How to choose the right data Visualisation
In the world of analytics and reporting, data visualisation is the interface between raw data and the decision makers who rely on it.
If you go to the trouble of preparing your data for analysis, and then choose the wrong visualisations, then a lot of that effort is wasted. Choosing the right visualisation in each situation is a critical skill in becoming a more data-driven organisation.
This guide will strip away some of the mystery, and make it easier to decide on which visualisation you should be using. It provides an overview of some common visualisation types, and some pointers on deciding which fits your scenario.
All visualisations on this page are created with Power BI.
Line Chart
The line chart is perhaps the most basic of all chart types.
It provides a simple way of representing the trend of a single, numerical variable over time.
The standard line chart shows a point for each value, connected by sloped lines.
Axes
X-axis = A date or time variable
Y-axis = A numerical variable
Use cases for line charts
Showing a trend over time
Visualising time-series data
Showing the trend of sales over time
Visualising the uncontrolled impact of a change
Stepped Line Chart
A stepped line chart is like the standard line chart above, except that each time period is shown as a horizontal line, and these are connected by vertical lines.
It is a suitable alternative to a standard line chart for variables that are assessed at a point in time. A useful way of thinking about this is that ‘stock’ variables should use a stepped line chart, and ‘flow’ variables should use a standard line chart.
An example of a stock variable is inventory remaining, which might be assessed at each month’s end. Sales, on the other hand, is a common flow variable.
Axes
X-axis = A date or time variable
Y-axis = A numerical, point in time variable
Use cases for stepped bar charts
Showing the change in a stock variable over time
Showing the trend of a variable which is measured at a point in time
Showing changes in inventory value over time
Bar chart
A bar chart is a little more versatile than a line chart, because either a categorical or date variable can be used on the x-axis.
However, if you are showing a single, numerical variable over time, you should use a line chart to keep the roles of these charts separate.
Note that you may choose to show the values for each bar as a label rather than on the y-axis. This can make things clearer when there are relatively few categories.
Axes
X-axis = Categorical variable
Y-axis = Numerical variable
Use cases for bar charts
Comparing values between categories
Showing categorical data in an easily digestible way
Showing the breakdown of sales by category
Combo chart
A combo chart is a bar chart with a line chart overlaid.
This is useful for tracking two related variables over time that are on significantly different scales.
See Clustered Bar Chart for cases where the two variables can be represented on the same axis.
These charts are commonly used when you have one count or sum-based variable, and one proportion or percentage-based variable, and you wish to see how they move together over time.
As a rule, have your sum/count variable as bars, and the proportion/percentage variable as a line.
Axes
X-axis = A date or time variable
Primary y-axis = Numerical variable
Secondary y-axis = Numerical variable
Use cases for combo charts
Visualising sales quantity vs margin % over time
Visualising ad impressions vs click through rate over time
Visualising correlation between two variables
Clustered Bar Chart
A clustered bar chart has more than one dimension or metric shown for each category along the x-axis, and these are shown side by side.
In the first chart, each category along the x-axis has two metrics shown side by side. Note that these two metrics make sense on the same scale.
In the second chart, each category along the x-axis is broken down by a second dimension.
Axes
X-axis = Categorical variable
Y-axis = Numerical variable
Use cases for clustered bar charts
Showing categorical data for multiple metrics
Showing categorical data with an extra breakdown
Showing total sales and cost for each category
Stacked Bar Chart
The stacked bar chart is like the clustered bar chart, except that the different categories are shown stacked on top of one another, rather than side by side.
This is useful when you are mostly interested in the total values of the primary dimension (the one shown on the horizontal axis) but would also like an idea of a secondary category breakdown as well.
Axes
X-axis = Categorical variable
Y-axis = Numerical variable
Use cases for a stacked bar chart
Adding an extra level of detail to a standard bar chart
Showing the category breakdown for each sales channel
100% stacked bar chart
This is like a stacked bar chart, except that each bar adds to 100% rather than to the total value for each primary dimension.
This is useful when you are more interested in how each primary category is broken down by a second dimension, rather than the total values.
Axes
X-axis = Categorical variable
Y-axis = Numerical variable as a proportion
Use cases for 100% stacked bar charts
Showing categorical data broken down by an extra dimension
Showing how much each sales channel depends on different categories
Scatter plot
A scatter plot is a useful way of showing the correlation between two variables.
Be careful how you communicate these though, as a correlation does not mean that there is a causal relationship.
If you are comparing a ‘lever’ variable to a ‘target’ variable, it is standard practice to place the lever variable on the x-axis, and the target variable on the y-axis.
Helpful hint
If you’re showing daily data points in a scatter plot, make weekend data points a different colour.
These will often differ on one or both dimensions, so it will make the relationship clearer.
Axes
X-axis = Numerical variable
Y-axis = Numerical variable
Use cases for scatter plots
Visualising correlation between two variables
As an alternative, correlation focused version of a combo chart
Visualising the uncontrolled impact of a lever variable on a target variable
Table
Sometimes you just need a standard table of data, particularly in formal settings such as board or financial reports.
Here are some ideas to make tables more visually appealing and easier to interpret:
Use different font and background colours for the header and footer rows
Have alternating colours for rows
Use a sensible unit scale for what the table is being used for, and by whom
Add data bars for simpler interpretation of key variables
Use cases for tables
preparing financial reports
preparing board reports
presenting data in a formal setting
showing a detailed breakdown of categorical data
Conclusion
When armed with the right information, choosing the best visualisation for each scenario is some of the lowest hanging fruit on your journey to becoming a data driven enterprise.
When designing your reports, dashboards, and other analyses, make sure that your visualisation best communicates the point you are making. Sometimes the same relationship should be shown in two or more ways, to provide alternative perspectives, or to reinforce a point.