The process of using data visualization for data analysis is to "obtain boring flat data and turn it into reality through visualization." However, in the process of data analysis, many people have become keenly aware that visualization may become data Presented in the wrong way, some people even started to say: "Visualization is often used to disrupt the data analysis process."
With the rapid development of our Internet information, data is richer than ever and can be accessed at any time through the Internet, but when organizations publish misleading data visualization (intentional or unintentional) analysis, people’s sense of trust in the results of data analysis is Will be greatly reduced.
Therefore, in the process of using data visualization analysis, we need to pay attention to which design factors can misunderstand visualization, or how to clearly display the data analysis results through visualization.
1. Blind spots in data visualization
The excellence of the graphics is that it can provide the audience with the most creativity in the shortest time.
From a physiological point of view, human vision and cognition are one of the most incredible phenomena in nature:
- Light enters the eyes.
- The lens sends information from the light to the retina.
- The retina interprets information and sends signals to the optic nerve.
- The optic nerve transmits 20 megabits to the brain every second.
The leap from seeing to thinking is instantaneous. At such a rapid moment of observation and understanding, data visualization proves its value. Here, many visualizations tell viewers what they "should" see in the data, and the overworked brain cannot think carefully and will subconsciously agree.
To be fair, misleading visualizations are not always deliberately produced. It may be that some details are not noticed, and the errors caused by them appear. But even unconscious mistakes can mislead the audience. Eyes are impressive, and humans tend to reduce their own thinking in order to obtain quick information. Therefore, vision and cognition must be the key considerations in the design of all data visualization.
Two, 10 data visualization errors to avoid
1. Misleading color contrast
Color is one of the most convincing design elements. Even subtle shadow changes can cause strong emotional reactions. In data visualization, a high degree of color contrast may make viewers think that the difference in value is greater than the actual difference.
The high-contrast color scheme of this heat map makes the red area appear to be much larger than the value represented by the dark area, and the visualized heat map uses color to describe the magnitude of the value. Higher values are shown in orange and red, while lower values are shown in blue and green. The difference between the values may be small, but the color contrast creates a sense of heat and enhanced activity.
summary:
- Color is not just a way to distinguish data series.
- The high-contrast color pairing makes the viewer feel a greater degree of data difference.
Improper use of 2.3D graphics
3D graphics brings two serious problems in data visualization.
When a 3D graphic partially blocks another graphic, occlusion occurs. This is the result of simulating space in the natural world. In the natural world, the X, Y, and Z coordinates of the object are different. In data visualization, occlusion can obscure important data and create a wrong hierarchy. Among them, unoccluded graphics are particularly important.
Distortion occurs when 3D graphics shrink into or extend from the picture plane by shortening. In drawing, the pitch makes the objects look as if they occupy three-dimensional space, but in data visualization, it creates more erroneous hierarchies. The foreground graphics look larger, the background graphics are smaller, and the relationship between the data sequences is unnecessarily distorted.
summary:
- 3D graphics, they may obstruct important information and confuse the proportional relationship between data series.
- Unless you absolutely need 3D graphics, display the data in 2D.
3. Too much data
This is an eternal design problem-what to include and what to reduce in the process of seeking clear communication. Data visualization is no exception, especially when the data is rich and thought-provoking.
Attract attention? Deep conclusions can be drawn with the aid of a single visualization file.
Solve the problem? Humans do not have enough power to calculate the meaning of multiple values abstracted in visual form.
Excessive data in a single visualization file instantly overwhelms viewers
When the visualization contains too much data, the information will be overwhelmed, and the data will melt into graphics that most viewers cannot bear
summary:
- When information overload applies to data visualization. If too much is displayed at a time, the data area needs to be divided.
- Data analysis with multiple visualization objects is more effective.
4. Omit the baseline and cutoff scale
The data can sometimes vary greatly, such as when measuring income levels or voting habits based on geographic regions. In order to make the visualization effect more vivid or beautiful, the designer can choose to manipulate the scale value on the graph.
A common example is to omit the baseline or start the Y-axis somewhere above zero to make the data difference more obvious.
Another example is to truncate the X value of a data series to make it look comparable to a lower value series.
summary:
- Aesthetic attractiveness is subject to accurate data representation.
- It is unethical to ignore the baseline and truncated scale to deliberately exaggerate or minimize data differences.
5. Lean towards text
The suggested behavior is the art of persuasion. Tell someone what they should see in the image, they might see it. The text attached to the visualization (supporting copy, title, label, title) is designed to provide viewers with objective background information, not to manipulate their perception of the data.
summary:
- When drawing associations between data sets (often implying causality), biased text often appears.
- Usually, the biased text comes from the client, and the designer flags the question.
6. Choose the wrong visualization method
Each data visualization method has its own use case. For example, a pie chart is used to compare different parts of the whole. They apply to budget details and survey results (the same pie chart), but they are not meant to be compared between different data sets (different pie charts).
The pie chart can be used to visualize the earnings of three competing companies, but the bar chart can make the differences (or similarities) between the two companies more obvious. If the visualization is designed to show revenue over a period of time, a line chart will be better than a bar chart.
Pie charts are used to compare parts of the whole. Using them to compare different data sets (such as the revenue of different companies) will not give viewers any insight
summary:
- Data visualization methods are not one size fits all.
- Understand the variables that the visualization must convey.
7. Confusing correlations
Visualizing the correlation between data sets is a useful way to give viewers a broader understanding of the subject. One way to show the correlation is to overlay the data set on the same graph. When relevance is carefully considered, overlap can lead to aha moments. When there are too many overlays, it is difficult for viewers to draw connections.
It is also possible to visualize correlation in a way that incorrectly implies causality. A well-known example is that the increase in ice cream sales due to warm weather is related to a surge in violent crime.
summary:
- It may be helpful to highlight the relevance to multiple visualizations that are very close. This allows the viewer to evaluate the data and still establish a connection.
- It is worth repeating. Correlation does not mean causation.
8. Amplify favorable data
Data and time are inseparable. You can zoom in on the time range and display data that is conducive to a wider narrative. Visualizing financial performance is a common culprit. Consider a chart that shows strong numbers in a short period of time, making the business seem to be booming. Unfortunately, the shrinking shows that the company has only experienced a small rise and experienced a sharp and sustained decline.
- If the magnified visualization is inconsistent with the overall meaning of the data, please let the audience know.
9. Avoid common visual associations
Visual design elements affect human psychology. Icons, color schemes and fonts all have connotations that affect the viewer's perception. This rarely happens when designers ignore these connections or avoid them and instead use creative expressions.
Analyzing data visualization is mentally laborious. At critical moments of cognition, the brain may not need to spend time interpreting the reimagining of familiar design elements.
summary:
- There are countless ways to introduce creative experiments into data visualization. Don't distract the audience by forcing the audience to reinterpret common visual associations.
10. Use data visualization first
Data visualization shapes difficult-to-connect numbers. When the data is complex and there are multiple variables at work, they will reveal the meaning. But visualization is not always necessary.
If statistics can be used to communicate data clearly and concisely, then it should be. If the text description proves insightful, and the shape of the displayed data has little effect, then visualization is not needed.
summary:
- Data visualization is a communication tool. Like all tools, sometimes it is appropriate, while another tool is more appropriate.