A Complete Guide to Choosing the Right Visualisation

Introduction

A good data analyst knows all the visualisations. A good data visualiser knows when to use each one. Visualisations are an art form. And like all art forms, they’re meant to send a message. While traditional art conveys the abstract, data visualisation conveys objective information. This guide will show you how you can use data visualisation best.

Principles of choosing the right visualisation

• Purpose – Your visualisation should primarily focus on getting the intended message across. People should be able to access all the information they need in as little space as possible.
• Clarity – Visualisations should be easily understood by the intended audience.
• Simplicity – Make sure the viewer spends as little time and energy on your visualisation as possible.
• Context - Context is key in data visualisation because it helps your audience grasp what the data means and why it matters. By providing background information and framing the data within its relevant circumstances, you enable better understanding and more meaningful insights.
• Aesthetics – Aesthetics in visualisations are a subject of their own, but put simply, good aesthetics help communicate the message more effectively, and bad aesthetics draw attention away from the intended message.

Objectives of Data Visualisation

To know which visualisation to choose, the most important factor is the purpose of the visualisation. Some of the most common objectives are:

• Showing change over time
• Showing part-to-whole composition
• Showing how data is distributed 
• Comparing values between groups
• Observing relationships between variables
• Showing geographical data 

Changes over Time

Charts that show changes over time have time on one axis, typically the X axis (horizontal). Time can be shown in various continuous and discrete ways, depending on which you choose a visualisation.

Bar Charts

Bar charts are ideal when

• You have single data points on the X axis OR
• You have uniform time intervals or bins on the X axis
• You have 5-10 values on the X axis
• You want to show changes in composition (stacked bar chart) or multiple subcategories (grouped bar chart) over time

Choose a different chart when

• You have continuous data 
• Time intervals are irregular 
• You have lots of time points
• You want to emphasize trends rather than compare. While bar charts are acceptable at emphasizing trends, line charts or area charts excel for this purpose.
• You are not using a zero baseline

‍

Line charts

Line charts are generally considered the go-to charts to show changes over time. Use line charts when

• At least one of your axes is continuous 
• You have a lot of data points
• You want to emphasize trends and changes over time rather than compare data during different points of time
• You want to compare multiple datasets over a time period

Choose a different chart when:

• You have very few data points
• The X-axis is non sequential
• The X-axis is discrete

‍

Area Charts

Area charts have a similar use case to line charts, with a few differences: 

• Area charts are better than line charts at highlighting the magnitude of change over time compared to line charts.

• Stacked area charts are good to show how individual parts contribute to a whole over time.

Avoid using area charts when:

• You are comparing multiple series, particularly when the data overlaps. Line charts are by far a better choice.
• You want to emphasize trends rather than volume. Go for a line chart instead.

Box Plots and Violin Plots

Both box plots and violin plots are the best choice to highlight how the distribution of data changes over time.

Box plots are usually the better of the two for this purpose unless you want the full shape of data distribution over time. For a complete guide to these two charts, check out this article.

‍

Heatmaps

Heatmaps use colour to show changes over time. Here’s when you should use heatmaps:

• When you have multiple categories and want to track their performance over time
• When you want to show a relationship between the X and Y axes 
• Heatmaps are equally effective at showing change over time, like line charts as well as emphasizing the magnitude, like area charts, through colour gradients. 
• The colours on heatmaps can also be based on non-numeric, qualitative values like general qualitative levels of low, medium and high.

‍

‍

Heatmaps also have the benefit of being scalable, which is useful when you are comparing cyclical data such as seasons. The above chart is set over 20 years, grouped by month. Each cell reports a numeric count, like in a standard data table, but the count is accompanied by a colour, with larger counts associated with darker colourings.
The chart not only compares the amount of precipitation by month, but also the distribution of precipitation in each month, and the colour emphasizes magnitude. 

Part-to-whole composition

These charts emphasize how individual components contribute to a larger whole. These charts are best used when you want to understand the proportion of each part in relation to the total.

