Many students learn Python for data analysis and data visualization. However, to do a good job of visualization, to accurately and intuitively display data and rules, it is not enough to master the technology at the code level.
An excellent data visualization chart is more than simply listing and summarizing data. The real value of data visualization is to design data displays that can be easily understood by readers. Therefore, in the design process, every choice should ultimately be based on the reader’s experience, not the individual chart maker.
So, today we put aside the code, only the visual chart design level, to share 30 tips summarized by the predecessors. By listing some common mistakes that are easy to be ignored, you can finally quickly improve and consolidate your visual production level.
1. Tips for making charts that you have to pay attention to
1. The baseline of the bar graph must start from scratch
The principle of the bar graph is to compare the size of the value by comparing the length of the bar. When the baseline is changed, the visual effect is distorted.
2. Use simple and readable fonts
Sometimes, typography can enhance the visual effect and add extra emotion and insight. But data visualization is not included. Stick to a simple sans serif font (usually the default font in programs such as Excel). Sans-serif fonts are those fonts with no feet on the edges of the text.
3. Moderate bar width
The interval between the bars should be 1/2 the width of the column.
4. Use 2D graphics
Although they look cool, 3d shapes can distort perception and therefore data. Insist on 2 dimensions to ensure accurate data.
5. Use table digital fonts
The table spacing gives all the numbers the same width so that they can be aligned with each other when arranged, making comparison easier. Most popular fonts have built-in tables. Not sure if the font is correct? Just see if the decimal point (or any number) is aligned.
6. A sense of unity
A sense of unity makes it easier for us to receive information: color, image, style, source...
Many people learn python and don't know where to start.
Many people learn python and after mastering the basic grammar, they don't know where to find cases to get started.
Many people who have done case studies do not know how to learn more advanced knowledge.
For these three types of people, I will provide you with a good learning platform, free to receive video tutorials, e-books, and course source code!
QQ group: 721195303
7. Don't be overly passionate about pie charts
Show the proportion of multiple blocks, the sum of all blocks (arcs) is equal to 100%. But it is best to avoid using this chart, because the naked eye is not sensitive to the size of the area.
8. Use coherent lines in a line chart
Dotted lines, dotted lines are easy to distract. On the contrary, using solid lines and colors makes it easier to distinguish the difference from each other.
9. Respect the proportion of the whole
When people choose multiple choices, there will be overlaps in proportions, and the sum of the percentages of different choices is greater than one. In order to avoid this situation, the ratio cannot be directly made into a statistical graph. Rather than presenting numerical values, some diagrams focus more on the relationship between parts and the whole.
10. Area and size visualization
Distinguish the length, height or area of the same type of graphics (such as column, ring, spider, etc.) to clearly express the comparison between the corresponding indicator values of different indicators. When making such data visualization graphics, mathematical formulas are used to express accurate scales and proportions.
11. Use size to visualize values
Size can help emphasize important information and add contextual hints. Using size to represent values is also very effective when used with maps. If you have multiple data points of the same size in your visualization, they will be mixed together, making it difficult to distinguish the values.
12. Use the same details
The more details (and numbers) added, the longer the brain will take to process it. Think about what you want to communicate with your data and what is the most effective way.
13. Use basic graphics
A good rule of thumb is that if you can’t understand efficiently, your readers or listeners may also be difficult to understand. Therefore, stick to the basic graphics: histogram, bar graph, Venn diagram, scatter plot and line graph.
14. Number of views
Limit the number of views in your visualization to three to four. If you add too many views, the big picture will be overwhelmed by detailed information.
2. About the color matching of the chart, you can refer to 5 guidelines
1. The shade of color
It is a common method of data visualization design to express the strength and size of the index value through the shade of the color. The user can see at a glance which part of the index data value is more prominent.
2. Use the same color
Too much use of colors will add unbearable weight to the data. On the contrary, designers should use the same color system or analog color.
3. Avoid using bright colors
Bright and bright colors are like capitalizing all the letters for emphasis. Your audience feels like you are selling them loudly. Monotonous colors, on the contrary, can be used well for data visualization, because they can allow your readers to understand your data without being overwhelmed by the data.
4. Labels are distinguished by different colors
In some cases, in a period of time or a series of values, we may measure different kinds of objects. For example, suppose we measure the weight of dogs and cats over the past 6 months. At the end of the experiment, we want to plot the weight of each animal, distinguishing cats and dogs with blue and red.
5. Number of colors
Do not use more than 6 colors on a picture.
Three, the standard visualization chart must have annotations
1. Explain the code
Through the combination of certain shapes, colors and geometric figures, the data is presented. In order for readers to read clearly, the chart designer must decode these graphics back to data values.
2. Axis label
This may seem unnecessary or not very helpful, but you can’t imagine that if your graph is a bit confusing, or the person who sees the data is not very familiar with it, how many times you will be asked what the x/y axis represents what. According to the previous two drawing examples, if you want to set a specific name for the axis.
3. Title
If we want to present the data to a third party, another basic but key point is to use the title, which is very similar to the previous axis labeling.
4. Annotate key elements
Usually, it is not very clear in itself to use scales only on the left and right sides of the chart. Labeling values on the graph is very useful for interpreting the graph.
5. Important view position
Place the most important view at the top or upper left corner. The eye usually notices this area first.
Four, excellent visualization charts, 6 principles to follow
1. The data is sorted in order
The data categories are sorted in alphabetical order, size order, or value, and guide readers to understand the data in a logical and intuitive way.
2. Compare the data
Comparison is a great way to show the difference in data, but if your readers can't easily see the difference, then your comparison is meaningless. Make sure that all the data are presented to the readers and choose the most appropriate comparison method.
3. Don't distort the data
Ensure that all visualization methods are accurate. For example, the size of the bubble chart should expand according to the area, not the diameter.
4. Display data
Let the reader see the data, this is the focus of visualization. Ensure that no data is lost or designed. For example, when using a standard area chart, transparency can be added to ensure that readers can see all the data.
5. Delete variables
Many times, too much information will affect the reader's attention. It is a good idea to remove implicit information from the visualization. In this case, I don't think we need to include the name of the variable in the axis.
6. Avoid data noise
Minimize or remove unimportant things. This includes weakening or removing graph lines, changing the color of axis and graph lines, and depicting spreadsheet rows in light gray. The "data ratio" can reach a very high level, and the audience will more easily understand the data situation.
V. Summary
Good data visualization should communicate data clearly and effectively through the use of graphics. Optimal visualization allows you to easily understand the data at a glance. They decompose complex information in a simple way so that the target audience can understand and make decisions based on it.
"The basic test of design is that it helps to understand the content, not its fashion.-Edward R. Tufte" Data visualization should adhere to this concept in particular. The goal is to enhance data through design, not to draw attention to the design itself.
As the saying goes, practice makes perfect. Think more about the details in the production process of data visualization every time, what details need to be paid attention to, and whether the processing of these details is reasonable, I believe your data visualization level will be greatly improved!