One figure means ten thousands words.
Data is abstract. Especially, huge volume data can hardly be analyzed by our brain to have an impression. However, data could be surprisingly beautiful in another sense. Many tools are designed to visualize data. They show us the beauty of data in an unbelievable manner. Fig. 1-3 are taken from Webdesignerdepot.com, where more examples are shown.
Visualization of data is widely used in high energy physics. The first thing learned by many graduate students is how to make plots using ROOT (or PAW in the past). Although rigorous statistical methods are used in physical analysis, scanning events by eyes is always necessary for suspicious ones. Fig. 4 shows an event in Super Kamiokande detector. It looks very beautiful, and actually very interesting, because it looks like a long-sought proton decay event.
Presenting the physical results with an elaborate figure, complex relation of data could be shown intuitively. There are countless examples in particle physics. Sudbury Neutrino Observatory presented their solar neutrino study as Fig. 5. Perhaps this result is not difficult to express in numbers. But this figure is much more impressive than numbers and text. It concentrates 10 years efforts of a hundred scientists. Whenever we see this figure, we know it is SNO.
No matter the results are relatively simple, or somewhat disorder, it is always worthy to spend a lot of time to compose a figure, which should be beautiful, intuitive, and rich in content and data relation.
Fig.1, Narratives 2.0 visualizes music. Different music tracks are segmented into single channels that are then shown in a fan-like structure.
Fig.2, Schemaball visualizes SQL database schema. Relationships are shown based on foreign keys within tables. It’s capable of showing schemas with hundreds of different tables and relationships.
Fig.3, The Hierarchical Structure of the Internet was a study that looks at how the Internet is organized, both in terms of structure and connectivity. It shows how the central core of the Internet is made up of about 80 core nodes, but that even if those nodes failed, 70% of the other nodes would still function via peer-to-peer connections.
Fig.4, Super Kamiokande event display. It looks like a proton decay event. The Cherenkov ring at the bottom is probably a positron. In the upper half, there may be two rings formed by pion decay to two gammas.
Fig.5, Flux of B-8 solar neutrinos which are muon or tau neutrinos vs flux of electron neutrinos, by SNO.
数据之美
一张图抵一万句话。
数据是抽象的，尤其是海量数据，人的大脑很难直接对大量数据进行分析并获得印象，然而从另一个角度看，数据也可以异常美丽，人们设计了很多工具，让枯燥的数据图形化，它们以令人难以置信的方式让我们看到了数据美丽的一面。图1-3摘自Webdesignerdepot，更多的例子请见原文。
数据的图形化在高能物理里广泛使用，以至于我们把它当成最自然的选择。许多研究生学的第一件事就是如何用ROOT（以前是PAW）来画图。在数据分析中尽管总是采用严格的统计方法，对可疑事例进行肉眼扫描仍然是必不可少的。图4是一个超级神岗事例，看上去非常漂亮，而且非常有趣，因为它像一个我们一直在寻找的质子衰变事例。
用一张精心构思的图来表达物理分析结果，复杂的数据关系可以得到直观的体现。这种例子在高能物理中不可胜数。例如Sudbury Neutrino Observatory (SNO)太阳中微子实验的结果见图5。也许这个结果用数据来表达并不困难，但远远没有这张图令人印象深刻。它浓缩了上百名科学家10年的努力。每当我们看到这张图时，只要一眼，我们就知道这是SNO。
无论结果是相对简单，还是显得杂乱无章，总是值得你花很多的时间去构思一张图，它应当漂亮、直观、包括尽可能多的内容和数据关系。
