Visualizing Data with SAP Analytics Cloud

[MarcZauritz]

This article is part of a blog series on Dashboard Design BestPractices, created to assist you in designing and building meaningful dashboards with SAP Analytics Cloud. Click here to open the Dashboard Design Best-Practices blog series hub page in a new window

Content
This blog page focus on the art of visualizing data through the selection of appropriate
chart types to convey the intended message:

1. Comparison
2. Time-Analysis
3. Part-to whole
4. Cumulation
5. Deviation
6. Distribution
7. Correlation

First, let’s start with some general rules of thumb and quotas that you may consider when creating analytical content.

• Show the necessary information comprehensively.
• Everything on the screen has to make sense.
• Support data comparison capabilities.
• Everything that belongs together is grouped visually.
• Show variances.
• Use graphics whenever possible.
• Use consistent terms.
• Group with empty space.
• Provide clear support of the information hierarchy.
• Place time-series on the x-axies
• Visually highlight the most important information.
• Provide clear info where the action is actually needed.
• Focus on aesthetically pleasing, clear visual design.
• Avoid abbreviations (unless these are colloquial).
• Do not use semantic colors in the design, e.g., green, yellow, and red (unless it's company branding).
• Lead with purpose (guide the user to the appropriate actions).

Quotes

• „The pieces on the dashboard have to be arranged in relation to one another“ - Stephan View
• "Overview first, zoom and filter, then details-on-demand"
  -> Visual information seeking mantra by Ben Shneiderman" 
• "Perfection is achieved, not when there is nothing more to add,
  but when there is nothing left to take away"
  -> Antoine de Saint-Exupéry from "The little Prince"
• "Less is faster" - Hick´s Law

Chart Types

Chart types are important tools for data visualization because they help transform raw data into visual formats that are easy to understand, analyze, and communicate. The correct chart type depends on the type of data you're working with and the message you want to convey.

1. Comparison (Structure Analysis)

Choose Bar and Column Charts as the most effective visualization for structure analysis, also called nominal comparison or category comparison. 

Note:
When showing development over time, always use charts with horizontal category axes (Column Chart).
This is one of the most well-known guidelines established by IBCS (International Business Communication Standards) in its practical guidelines for the consistent design of reports.

They also propose the usage of semantic patterns to visually distinguish different types of values ​​when comparing them.

• Actual Values: What values are (solid pattern with dark color).
• Previous Year Values: What values were (solid pattern with neutral grey color).
• Forecast Values: What values might be (hatch pattern).
• Plan Values: What values should be (empty pattern).

2. Time Analysis 

Use a line chart if you want to emphasize the trend over time.

Use a column chart if you want to emphasize the values themselves.

Note that there are other time-related categories that have an intrinsic order and indicate progression or trend, such as age and age group.

3. Part-to whole

When thinking about visualizing parts of a whole, you are likely to turn to pie or donut chart.
However, in many cases bar chart is a far better option because you can:

• Compare parts to each other
• Display category labels and value labels associated with each part
• Display multiple values
• Display small values in a better way
• Cut down on the usage of colors and corresponding legends you have to use

Part-to-Whole Analysis by Bar Chart

Part-to-Whole Analysis by Stacked Percentage Bar Chart

 

Donut Chart

Donut charts work best with a small number of categories (5-7).
Too many categories can make the chart hard to interpret (see Pic. 1).

Note that when using pie or donut charts, you should be careful when enabling the "Top N" options, as this will change the original percentage values ​​and recalculate each segment's contribution to the total of the "Top N" categories (see Pic. 2).
Instead, use the group remaining function in chart widget menu to create a segment that represents the "Others" category (Pic. 3)

Tree Map

Tree map is a good visualization to present hierarchical data, with nested figures rendered by rectangles of different sizes.

In the example below, each rectangle signifies a department and its size in proportion to the number of department employees. You can use colors to differentiate employee genders, but here the chart displays only female employees. 

 

4. Cumulation

You can use waterfall charts to analyze cumulative values. A waterfall chart accumulates successive values and shows how the cumulative value changes from an initial state to a final one.

Variance waterfall chart in vertical orientation displays the variance between different scenarios in structure.
In the example below, the variance waterfall chart shows the net sales in Europe from previous year (PY) to this year (AC). You can see that this year's net sales value is 14 million USD higher than the previous year and the difference in Spain is the largest.

Variance waterfall chart in horizontal orientation displays the variances across different time periods.
In the example below, you can see that shipping cost keeps growing from Q1 to Q4 in the whole year.
Q4 is expanded to show changes in months.

 

5. Deviation

Deviation charts, also known as variance charts, are used to visually display how a value deviates from a baseline or expected target.
They are particularly helpful for understanding the differences between actual results and a reference point, such as a goal, target, or historical data. 

Actual and Plan Values and Deviation in Bar Charts Side by Side

Numeric and Percentage Deviation in Table with In-Cell Charts

Focus on Trend
If you want to focus on the trend of the variation, you can use a line chart as the examples below.

6. Distribution

Histogramm
A histogram displays the numbers of data points that lie within respective ranges, which is extremely helpful in emphasizing data frequency. In the left example, the axis displays age groups, and the columns show the numbers of employees that fall in different age ranges.

Column Chart
If your use case is to display the distribution of employees of each age group in percentage,
a column chart is a good option.

Stacked Column Chart
Distribution of Two Series by Stacked Column Chart.

You can use a frequency polygon with multiple lines when you want to compare the distribution of multiple sets of values.

Box plot
A Box Plot is an ideal option for displaying the concentration of data, which is constructed from minimum, first quartile, median, third quartile and maximum.

 

7. Correlation

Scatterplot and bubble chart are the most frequently used chart types to visualize correlation,
though line chart or bar chart side by side is also an acceptable solution.