To realize
data visualization, the steps are not complicated, just like putting the elephant in the refrigerator, it is also divided into three steps: data preparation, visual design and content distribution.
1. Data preparation
The purpose of data analysis is to solve the problem, so as to provide companies and departments with analysis content with reference value. The basis for completing the above content is data and various data.
Data preparation is to clarify the data range and reduce the amount of data. Through collection, statistics, analysis and induction, we sort out the data result table we need.
To sort out the data content, the storage method is simply to use Excel, but also to use MySQL or Hive, etc., which needs to be selected according to the data volume and query performance requirements.
When data analysts use data tables, they can complete the data analysis work through single-table query or multi-table association, and then they can enter the visual design link.
2.
Visual design
If a worker wants to do his job well, he must first sharpen his weapon.
The most frequently used visualization tool may be Excel. In addition, it can also be implemented through R code, Python combined with JS, and code. However, these implementation methods have slightly higher learning and operation costs.
There are many visual chemicals available on the market today: Tableau, Haizhi BDP, Sailsoft FineBI, PowerBI, Netease, etc., through basic SQL capabilities combined with mouse drag and drop operations, you can complete visual design.
Use the mouse to drag and drop data table fields, you can achieve the setting of dimensions and indicators, you can also add filter conditions, and then combined with SQL queries, completed the production of visual reports.
Under the premise that the tools are becoming more and more practical, we also test our design and aesthetic capabilities. What we need to pay attention to are:
A. Reduce digital noise and choose the right chart
There are many options for charts, and it is not complicated charts that count as grade. The simplest way is, the simpler the chart, the more the user can understand what we want to express.
To summarize briefly:
Basic charts: line chart, scatter chart, bar chart, bar chart, bubble chart, combination chart, area chart, pie chart, etc.
Complex charts: dashboards, maps, flow maps, heat maps, tree charts, frame charts, funnel charts, Gantt charts, word cloud charts, radar charts, etc.
When choosing a chart, you need to understand the pros and cons of different charts and their suitable application scenarios. Try to reduce the data noise as much as possible and do n’t give users too much content at the same time.
But it is not impossible to use complex charts. In some scenarios, complex charts may be able to express the meaning behind the data more clearly. It is not bad to use it.
B. Reasonable color matching, multi-dimensional interactive matching
Regarding color matching, benevolence sees benevolence, and wise sees wisdom, there are various schools.
As far as the author is concerned, I prefer to use as few shades as possible, and use gradient colors as much as possible to ensure recognition, and the data to be compared can be selected from contrasting colors or complementary colors.
It is recommended that you go to some color matching websites, such as Material Palette, Material UI Colors, etc., or you can go to Dribbble, sugar pile, petals, or Thousand Maps.
The more you look at it, the more you will feel, and get an inspiration from accumulation.
As for multi-dimensional interaction, the following are common:
Screening: By setting filter conditions, the combination of different dimensions of data can be displayed;
Drilling: Realize hierarchical display of data at different levels, such as upper and lower departments;
Linkage: Through the change of one graph, the changes of other graphs are linked. For example, if you select a color block in the pie chart, the trend table at the bottom shows the trend of the corresponding content.
Of course, there are many other interactions, as long as they can show the user effective data content and meet the user's business needs, that is a good interaction.
Beautiful and easy to use is the meaning of data visualization.
3. Content distribution
The final product of data visualization is a report with many pictures and figures. We can communicate it to users in many ways. The easiest way is to directly provide source files or screenshots, but this is too bulky and inefficient.
Data platform products assume the responsibility of efficiently distributing reports, such as BI platform, mobile BI platform, etc., that is, the control of report viewing rights is realized, and the control of report data rights is also realized.
We can also directly use third-party tools to directly complete content distribution. Tools such as Tableau can achieve content distribution on the basis of localized deployment. The company still chooses to use its own research to build its own data visualization system.
