During the exploration phase, you get to look at your data from a variety of angles and browse various facets, without having to dwell on charting standards and clarity. You understand a chart better because you know more about the data after you examine lots of other quickly generated charts. However, when you use graphics to present results to other people, you must make your graphics readable to those who don’t know your data as well as you do.
A common mistake is that all visualization must be simple, but this skips a step. You should actually design graphics that lend clarity, and that clarity can make a chart “simple” to read. However, sometimes a dataset is complex, so the visualization must be complex. The visualization might still work if it provides useful insights that you wouldn’t get from a spreadsheet.
As an effort toward clarity, people often preach removing all elements of a graphic that don’t help you interpret the data. When you “let the data speak,” you have done your job. This is fine, but it assumes the only goal of visualization is quick analytical insight, which is a small subset of what you can get out of data. It’s okay to ponder and reflect, and elements that are not helpful in one situation might be helpful in another.
That said, whether it’s a custom analysis tool or data art, make graphics to help others understand the data that you’ve abstracted, and try your best not to confuse your audience. How do you do this? ...