1

I have a bunch of data I would like to display in several static graphs. In this particular case, it's the amount of visits different parts of a website has throughout time, for analytics purposes.

For example, this graph shows the amount of visits each part has had since I started collecting data:

Graph

And this graph shows the amount of visits each part has had, during its first 72 hours, adjusted to local time:

Graph 2

I started by plotting each part of the website as a different line in a line graph, with the label for each line at the right end of the each line.

Initially it went pretty well, but then the labels started covering each other on similar data sets. I made it so the labels would never cover each other, while keeping the order of the last point in the data set. This made an improvement in some graphs, like the bottom one, but made a mess in other graphs where data is more crowded like the top one.

Now, the graph has 26 data sets, the labels are really crowded and this will only get worse as more and more data sets are added.

I know that it's a lot of data to cram into one single graph, but for me it's important both to know detailed information (such as when did one data set overtook another one), as well as general tendencies (such as the increase in pretty much every data set around the 5月24日 mark in the top graph)

Also, I've had to reuse colors for the datasets, and some parts become quite ambiguous, such as the two green lines crossing at the top right part of the top graph. I could reassign the colors based on another criteria, but since the order of the lines varies from graph to graph, and I would like to keep datasets having the same color across graphs, this problem will likely persist.

I would like to know if there is a more intelligent way of showing this data. Ways of improving the line graph would be nice, but I am open to completely different ways of showing the same information.

Finally, I would prefer to keep the graph static. This is not a hard requirement, but more of a preference to keeping it simple. Solutions that involve interactivity are welcome as well.

Does anybody have any ideas how can I show this information in a better way?

  • "adding interactivity to these graphs is unfortunately not much of an option." Therefore, I think this is more of a graphic design problem than a interaction design problem. You could try posting your question here: graphicdesign.stackexchange.com – Ruudt Sep 11 '14 at 11:53
  • You may be right, as this is not about interaction. But I have seen several questions in here regarding how to better display data that stay in the static display domain, as in infographics. In any case, my preference for static is simply that, a preference, not a hard requirement. If you have an idea that involves interactivity, it would be welcome. – Panda Pajama Sep 11 '14 at 11:57
  • I don't have a total solution, sorry. However, here are some thoughts: You could solve the color problem by using dotted lines. For precise data you could consider using tables besides the graphs. – Ruudt Sep 11 '14 at 11:58
  • And as far is interaction goes: highlighting a line on roll-over could really help users identify the line alongside the other lines. Be aware of touch-based devices though (maybe highlight on touching the label?) – Ruudt Sep 11 '14 at 12:01
  • That's what I'm going on about. I will be adding interactivity to enhance the viewing, but I am looking for a much more general solution that applies even if there was no interaction. – Panda Pajama Sep 11 '14 at 13:17
1

I think what you are really after is a decision cube with a graphic part. That way, your users can slice and dice the data and only see the parts they're interested in.

This will not only simplify your graph (because, anyway, no one is going to want to look at all of it) but will also make it cleaner and more rational.

  • Welcome to the site, @Andrea. Can you clarify what you mean by "decision cube" and perhaps provide an example? – Graham Herrli Sep 11 '14 at 14:18
  • Thank you. IF you have never seen a decision cube, it's a sort of grid that hosts 3 different kinds of data: rows and columns plus a client area. You put all of your relevant values (visits, pages, etc.) in the client area but then you can drill down into a the rows and columns. You can also filter the rows and columns. It's a bit hard to start with, but once you get the hang of it it's extremely powerful. Also, most decision cube component packages allow for the creation of charts and they change as you drill down/up in the data. It's very powerful. – Andrea Raimondi Sep 11 '14 at 14:25
  • I have to admit that I had never heard of the term before, and most google results are clearly irrelevant. I suppose that is a term for a general system instead of a graph type, but I am interested in learning more about it. Do you have any images, videos or anything that explains the concept much more thoroughly? – Panda Pajama Sep 12 '14 at 2:55
  • Try this: technet.microsoft.com/en-us/library/aa216365%28v=sql.80%29.aspx . They are sometimes called "Pivot cubes". – Andrea Raimondi Sep 12 '14 at 8:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.