7 Best Practices of Data Visualization

In order to make the best out of your data, you need to adhere to certain best practices that would make your graphs or charts more attractive and eye-catching. This in turn would help in better decision making and better performance.

1.Encode Meaning

The main intention is to ensure that your data stands out and visually mean something. Our mind is good at perceiving differences and contrasts, as our eyes get drawn to familiar patterns. There are two types of data; one being the data that is really easy for our mind to grasp at once and the other is the data that may take more time than usual to understand. They are very precise data and not very precise data.

Precise data our very easy for our mind to understand. Data such as length or 2-D position can be easily grasped as it is to the point.

Not very precise data such as width, size or intensity are very difficult for our mind to understand. Hence, it will take more time for our brain to process the data.

Our brain also can naturally extract meaning from patterns. Patterns can tell us if a trend is going up, going down, or if it has been flat or fluctuating. It can also help us to identify clusters and gaps in between.

Revealing patterns is what any good data visualization should do.

2. Choose the Right Chart

Choosing the right chart type is the most crucial part of it all. If you do not, then the purpose of the chart will also be gone for a six! It is all about choosing the right chart type to explain your data more accurately. Right type of chart really depends on what type of question you’re trying to answer through the chart. It is always good to align your chart with what you really want to show.

  • Big Number Chart

This chart is very useful and it gives the viewer an answer with a single metric. It is most commonly used in aggregation such as to display the sales this much, or it could also be used to display an average, minimum or median.

It is good to answer direct questions such as how much? The example shown above displays the YouTube views this month. It can also be used to monitor KPIs of an organization, where it clearly displays where the organization stands. The key trick for big number charts is to make it big and beautiful. This will ensure that the number or the question you are trying to answer really stands out.

Big number charts can also be used with a combination of other charts.

The chart on to the left displays how sales have changed over time. Here, the big number chart shows the total yearly sales.

It is always nice to use big number charts with images. It makes your report more pleasant and enjoyable to look at.

  • Line charts

These type of charts are a frequently used chart type. It is best to use when displaying metric over a period of time. There can be many variations of Line charts. It is illustrated in the diagram below.

The first chart represents a line chart chart. It basically is used to highlight trends with aggregated data. The second type of chart displays an example of a step chart. These type of charts represents granular time data. It is best to represent where there are long time intervals where there is no change and suddenly there is a big change in the trend. The final line chart type is Area charts. It shades the area underneath the line chart.

When you add additional metrics or dimensions to your visualization, then it is better to use Stack Charts.The chart on the left side represents a stack chart where it allows you to see what makes up the total area of the chart.

The graph on the right side represents a Multi-line Chart. This chart type is used compare diff categories against each other. The start point for the lines arise at the same point in the y-axis.

  • Column Charts/ Bar Charts

This is once again considered as the most used chart type. If the requirement is to display trends and categorical data, then the best type of chart to use are Column charts. You need at least one metric or one dimension to build up a column chart.

Column charts are best visualized when the number of categories are limited. Depends on the no. of categories we have.

The chart above compares different sales persons against each other. It is evident that the clarity of this chart is less as the full name of the sales persons has not been displayed as all the names cannot be fitted accurately to the graph. Therefore, a point to note when drawing column charts are when the size and the data labels increase, the effectiveness of the of the chart decreases.

As a solution what you can do is, you can make it more visible by rotating the horizontal text. However, this practice is not the best as it is not very user-friendly. The person who is reading the graph will have to tilt his/her head to the right and see the data labels. I’m sure if they continue to read the graph for a long time tilting their heads to one side, their going to end up having neck pains, and I’m sure this is not what’s expected!

So, when you have data labels that cannot be fitted in to a column chart, we use a Bar Chart. This is definitely easier to read than tilting your head to one side for a long time. It is very user-friendly and has better clarity when reading a graph.

As a general rule of thumb, the best practice to be followed when using column and bar charts is to use a column chart for less than ten (10) categories and to use a bar chart if there are more then ten categories.

  • Bar Charts with Multiple Dimensions

If there is a requirement to add additional dimensions to bar charts and column charts then it is best to consider using Stacked Column charts or Clustered charts.

Stacked Column charts (refer left chart) can be used to see how the total is made up of those different categories. Clustered charts (refer right chart) is better for comparing those different categories against each other.

The power of cluster charts tend to decrease when the number of categories increase. As displayed in the above chart, it is significantly harder to read when there are colors all over the place. If we have to read a chart having such an output, it is difficult for our mind to read and try to understand what is going on.

  • Trellis chart

When your chart starts to look like a double rainbow, a Trellis chart can be used. Trellis charts are much similar to clustered bar charts. However, the categories in this chart are grouped in a more intuitive way, rather than been broken down by color.

The chart on the left hand side displays the sum of revenue region by revenue type and the chart on to your right displays revenue by region type. As you can see, it is more easier to read and understand as the colors used for the chart is static and does not look like a rainbow.

  • Pie Charts

When to use Pie Charts?

Pie charts are most frequently misused. To illustrate this chart, a metric and a dimension is mandatory. Pie charts should not be used if the categories are of similar sizes or when the number of categories increases.

As you can see below, sales of Central Asia and South America region are almost equal and by the look of it we cannot depict which region has has made more sales. Therefore, it is best to use a column chart as displayed on the right hand side chart, we can conclude Central Asia has more sales by region than South America.

It is best to use Pie charts when you are trying to compare two categories. By looking at the graphs below, it can be concluded that women are more likely to attend day classes, while men are more likely to attend evening classes. Wonder why? Another option that can be used instead of pie charts are ring charts which is also known as a doughnut. More white space is available in a doughnut chart and therefore you can use a big number chart to depict the attendees.

