4

What data visualization would you say is suitable for my purpose? I'm looking for a chart or something similar to present queued items with different states, some is queued, some is processing and some has failed. I have almost 10 different queues, each with three different states.

Right now I have pie charts showing the relation of states in each queue, but this is not good for the comparison between the different queues.

A queue has 10 items total, and if 9 of them has failed it means that 90% of the pie will be red, which will make the user think that something has gone really wrong. But if there is a queue with 100 items and 10 of them is expired, the red area will be significantly smaller than the first queue, and the user will think there is a bigger problem with the first one, although there are a bigger number of failed items in the second queue.

I'm not a fan of the bar chart since it is more of a "information view" than a "monitoring view". Is there any other appropriate approaches?

Improve this question
1

2 Answers 2

Reset to default
1

As Michael Lai has already said, bar or column charts seem perfect for what you are describing.

You can use stacked charts to show the relationships between items within each queue:

enter image description here

I threw this together in about 2 minutes using Google Charts and didn't really have the time to go and fiddle with the colours or ordering but you could quite easily show the items stacked in each queue as waiting until they are either complete or failed. You could show failures, successes and waiting items in whatever order and whichever colour suits your particular case.

Improve this answer
0

I am not sure why you think bar charts are not good for monitoring views, since the monitoring view requires only information to be provided in real-time and in a way that is easy to pick out the trends and patterns. A stacked bar chart is in essence a pie chart stretched out linearly. I find that when you are comparing between more than two statuses that can have varying values it becomes easier to gauge values linearly compared to working out the proportions of a pie (i.e. the sectors). Bar charts also take up less space compared to having 5 different pie charts.

I have put in some sample data and created some visualization for 5 queues to give you an idea of the way different bar chart representations can provide different views and address the concerns you have around actual number of fails versus % of fails in each queue (I believe that both are equally important to know).

So on the left hand side are the charts of statuses grouped by the queues, with each grouping representing one queue, and the breakdown of the statuses expressed as absolute numbers (top left chart) or % (bottom left chart), while the right hand side shows a graph of queues grouped by statuses (top right chart). I think between the three graphs you can identify the queue with the most number of fails and the queue with the highest percentage of fails.

enter image description here

Improve this answer
3
  • Hi! Thank you for input! :)The biggest reason for why I don't think bar charts are suitable is that there are no "warning-feel" to it. If something is bad, it's hard to communicate the "you-have-to-do-something-to-make-it-right-again" with a bar chart, I think.
    – efrethe
    May 23, 2016 at 13:22
  • One of the problems I see with this is that it might not be suitable for bigger data. I have at least 10 queues and the number of items in each queue might be 5 but it might also be 10 000. The bar of 10 000 items, compared to the par of 5 items, will be too tall for the screen. Another way is to zoom out, so the user actually can see the entire bar of 10 000 items, but what about the bar with 5 items? Can the user even see that, or will it be too small?
    – efrethe
    May 23, 2016 at 13:22
  • 1
    @tessfred You can use the stacked bar chart to show the percentages, and then display the actual numbers not in a bar chart but as values instead (as a label). If you want to display absolute numbers for a very large range there are other things you can do like use a log scale. Regardless of the type of chart you use, having a very large number of datasets that can have a very large range of values will always be problematic. Imagine trying to fit 10 pie charts nicely in a responsive web page compared to just one bar chart for instance.
    – Michael Lai ♦
    May 23, 2016 at 14:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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