Which chart type works best for your data?

Purpose

It starts with a simple question: what is the purpose of your chart? Which aspect of your story do you aim to illustrate with your chart?

In essence, there are 6 guiding principles that can steer data visualisation:

1. Comparison
2. Composition
3. Distribution
4. Deviation
5. Relationship
6. Trend

Let’s look at each of them and see how these principles can help you create better charts and graphs.
‍

1. Comparison

You want to compare one set of values with another. A comparison tries to set one set of variables apart from another, and display how those variables interact, like the number of visitors to five competing websites in a certain period of time.

These 4 chart types can best achieve this:

Column chart

‍Column charts are the standard for showing chronological data, such as growth over specific periods of time, and for comparing data across categories. Grouping a small number of subsets (max. 6) gives additional information about the distribution of data over smaller units, i.e. quarters or regional offices.

Bar chart

This chart type is typically suitable for comparing the size of data (horizontal axis) for a medium to large data set (vertical axis). Keep data in an order that makes sense. Either list by value or, if that’s not the strength, choose a logic for the labels that makes sense, like listing them alphabetically. The added benefit of this chart type is that it accommodates for long data labels.

Line chart

‍The purpose of a line chart is to show trends, accelerations (or decelerations) and volatility. They display relationships in how data changes over a period of time. Choose a line chart also when the number of data points is very high and a column or bar chart would look too cluttered.

Two-axis chart

‍This combines two chart types in a single chart: a column chart series and a line graph series. There are two Y-axes: each axis has its own unit and magnitude, and each data series conforms to one of these axes. This chart type is ideal if you have two (or more) variables that you want to show in the same period of time, like temperature and rainfall, and how they correlate over time.
‍

2. Composition

You want to show how various parts of the data comprise the whole. A composition tries to collect different types of information that make up a whole and display them together in one chart.

These 4 chart types can best achieve this:

Pie or donut chart

‍This chart type is typically suitable when you need to visualise a part to whole relationship. Pie charts work best if you have no more than 6 categories and when the categories have clearly distinguishable values. The donut chart merely adds a hole to the middle of the pie chart.

Stacked column chart

‍This is a variation of the column chart where each column is broken into smaller units (i.e. quarters or regional offices) showing the development of the whole as well as that of the smaller units within the data set. Use a 100% stacked column chart when you want to focus on the differences between units.

Stacked area chart

‍This is a variation of the stacked column chart showing the development of the whole as well as that of the smaller units within the data set in a more continuous flow. Use a 100% stacked area chart when you only want to show relative differences.

Waterfall chart

This chart type is used to visualise changes in performance. A typical use is to show how an initial value is increased and decreased by a series of intermediate values, leading to a final value.

3. Distribution

You want to show the distribution of a set of values. A distribution tries to lay out a collection of related or unrelated information in order to find out how it correlates, if at all, and to understand if there's any interaction between the variables. For instance, in order to detect the outliers or to define normal ranges.

These 4 chart types can best achieve this:

Column chart

Column charts are the standard for showing chronological data, such as growth over specific periods of time, and for comparing data across categories. Grouping a small number of sub-sets (max. 6) gives additional information about the distribution of data over smaller units, i.e. quarters or regional offices.

Scatter plot

This chart type works best to find and show the correlations in a fairly large data set. The data sets need to be in pairs with a dependent variable and an independent variable. The dependent (the one the other relies on) becomes the y axis and the independent, the x. Adding a trend line will help show the correlation and how statistically significant it is.

Line chart

The purpose of a line chart is to show accelerations (or decelerations) and volatility. They display relationships in how data changes over a period of time. Choose a line chart also when the number of data points is very high and a column or bar chart would look too cluttered.

Two-axis chart

This combines two chart types in a single chart: a column chart series and a line graph series. There are two Y-axes: each axis has its own unit and magnitude, and each data series conforms to one of these axes. This chart type is ideal if you have two (or more) variables that you want to show in the same period of time, like temperature and rainfall, and how they change over time.
‍

4. Deviation

You want to show which values deviate from the norm. This is a variation of 1. Comparison, but aims to highlight the outliers or problem areas that need special attention.

These 4 chart types can best achieve this:

Column chart

Column charts are the standard for showing chronological data, such as growth over specific periods of time, and for comparing data across categories. Deviations will be clearly visible if one or more columns are much shorter or longer than the others.

Bar chart

This chart type is typically suitable for comparing the size of data (horizontal axis) for a medium to large data set (vertical axis). Keep data in an order that makes sense. Deviations will be clearly visible if one or more bars are much shorter or longer than the others.

Line chart

Line charts are more suitable than column charts or bar charts for showing deviations when the differences are fairly small or if there are a lot of data points.

Area chart

This is a visual alternative for the line chart, showing not only the development of data on a graph over time but also the volume of that data.

5. Relationship

You want to show the relationship between series of variables. This is again a variation of 1. Comparison, but takes the comparison one big step further in aiming to show the direct correlation (relationship) between the compared variables.

These 3 chart types can best achieve this:

Scatter plot

‍This chart type works best to find and show the correlations between two variables. The data sets need to be in pairs with a dependent variable and an independent variable. The dependent (the one the other relies on) becomes the y axis and the independent, the x. Adding a trend line will help show the correlation and how statistically significant it is.

Bubble chart

‍This chart type is a variation on the scatter plot, adding a third (dependent) variable which indicates volume or size.

Two-axis chart

‍This combines two chart types in a single chart: a column chart series and a line graph series. There are two Y-axis: each axis has its own unit and magnitude, and each data series conforms to one of these axes.
‍

6. Trend

You want to understand the trend over time of some data variables. You will need to plot the time line on the X-axis.

These 4 chart types can best achieve this:

Line chart

The line chart displays relationships and the resulting derivative trend of how data changes over a period of time. Choose a line chart also when the number of data points is very high and a column or bar chart would look too cluttered.

Column chart

‍Column charts are the standard for showing chronological data, such as growth over specific periods of time, and for comparing data across categories. Grouping a small number of sub-sets (max. 6) gives additional information about the distribution of data over smaller units, i.e. quarters or regional offices.

Scatter plot

‍This chart type works best to find and show the correlations between two variables. The data sets need to be in pairs with a dependent variable and an independent variable, in this case time. The dependent (the one the other relies on) becomes the y axis and time becomes the x. Adding a linearly plotted average will show the trend.

Area chart

‍This is a visual alternative for the line chart, showing not only the development of data on a graph over time but also the volume of that data.

Concluding

This should be a fairly concise overview of data visualisation purposes and of the most often used chart types related to those purposes. This overview will surely help those who venture into the rich world of data visualisation.

‍