How to choose a data visualization tool ? Before answering this question, another question needs to be answered, what do you need to use these data visualization tools for and for what purpose.
Maybe you have a complete idea that has been verified and needs to be presented in a more intuitive and understandable way to tell a logic or a story; maybe you have a lot of data, how do you want to discover from the data, Mining and showing the knowledge or insight behind some data; maybe you have all kinds of data, but you don't know data modeling, programming, or data cleaning, or even SQL optimization, etc. You need an easy-to-use data The visualization tool realizes the visualization of data by dragging and dropping, and can give the most suitable display graphics; there may be other various scenarios, but all data visualization tools have a scenario of their core services, which is beautiful and easy to use , Simple, Collaborative, Smart and more. Every data visualizer has a tab for positioning. The choice should be made by the core needs we need. Do a simple classification:
1. Easy-to-use and diverse display tools with clear goals, such as Tableau;
2. It can support flexible and customized display types, such as icon library D3;
3. Data exploration with unclear goals, such as explore of google spreadsheet;
4. There are industry demands that can both visualize analysis and data exploration, such as FanRuan FineBI ;
5. Data visualization according to industry or functional requirements, such as DOMO, Qlikview;
I saw that a netizen used 24 tools to make the same chart, compared 12 visualization software and 12 programming/chart libraries, and focused on the tool/chart library's emphasis, flexibility, chart innovation, and interactive effects. In general, I wrote an excellent article.
In May of this year, the girl set herself a challenge: try to use as many programming languages or software as possible for data visualization. To compare the tools, she repeated the same scatterplot with them. Based on the results, she also published two articles: one on making an identical graph with 12 softwares, and another on making an identical graph with 12 programming/graphing libraries. The image below shows her using 12 different software to create the same scatterplot:
这是12种编程/图表库制作出来的效果:
她从这些可视化软件/图表库中认识到:没有十全十美的工具,但是如果确立(可视化)目标,就能找到合适的工具去实现。下面是她在制作中曾遇到过的一些矛盾,也是数据可视化工作者常常遇到的情况。
1、分析 VS 展示:
是想使用工具(R, Python)来分析数据,还是更注重于构建可视化效果(D3.js, Illustrator)?有些BI工具(比如说FineBI, Tableau, Plotly)试图在这其中谋求平衡,既可分析又可展示。她根据分析和展示上的侧重性对可视化工具和编程语言们进行了排列:可以看到工具类的往往更注重展示,而编程类的比较平均,各有侧重。
2、数据管理
如果制作可视化的时候需要更改源数据怎么办?在这方面,这些工具或编程语言的灵活性如何?
低灵活性:比如在Illustrator中,即使你只是轻微修改了数据,也需要重头开始制作图表,这种工具还不方便进行数据管理。
中灵活性:比如在D3.js中,可以单独处理或修改数据,然后再重新导入数据文件来更新可视化结果。
高灵活性:比如在FineBI中,数据分析的处理如数据建模,数据清洗,甚至是SQL的优化,大数据量的处理都可以在一个平台完成,同时易用,拖拽就能完成数据的可视化。
3、传统图表 VS 创新图表:
如果你只需要基本的图表类型,如柱状图或折线图,Excel完全可以满足啦~但你如果想创建表现形式更为丰富的互动图表,比如点击可以出现酷炫的交互效果,像D3.js之类的编程语言就更适合啦,但是学习此类工具的门槛也往往更高,有着陡峭的学习曲线和冗长的代码。或者也可以使用Processing,用它制作这张散点图的代码长度只有D3.js的一半。还有Lyra,它不需要任何代码基础,但也可以让你轻松修改数据有关的视觉元素。
4、交互图表 VS 静态图表:
你是需要创造基于网页的交互图表(如D3.js, Highcharts能做到的),还是PDF/SVG/PNG形态的图表就能满足你 (R和Illustrator可以做到)?几年前,互动图表曾受到高度追捧,但现在关注焦点慢慢从“看起来怎么样”转移到“什么才更有意义”。对于分析部分,交互特性往往也是很有必要存在的。Plotly和R的库Ggvis就可以让读者轻松地将鼠标悬停在可视元素上来查看基础数据。下图是作者对于软件/编程的在静态和交互的划分:
本文首发CSDN:http://blog.csdn.net/liukecun0614/article/details/73087903