Understanding Visualization Context and Accordingly Select The Visual Types
Understanding Visualization Context and Accordingly Select The Visual Types
visual types
Relationship
Distribution
Composition
Comparison of data
By asking yourself what kind of story you want to tell with your data and
what message you want to convey to your audience, you’ll be able to
choose the right data visualization types for your project or initiative. And
ultimately, you’re likely to enjoy the results you’re aiming for.
For more on data storytelling, check out our full guide for dashboard
presentation and storytelling.
After gaining a greater level of insight into your audience as well as the
type of story you want to tell, you should decide whether you’re looking to
communicate a particular trend relating to a particular data set, over a
predetermined time period. What will work best?
st?
Line charts
Column charts
Area charts
Pie charts
Waterfall charts
Stacked charts
Map-based graphs (if your information is geographical)
While most types of data visualizations will allow you to compare two or
more trends or data sets, there are certain graphs or charts that will make
your message all the more powerful.
If your main goal is to show a direct comparison between two or more sets
of information, the best choice would be:
Bubble charts
Spider charts
Bar charts
Columned visualizations
Scatter plots
6. Is timeline a factor?
It’s important to ask yourself how you want to showcase your key
performance indicators as not only will this dictate the success of your
analytical activities but it will also determine how clear your visualizations
or data-driven stories resonate with your audience.
Consider what information you’re looking to gain from specific KPIs within
your campaigns or activities and how they will resonate with those that
you’ll be sharing the information with – if necessary, experiment with
different formats until you find the graphs or charts that fit your goals
exactly.
Here are two simple bonus questions to help make your data visualization
types even more successful:
“Visualization gives you answers to questions you didn’t know you had.” –
Ben Shneiderman
1) Number Chart
When to use Number Charts
What to avoid
Number charts are often the first thing people see and are the quickest to
read, so if there are too many, your narrative can get diluted. Using too
many can also make your dashboard a little superficial. If you want more
in-depth information, limit the number of number charts and leave room
for other types of data visualization that drill down a little deeper.
When you add a trend indicator, we suggest you compare numbers from
the same period. For example, if you are tracking total sales for the
current quarter, compare that data to the same quarter last year (or last
period – depending on your story). If you select a target manually
(perhaps you have no accurate past data), be sure to set realistic goals to
be able to get on top of your KPI management practice. Again, remember
to label the trend indicator clearly so your audience knows exactly what
they are looking at.
2) Line Chart
When to Use Line Charts
What to avoid
Too many lines (variables) can make your chart complicated and hard to
decipher. You may also find your audience constantly referencing the
legend to remind them which one they are looking at. If you have too
many variables, it’s time to consider a second (or even third) chart to tell
this story.
When it comes to layout, keep your numbers relevant. When you set up
your axis scale, keep it close to the highest data point. For example, if we
had set the y-axis above to track all the way to 200K (when our highest
data point is just over 90K), our chart would have been squished and hard
to read. The top half would have been wasted space, and the data
crammed. Let your data breathe a little!
A great feature of line charts is that you can combine them with other
types of data visualization, such as bar graphs. Using a double y-axis, one
for the bar graph and one for the line, allows you to show two elements of
your story in one graph. The primary y-axis below shows orders (bar
graph), and the secondary y-axis is sales totals (line). The metrics are
different and useful independently, but together, they tell a compelling
story.
3) Maps
Maps are great at visualizing your geographic data by location. The data
on a map is often displayed in a colored area map (like above) or a bubble
map. Because maps are so effective at telling a story, they are used by
governments, media, NGOs, nonprofits, public health departments – the
list goes on. Maps aren’t just for displaying data; they also direct action.
This was seen most recently through the Zika outbreak. Mapping the
spread of the disease has helped health officials track it and effectively
distribute resources where they are most needed.
Even if you aren’t saving the world from Zika, maps can help! For
example, they are great at comparing your organization’s sales in
different regions.
What to Avoid
Everyone loves maps. However, that doesn’t mean you always need to
display one. If the location isn’t a necessary part of your data-story, you
don’t need a map. They take up a lot of room, so only use them when
necessary. Also, don’t just fill your maps with data points. Clickhole did a
good job of satirizing this common data visualization type by placing 700
red dots on a map. Filling your map with data points doesn’t tell a data
story; it just overwhelms the audience.
4) Waterfall Chart
What to Avoid
5) Bar Graphs
There are three types of bar graphs: Horizontal (left to right), Column (up
and down), and Stacked (which can be either). Although all are in the
same chart family, each serves a distinct purpose.
Horizontal charts are perfect for comparative ranking, like a top-five list.
They are also useful if your data labels are really long. Keep them in an
order that makes sense, though. Either list by value (like we did above) or,
if that’s not the strength, choose a logic for the labels that makes sense,
like listing them alphabetically.