Pie-chart

Pie-charts are one of the simplest and most well known charts, but they have a fairly narrow use case. Use pie charts:

• When you have 5-7 categories or less
• If the differences in the values are significant enough that it can easily be spotted on a pie chart
• When your goal is to focus on how each category contributes to the whole rather than compare categories to each other

Avoid using pie charts: 

• When you have too many categories. This can make the pie chart look cluttered. Tree maps are better suited for this purpose.
• When the differences are subtle. This makes it difficult to differentiate between the contribution of each category to the whole. Use a bar chart or waterfall chart instead.
• When comparing composition over time: Use stacked bar charts instead.
• When comparing data across multiple variables. Use stacked bar charts instead. 
• When the categories don’t add up to a meaningful whole.

For more information on pie charts, check out our comprehensive guide.

Stacked Bar Chart

A stacked bar chart is like a regular bar chart but with a twist – it lets you compare two categories at once. Instead of just showing one set of values, each bar is split into sections, or "sub-bars," that stack on top of each other. Each section represents a different group within the second category, making it easy to see how everything adds up across both categories.

Use stacked bar charts when:

• You want to compare the composition of different categories, in addition to the total
• You want to show how the composition of something changes over time
• You want to show the total in addition to the segments
• You want to emphasize the differences between segments

Avoid using stacked bar charts when:

• Your sole aim is to show the composition of something at a single point of time. Use a pie chart or a tree map instead.
• You have zero or negative values
• When there’s a small difference in categories. Use a tree map instead.

‍

Stacked Area Chart

A stacked area chart shows how different parts contribute to the whole over time. A stacked area chart is similar to a regular area chart, but the areas representing different categories are stacked on top of each other.

Tree Maps

A tree map is composed of nested rectangles. Each rectangle represents a category or subcategory, and its size is proportional to the value or importance of that category in relation to the whole. 

Tree maps can show both categories and subcategories easily and are more suitable than pie charts for showing a large number of categories.

Waterfall Charts

A waterfall chart is a data visualisation that shows how an initial value is affected by a series of intermediate positive or negative values. 

Waterfall charts have specific use-cases as they require an initial bar, or “anchor”.
Here are some of the best uses for waterfall charts:

• Profit and Loss Analysis: Great for showing how different factors like sales, costs, and other expenses contribute to a company’s profit.
• Budget Breakdown: Illustrates how different expenses reduce a budget, showing the remaining funds.
• Progress Towards a Goal: Tracks how small achievements or setbacks affect progress toward a target.
• Cumulative Effects: Shows the incremental effect of sequential factors, helping you explain how individual contributions add up to the whole.

Data Distribution

One great use for visualisations is to show how data points are spread out or distributed. This comes in especially handy when you're exploring your data and trying to get a better feel for its characteristics.

Bar Charts

Bar charts are handy when you want to show how things are spread across different categories. They're great for visualising how many times each category appears in your data—like showing how many people fall into different age groups, how often certain products are sold, or how many votes each option got in a poll.

Bar charts are used when your data is split into categories or groups.

Histograms

Histograms are perfect when you want to see how data is spread across different ranges or intervals, especially for numbers. Instead of categories like in a bar chart, histograms group continuous data into "bins." So, if you’re looking at something like ages or exam scores, a histogram can show you how many data points fall within specific ranges—like how many people are between 20-30 years old, or how many students scored between 70-80 on a test.

You’ll use a histogram when you want to get a sense of the distribution of numerical data and spot patterns like whether most of the data is clumped in one area, spread out, or if there are any peaks or gaps. It’s great for seeing the shape of your data, like whether it’s normally distributed or skewed.

Density Curve

A density curve is great when you want to see the overall shape of how your data is spread out, but in a smooth, continuous way. Instead of grouping data into bins like a histogram, a density curve estimates the probability of values falling in different areas, giving you a smooth line that shows where the data is more concentrated.

You’d use a density curve when you want to get a feel for the distribution of continuous data without the blocky look of a histogram. It’s perfect for spotting where the data tends to cluster, and it’s useful for seeing patterns like whether the data is spread evenly, has multiple peaks, or is skewed in one direction.

Box Plot

Box plots are useful when you want to get a quick snapshot of how your data is spread out, including the middle range, extremes, and any outliers. They show you the median, the spread of the middle 50% of the data (called the interquartile range), the minimum and maximum values and outliers.

