Reasons to use dot graphs | Hacker News Reader

Reasons to use dot graphs

maartenlambrechts.be

54 pointssebkomianos9 comments

While doing scientific research I usually just use scatter plots. If you summarise things into statistics it's too easy to miss things. However, some of those plots were unreadable -- what the hell do the colours mean in that unemployment rates graph?

The graph is titled age, sex and region, while the axis labels are region and age, so the colors must represent gender.... But the importance of the gradients are beyond me.

Yes, the missing legend doesn't help readability unless you're familiar with Eurostat dataset terminology. Actually, that's total unemployment rate for >15y at a regional level. Ignore the sex and age part - it's the total number here. Colors stand for magnitude of total unemployment in a given region.

Here's the original version of the graph from Eurostat: http://ec.europa.eu/eurostat/cache/RSI/#?vis=nuts2.labourmar...

Strongly disagree with replacing bar charts by scattered points with arbitrary position on the horizontal axis (2 and 3). It is very difficult to disassociate horizontal proximity from vertical proximity, and the former is meaningless and distracting here.

Strongly disagree with your disagreement. The scattered points are very readable. They present more data, data that's simply missing from the bar charts. I always feel a little bit bad presenting, say, just a box and whisker plot when the original data is so much richer.

Interesting. I intuitively disregarded the horizontal placement of the dots at first sighting. Not sure why. The dot graphs are much easier for me to visually process and understand.

It's a very important message. We, humans, are great at aggregating visual patterns. So, most of the time, it makes sense to show raw data (and we can make our own conclusions/observations) rather than aggregate it to much (there is no possibility to go the other way).

As someone who works with data, this makes a lot of sense.

I challenge the applicability to people who do not. It is difficult for significant parts of the population to digest simplified data (or file their taxes). Reducing information density by turning dots into a histogram makes it easier to consume (and hide the weaknesses in your paper).

This brings us back to the challenge of accessibility of published papers vs scientific reporting.

I would suggest working in dots and including them as appendices, but producing simplified (honest) visualizations to reach a greater audience.

Hey if the author reads this nice article! On my iPhone 5S running the latest iOS (8.3) and using Safari the pictures seem to cut off a bit. It might be worth checking out if you're interested.

I enjoyed checking out the different uses of dots though. After the semester and once my blogs up I hope to do some data projects so it will be fun to work with different ways of displaying some of the information that comes out.

Reply on news.ycombinator.com