What to avoid
Because time is best expressed left to right, it’s better to leave showing
an evolution for the column chart. Also, like many charts, when you have
too many values, a horizontal bar graph quickly becomes cluttered.
b) Column Graphs
Column bar graphs are the standard for showing chronological data, such
as growth over specific periods, and for comparing data across categories.
In our sales data analysis example, Amount of Sales per Channel and
Country (last year), it is clear that we are comparing six regions and five
channels. The color coding keeps the audience clued in to which region
we are referencing, and the proper spacing shows the channels (good
design is at the heart of it all!). At a glance, you can see that SEM was the
highest-earning channel, and with a little effort, the Netherlands stands
out as the region that likely enjoyed the highest sales.
c) Stacked Column Chart
What to avoid
Aesthetically speaking, when you have too much data, columns become
very thin and ugly. This also leaves little room to properly label your chart.
Imagine we had 10 different age ranges per column. Some results, if not
most, would be only slivers. To make your chart easy to understand, use
good colors, proper spacing, and a balanced layout. This invites people to
look at your chart and even enjoy it. A pretty chart is a much nicer way to
consume data than squinting at a table.
6) Pie Charts
The much-maligned pie chart has had a bad couple of years. In fact, it has
become pretty cliché to talk about how bad pie charts are. We understand
the pie chart doesn’t do a lot, but it does do some things quite well. Pie
charts are useful for when demonstrating the proportional composition of
a particular variable over a static timeframe. Let’s look at some particular
cases:
What to Avoid
Data visualization guru Edward Tufte famously declared that “pie charts
are bad, and the only thing worse than one pie chart is lots of them.” We
already talked about the pros of pie charts and why we don’t adhere to
this strict no-pie-chart philosophy. We should also state that there are
plenty of instances where you should not use a pie chart. First off, pie
charts portray a stagnate time frame, so trending data is off the table with
this visualization method. Make sure your audience understands the
timeframe portrayed and try to document or label this applied filter
somewhere.
Pie charts are also not the best data visualization type to make precise
comparisons. This is especially true when there are multiple small pieces
to the pie. If you need to see that one slice is 1% larger than another, it’s
better to go with a bar chart. Another thing about multiple pieces to your
pie – you don’t want too many. Pie charts are most effective when just
displaying two portions. They lose presentation value after six segments.
After six, it is hard for the eyes to decipher what the slices proportion. It
also becomes difficult to label the pie chart, and valuable online
dashboard/reporting real estate is often wasted in the process.
This brings us to the last issue: circles take up space. If you are using
multiple pie charts in a dashboard, it is probably best to more effectively
combine the data in one chart. We recommend checking out the stacked
bar chart for these cases. You can also have a look at the different pie
charts that are commonly used and explore the disadvantages of pie
charts.
7) Gauge Charts
When to Use Gauge Charts
Gauge charts are also known as dial charts or speedometer charts. These
charts use needles and colors to show data similar to a reading on a
dial/speedometer, and they provide an easily digested visual. They are
great for displaying a single value/measure within a quantitative context,
such as to the previous period or to a target value. The gauge chart is
often used in executive dashboards and reports to display progress
against key business indicators. All you need to do is assign minimum and
maximum values and define a color range, and the gauge chart will
display an immediate trend indication.
What to Avoid
Gauge charts are great for KPIs and single data points. After that, they can
get a bit messy. With only one data point, you can’t easily compare
different variables. You also can’t trend data using gauge charts. All of this
makes taking actionable insight from a gauge chart difficult. Furthermore,
they take up a lot of space – if your live dashboard has precious real
estate, it may not be most efficient to fill it with multiple gauge charts.
You will likely get more bang for your buck using one chart to summarize
multiple KPIs.
8) Scatter Plot
When to use Scatter Plots
Scatter plot is not only fun to say – it’s what you need when looking for
the correlation in a 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-axis.
When the data is distributed on the plot, the results show the correlation
to be positive, negative (each to varying degrees), or nonexistent. Adding
a trend line will help show the correlation and how statistically significant
it is.
What to avoid
Scatter plots only work when you have a lot of data points and a
correlation. If you are only talking about a few pieces of information, a
scatter plot will be empty and pointless. The value comes through only
when there are enough data points to see clear results. If you only have a
little data or if your scatter plot shows no correlation at all, this chart has
no place on your business dashboard.
9) Spider Chart
When to Use Spider Charts
What to avoid
This is not the easiest chart to pull off, but it really impresses when done
correctly. Using this chart if you have more than five values in your
dimension (five “things” to evaluate) makes it hard to read, which can
make it pointless altogether. Whether you use solid lines or shaded areas,
too many layers are difficult to interpret. Naturally, it is not a choice when
you want to show time (the whole circular thing…).