You’d use a box plot when you want to compare distributions across different groups or just get a summary of the data’s spread without diving into every single detail. It’s especially handy for spotting outliers and seeing if the data is skewed or evenly distributed. 

Violin Plot

Violin plots are a combination of box plots and density curves. They show the same summary as a box plot (median, quartiles, etc.) but with added detail on the shape of the data’s distribution, thanks to the smooth curve on each side, which is essentially a density curve. That curve tells you where the data is concentrated.

You’d use a violin plot when you want the detailed distribution view of a density curve, but still want the clear summary stats from a box plot. They’re especially helpful when comparing multiple groups, as you can see both the spread of the data and its shape at the same time. Violin plots make multimodal distributions easy to spot.

Charts for comparing values between groups

Another very common application for a data visualisation is to compare values between distinct groups. This is frequently combined with other roles for data visualisation, like showing change over time, or looking at how data is distributed.

Bar Charts

Yet another use for the versatile, yet simple bar chart. The method is simple – the length (or height) of the bars show the values of the variable. Grouped and stacked bar charts allow us to include subcategories as well.

As always, bar charts are used to show the value of groups or categories rather than continuous numeric data.

Dot Plot

Think of dot plots like a minimalistic alteration of a bar chart. The dots are where the tops of the bars would be.

Dot plots are primarily used when including a vertical baseline, such as a zero line would not be meaningful. 

Line Chart

We already know that line charts are used to show how the value of a variable changes over time, and with multiple lines, we can compare the values and changes of multiple variables. 

Box Plots and Violin Plots

These charts are used to compare data distribution of different categories or groups.

‍

Charts for observing the relationship between variables

Scatter Plot

The scatter plot is the go-to chart to observe a relationship between 2 variables. You plot a series of points in a standard, 2D graph with 2 variables on the x and y axis, and there you have it – a scatter plot.

‍

Scatter plots are also useful for showing clusters, gaps and outliers.

‍

Scatter plots work best when you have many data points.

Bubble Charts

When you’ve got a third variable with numeric values, a common way to show it in a scatter plot is by changing the size of the points. This gives it the name, “bubble chart”. The size of the bubble gives you the value of the third variable.

Gradient Scatter Plot

Another alternative when you’ve got a third numeric variable is to represent the value of the third variable in a gradient form. The standard practice is to use a single colour, where dark colours represent higher values, and light colours represent lower values.

‍

Coloured Scatter Plot

This option is used when you have a categorical third variable. The different colours represent different categories. These are often used in conjunction with bubble charts.

Shapes

In instances where colours are not possible, like black and white printed charts, shapes are used to distinguish between categories.

Dual Axis Plot

Dual axis plots use 2 y-axes (and rarely, x-axes) on the left and right (or top and down in the case of x-axes) which are synchronized. The most common application for this is a line and bar chart, which is superior to grouped bar charts at showing the relationship between 2 variables.

Heatmaps

Heatmaps are also excellent choices for showing the relationship between variables. A correlogram is a specific type of heatmap designed for this very purpose. 

‍

Charts for showing geographical data

Geospatial Heatmaps

The grid isn’t the only way a heatmap can be visualised. Rather than plot colours against a 2-dimensional graph, geospatial heatmaps plot colours against a geographic map. 

The above image shows a zoomed in geospatial heatmap. The chart highlights the most photographed areas of Edinburgh, with red indicating zones of high photographic activity, gradually fading to blue in less photographed areas. Notably, the historic city centre, shown in vibrant red and orange, stands out as one of the most photographed locations, while other parts of the city attract less attention.

A birds-eye view of a heatmap would look something like this:

‍

Choropleth

Another geographical variant of a heatmap, choropleth maps use distinct geographical areas filled with colours that show aggregated values. 

The graph above shows a bivariate choropleth. 

Conclusion

Choosing the right data visualisation graph differentiates a great visualiser from a competent one. It’s not only important to understand the visualisation and the data itself, but also the target audience. Keep these in mind, and you’ll be well-equipped to make informed decisions and create visualisations that stand out (in a good way, of course).

This article is part of our visualisation series. For a more in-depth look at each type of visualisation, check out our comprehensive guides.

‍