a good example.
• Scatter plot
• Box and whisker plot
• Histogram

2) Format of the graphs
The composition of the graphs makes it easier or difficult to understand. The composition is therefore, of importance. The composition of a graph lies in the labelling, the colour and the structure of the graph itself.
Colour
Colours are necessary when it comes to graphs and charts. Some things to keep in mind while using colours in graphs are as following
• When representing a variable with a colour, stay consistent throughout the entire report. With a colour assigned to a variable, the user gets familiarized and is not confused.

• When representing the intensity of the same variable, a good idea would be to use different tints and tones of the same colour. A contrasting colour to the background makes the data point jump at the user. Use it to highlight data of importance.

• When representing multiple variables in the same chart, a good idea is to use contrasting colours to easily distinguish between them.
• While choosing colours, be intuitive. Our minds associate with green as good and red as bad. Use this to your advantage.

• Consider the possibility of the user being colour blind and pick a colour palette that makes it colour blind friendly.

Labelling
• Make sure you label the x-axis and y-axis with units. A person who has no context should be able to understand what data is on the graph.
• If there is more than one colour, make sure you add a legend on the graph.
• The size and the font of the text used in the graph must make the text visible and pleasing to the eye. It must also be uniform throughout the dashboard to avoid confusion.
• Add axis to the graphs to make it easier to read. Sometimes the axis makes the visual too congested. Be wary of it. Reduce the scaling of the graph if necessary.
• Label important data points and use a contrasting colour to make it look striking.
• Make sure there is not too much labelling as this makes the graph overcrowded.

3) Dashboard Structuring
The dashboard must be built in a user-friendly way. If the consumer is not directed on what he’s looking for the, dashboard has not competent.
• A dashboard must not contain more than 3 visuals. If the dashboard is crowded with graphs, the size of each graph is compromised and hence the intelligibility is compromised.
• The order of the categories in a graph is important. It should either be arranged in a contextual context of importance, size or general ordinality.
• Comments on the graphs are not necessary but when used wisely, can make the overall comprehension greater.
• Rather than labelling the graph to describe the graph, explain the insight. Show the user what he’s looking for.
• Use a uniform background colour. Keep the number of colours on a dashboard to a minimum. Contrasting colours to highlight the things of importance. Choose the colours sensibly.
• Reference lines on graphs are great for easy consumption. Reference lines on graphs to give a context on the existing graphs to compare the figures easily.
• Interactive graphs are great but remember that the graph must work even without them.

Though data visualization proves useful in finding the insights, it must be designed keeping in mind the variety of people who will look at it. There are people who scrutinize the data using these graphs and there are people looking for the insights these graphs produce. There designer must keep in mind the audience he is making the graphs and charts for. The above tips will improve the design of the graph in case of any audience.