10) Tables
Remember – just because you are using a table doesn’t mean it can’t be
visually pleasing. You can use various colors, border styles, font types,
number formats, and icons to highlight and present your data effectively.
What to Avoid
There are many reasons to use a table, but there are also many instances
where different data visualization types are a better choice. It all comes
down to our eyes and brain. Tables interact primarily with the verbal
system – we read tables. This reading includes processing the displayed
information in a sequential fashion. Users read down columns or across
rows of numbers, comparing one number to another. The keywords here
are reading, processing, and time. Tables take longer to digest.
Graphs, on the other hand, are perceived by our visual system. They give
numbers shape and form and tell a data story. They can present an
immense amount of data quickly and in an easy-to-consume fashion. If
data visualization is needed to identify patterns and relationships, a table
is not the best choice. Also, while it is fun to get creative with colors,
formatting, and icons, make sure your formatting and presentation
choices are increasing perception. Tables are hard enough to read as is!
The area chart is closely related to the line chart. Both chart types depict
a time-series relationship, show continuity across a dataset, and are good
for seeing trends rather than individual values. That said, there are some
key differences between the two. Because of these differences, “when to
use area charts” does not equal “when to use line charts.”
Line charts connect discrete but continuous data points through straight
line segments. This makes them effective for facilitating trend analyses.
Area charts technically do the same, except that the area below the
plotted lines is filled with color. In this case, an un-stacked area chart is
the same thing as a line chart – just with more coloring. The problem you
run into here is occlusion: when you start comparing multiple
variables/categories in an unstacked area chart, the upper layers obscure
the lower layers. You can play around with transparency, but after three
variables, un-stacked area charts are hard to read.
This brings us to the most commonly used area chart: the stacked area
chart. Like stacked bar charts, stacked area charts portray a part-to-whole
relationship. The total vertical of a stacked area chart shows the whole,
while the height of each different dataset shows the parts. For example, a
stacked area chart can show the sales trends for each region and the total
sales trend. There are two different stacked area chart types you can use
to portray the part-to-whole relationship.
Traditional Stacked Area Chart: The raw values are stacked, showing
how the whole changes over time.
What to Avoid
As we hinted earlier, for the most part, you should stay away from un-
stacked area charts. If you are just comparing 2-3 different variables that
don’t obscure each other, then go ahead. But in general, they are often
messy and don’t follow data visualization and dashboard design best
practices. When it comes to stacked area charts, don’t use them when
you don’t need to portray a part-to-whole relationship – use a line graph
instead. Also, if you are trying to compare 7+ series, a stacked area graph
becomes hard to read. In this case, you should once again turn to the line
graph.
Bubble charts, or bubble graphs, are among the best data visualization
graphs for comparing several values or sets of data at a glance. If you’re
looking to show the relationship between different product categories,
revenue streams, investment risks, costs, or anything similar, bubble
charts or plots are incredibly effective.
Here, you can tell that the TV & Home Theater product category has the
highest number of orders (around 3,000 as you can see from the number
scale on the left) as well as the highest profit margin, and therefore, it is
the biggest bubble on the chart. Comparatively, the camera category
shows the lowest number of orders in addition to the smallest profit
margin and naturally is the smallest bubble on the chart.
The bubble plot is extremely powerful for visualizing two or more variables
with multiple dimensions. And here, the bigger the bubble, the higher the
profit margin. Not only are bubble plots visually stimulating, but they are
also incredibly effective when building a comparative narrative for a
specific audience.
What to Avoid
It’s difficult to go too far wrong with bubble charts, but the most common
mistake with these types of data visualization graphs is focusing on
varying the “radius” of the values rather than the “area” they take up on
the chart. Doing so sometimes makes the bubbles on the plot
disproportionate on the graph, making the information misleading at a
glance. In short, your bubbles should be accurate in terms of size
compared to the values. Get this right, and you’ll get the results you
deserve.
Your part in creating an effective design for your data visualization graphs
boils down to choosing the right data visualization types to tell a coherent,
inspiring, and widely accessible story. Rarely will your audience
understand how much strategic thought you have put into your selection
of dashboards – as with many presentational elements, the design is often
undervalued. However, we understand how important this is, and we’re
here to lend a helping hand.
Aside from what we’ve covered in this guide, setting yourself up with a
varied and value-driven set of questions will give you the best start to
telling your data story the right way.
Is your audience going to be actively using the data featured within your
dashboard? Or will they merely be viewing it to make more informed
business decisions? Establishing this before you start designing your
charts will help you decide which KPIs you want to showcase and which
you want to highlight the most within your story.
To summarize, here are the top data visualization types you should know: