3

Question to UX/Data/Subject Matter Experts: I'm working on a Call Analysis app where one of the chart is to present the aggregate value of "Call Quality by Client Health by Call Type".

These are the values for each of the parameters:

  • Call Quality: Good, Fair, Poor, Unknown (> 70 is good, 30 and 70 is fair, < 30 is poor)
  • Client Health (eg: PC, laptop, mobile etc): Good, Fair, Poor, Unknown (> 50 is good, 25 to 50 is fair, < 25 is poor)
  • Call Type: Audio, Video, Desktop Sharing

If you look in this chart below, in the x-axis I have Call Type, and in y-axis I have Client health on the left side and Client Quality on the right side.

If you look at the Audio vertical bars and Client Health ... What we are planning to show on this chart are aggregate numbers ... How do we do that in the current instance where Audio is a vertical bar and Client Health on the left side Y-Axis (leaving aside Call Quality for now) is divided into Good, Average and Poor?

How can we say that there are 40 Audio calls with good CH, 30 with Average CH and 40 with Poor CH? Likewise for Call Quality also?

enter image description here

Is there any better way of representing this data?

Thanks in advance!


Updated:

Bubble Charts as suggested by Crissov

Split/Panel Chart - Is this doable?

  • Can you explain what client health means and the relationship between client health and call quality? I feel like this is important to understanding the best way to represent the data. Also, it would help to understand the message or information that you want to convey as there are several different datasets being combined together so the message sometimes gets lost. – Michael Lai Jul 20 '16 at 21:39
  • 1
    I like your new representation much much better! It is clearer and visually appealing!!! Great job!!! – Wen G Jul 25 '16 at 14:00
  • Is split/panel type chart (as shown in the mock attached) acceptable? I have seen broken y axis chart like this. But not sure if high charts provides this in the first place. – essdeepee Jul 26 '16 at 2:48
3

Bubble chart

Put Client Health and Quality on the x and y axes. Each Call Type is represented by a bubble in the resulting coordinate system, i.e. a circle whose radius is relative to the total amount of such calls.

You can add some labels or colors to indicate the “good” and “poor” areas of the sector. Instead of actual circles you could also scale icons like 🎙, 🎥 and 🖥.

  • Please note that this recommendation wa sbased on the assumption that each Call Type would be represented by a single cumulated bubble. – Crissov Jul 25 '16 at 20:39
  • Could you elaborate on this? How will a single cumulated bubble tell me that "30 audio calls with fair client health had good call quality" If you don't mind a rough sketch would help. Also, do you think that the bubble chart that I put together is hard to read? Not sure how to simplify this, because people complaints that the chart looks complex and hard to read because of the number of bubbles in the chart? – essdeepee Aug 16 '16 at 7:48
  • @essdeepee Do you have example data, e.g. in a Google Spreadsheet? – Crissov Aug 16 '16 at 10:30
  • [ { "call-type": "AU", "call_quality": "FR", "client-health": "FR", "count": 1 }, ] AU - Audio, FR - Fair Will this help? – essdeepee Aug 16 '16 at 17:02
  • Not really, sorry @essdeepee, I meant something like the actual data you used for the charts in the question. Anyhow, if each Call Type was represented by a single bubble it could still expand to constituents on demand. Gapminder supports that e.g. for continents and their countries. – Crissov Aug 16 '16 at 20:05
2

What are you looking for is Radar chart.

Quote from Wikipedia

A radar chart is a graphical method of displaying multivariate data in the form of a two-dimensional chart of three or more quantitative variables represented on axes starting from the same point. The relative position and angle of the axes is typically uninformative.

1

Have you considered splitting the information into several charts? The current chart looks sophisticated, which may show that you are clever and have great data analyzing skills. However, it may be hard to read by your intended audiences.

The whole purpose for us to do the data analysis is for other people to understand the data, not to show that we can produce beautiful charts ;) So based on my experience, we should avoid having more than two key points in one chart. The problem with your representation of the call analysis data is you have too many information going on in one chart. Thus I suggested splitting your analysis results into chunks. For example, you could use a pie chart to show the percentage of each call type, and then three other pie charts to show the percentage of client health with each call type separately. The same logic goes with the call quality.

If the three factors indeed cannot be separated in your case, then maybe consider a three-dimensional chart with an x-axis, a y-axis, and a z-axis (represented by color density). But it would be not intuitively readable by people who is not familiar with the raw data and the analysis. So I wouldn't recommend using that. After all, we are designing the chart to let others read, and grasp our ideas, usually in just seconds. If you still would like to go along that path, I am happy to talk to you about potential solutions.

You seem to have good data to work with :) Good luck with the analysis and presentation! :)

  • Sometimes combining charts facilitates better data analysis – Andrew Martin Jul 20 '16 at 21:08
  • 1
    Yes you are totally right! And it is a good idea to use a dual-y chart. However, we may want to keep in mind that sometime dual-y charts could be confusing. Here is an article that gave out great tips on dual-y charts. Hope it helps :) scribblelive.com/blog/2013/06/18/… – Wen G Jul 21 '16 at 2:23
  • The main objection in that article seems to be when the two Y axes are different or at different scales. OPs chart has both Y axes calibrated to exactly the same scale and units this making it much easier to read and analyse than two separate charts. – Andrew Martin Jul 25 '16 at 5:31
1

There are a few fancy solutions here but, based on the data you presented, I think the best option is probably a Dual-Y Column Chart.

Since the graduations for both Client Health and Call Quality are effectively the same ('Good', 'Fair', 'Poor', 'Unknown') you can easily represent both of these values on the Y axis without causing confusion with scales. Simply representing the two different values with different coloured (and textured if possible) columns for each Call Type should be extremely easy to read.

  • In this case, I would only get to know "50 calls with poor client health" or "25 calls with fair client quality" but not "30 calls with fair client health had good call quality". – essdeepee Jul 26 '16 at 9:51
0

enter image description here

A good rule of thumb to start out with is one or two extra dimensions, such as a color division on a bar chart, or a bubble chart with dimensions plotted by size and color (see an example here from www.inetsoft.com). Go beyond and you are entering the danger zone. So if you’re considering adding more than two extra dimensions to a chart, ask yourself, are you actually hoping to find patterns in the data across that many dimensions? Or are you just trying to save screen real estate? Screen space is often the reason why charts end up becoming too cluttered.

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.