  • Maps

The beauty of Maps are that it shows you where things are and where things are not. This type of visualization is good for identifying opportunities like “where my next door is going to be?” In order to make use of maps, the data should be encoded.

There are also other different types of maps such as;

  1. Bubble maps
  2. Thematic maps/ GIS maps
  3. Heat maps
Bubble Maps
Thematic/ GIS Maps
Heat Maps
  • Histograms

Histograms helps to depict the underlying frequency of a set of continuous data.

It also helps to look at the distribution.

  • Tree maps

Allows to see the total is made up of the sum of its parts.

  • Meters

These types of visualization methods are good for giving people a good glance on how they are performing vs a certain organization KPI. The meters can be formatted by applying conditional formatting to suit the requirement.

One key point to be noted is to NOT use 3D when using data visualizations. It is not recommended as it restricts clarity of the graphs that you are trying to visualize.

That concludes the 2nd best practice when using data visualizations.

3. Format Styles

After we choose the right chart type, next is to choose proper formatting. By using the proper formatting styles, it makes your chart easier to understand.

  • Colors

Color is one of the different ways where you can encode your data with meaning. Color can be used as a metric.

  • Sequential — color pattern is ordered from highest to lowest
  • Diverging — two different color schemes having a critical mid point (eg: budget, population avg)
  • Dolor by dimension — each category will have a unique color

Note: the color gradients mentioned above can be changed from company to company

  • Grid lines

Grid lines is a type of formatting style that is good to communicate thresholds. But too many grid lines in a single chart makes it harder to read.

  • Labels

Labels allows you to clearly see the exact number of the specific category. It also supports the visual representation.

  • Legends

Legends are commonly used when in charts where color is used and when it is used as either a dimension or metric. We should avoid having legends with too many categories in it.

Another additional point that needs to be noted is consistency. Please like consistency. Our brain is designed in such a way to quickly notice patterns and outliers. If consistency is not maintained, it is an added distraction for the viewer. Graphs with different fonts, colors or patterns are added distraction and people viewing notices it. Therefore, it is highly recommended to maintain consistency throughout your graphs/ charts.

4. Add Clarity

Making sure there is clarity in your chart ensures that more meaning is given to your charts.

  • Titles and Descriptions

Main purpose of adding titles and descriptions to your charts is to frame the story. It also outlines the purpose of what your chart means. By the glance of it, what you are trying to deliver from the chart is understood, thereby making the task of explaining the chart easier. It uses data to support your story.

  • Sorting

As a best practice, it is always good to appropriately sort your charts, thereby giving additional clarity as to what you are trying to visualize through the chart.

According to the diagram below, there are three ways to sort your charts.

  1. Alphabetical
  2. Ascending
  3. Descending

Alphabetical sorting method helps people find what they are looking for. As the chart describes, the different countries can be easily compared as it has been alphabetically sorted.

Once you sort your data to the ascending order, it allows us to tell a story of the growth. For an example, product iterations or growth of a particular product or company can be easily interpreted once your data is sorted to the ascending order.

Descending order sorting arrangement helps you to compare largest to smallest or to show what is biggest. The biggest column is aligned to the y-axis of the graph, hence making it easier for viewing purposes.

5. Highlight What’s Important!

Now that we have chose our chart type, added formatting styles and clarity, it is now time to highlight what is important in your chart. This section will help you identify methods of emphasizing what really needs to be highlighted from your chart. By highlighting what is important, it is an easier process to direct the attention of the audience, and what is important can be understood easily.

  • Conditional Formatting

These type of formatting can be applied for the data where you are trying to show if your data is above or below a certain threshold level. Once it reaches that defined threshold level, it will change colors.

  • Reference Lines

Reference lines are a good way to interpret to see if you are on target for a goal. The threshold is based on a metric, and based on the metric you can set your reference line.

  • Annotations

Annotations allow you to attach descriptions and comments. If your purpose if to share knowledge throughout the organization, then this is the best method to highlight what is important.

  • Highlight trends

This helps to understand what trends are happening in your organization. This can be used to track trends of the performance of employees, products etc.

  • Project forecasts

Data inputted will actually predict or forecast what is going to happen.

6. Make it interactive

  • Time sliders & Unit selection

This feature allows people to explore data using visuals.

  1. Unit selection — For an example it will allow you to change the visual from month to quarter and the graphs will be changed accordingly.

2. Time sliders — Allows you to change a time period

  • Filters & Brushing

Applying filters to your charts is quite a popular interaction used by many individuals. It allows peoples to include or exclude data from the chart. Filters can be of two types;

  1. Filter by brushing
  2. Filter that appears above or below dashboards
  • Zooming & Panning

This feature is popular with Maps. It allows you to pan sideways or zoom in to data, where you can get a better view of the data you are looking for. It also gives the users the ability to re-scale and go from high-level to low-level.

  • Drill through

When you click on a specific piece of data within the chart, another report filtered by the data you selected will be opened.

  • Drill down

This feature provides you the ability of drilling down for predefined hierarchies. Hierarchies such as from regent to country to state. The following examples shows you the different beer types available, when you select Beer column to drill down.

  • Series selection

Allows you to change a series or metric. The graphs below changes the series from Invoiced to Cost, as you can see the graph changes accordingly. This is a powerful way for exploring your data more faster.

7. Share

Everything is set to visualize your data. Now comes the most important thing, and that is to share it. Data is more valuable when it is shared among a group of individuals and it will serve the purpose. Data visualization has no power outside of you, unless it is shared. This method can be used to influence people, as it is very powerful.

There can be many methods used to share your data visualizations, have a look at the image below.

It has come to the end of the post, and hope this was of much use for you/your organization for better performance.

For more information visit Data visualization best practices by YellowfinBI — https://www.youtube.com/watch?v=w1Ts0vT5Pf0