9

Background

I am working on real-time data acquisition and visualization software for a scientific measurement device. The software gathers data from the device, and plots the data in real time. Data is generally gathered for a few minutes, and then stops. The user then saves the data to a spreadsheet file for further processing.

In some cases (not rarely, but not all the time), the data from the measurement device will be "out of range", i.e. beyond the measurable limit. Forcing the data to the value of the measurable limit is misleading and erroneous since it's value is not known.

Most of the time when the data is out-of-range it's for a short period of time such as a few seconds or less. In some cases it will just be a transient spike lasting only a few data points. This is generally not a problem, but indicates that the user should probably change some settings before they run the acquisition again.

If the data is out of range for an extended period of time, this is an indication that something is probably wrong. The longer the data is out of range for, the more likely it's due to a problem with the user's actual experiment or the device settings set by the user. That doesn't mean the user must abort it, but it does mean that they need to be clearly informed that there is a probable issue.

To complicate things further, there are 16 independent measurements being plotted at the same time on the same plot. Some may be out of range while others are not.

Question

With respect to the situations described above, what's a good way to indicate to the user that the data for a given channel is out of range in the real-time plot?

My Ideas and Attempts

One prototype I've written shows the text "Out of range" next to a channel in the legend on the plot while data being received is out-of-range. This only works if the user is watching the legend at the exact moment that out-of-range data is received. There's no lasting indication to the user that data was out of range if they weren't there to see it, until they view the saved data file at the end. As a modification on this, I may have the "Out of range" text slowly fade out if there is no more out of range data, to make it easier to see transients.

Another idea is to show a running tally of the number (or proportion) of out of range data points for each channel.

Improve this question
2
  • I'm a little confused. Does "out of range" mean "we couldn't get this data during this time slice", "the value we got is off the meter so we can't use/visualize it", or something else? Commented Nov 2, 2011 at 17:51
  • 1
    The second, as he writes: "i.e. beyond the measurable limit". Commented Nov 2, 2011 at 17:57

1 Answer 1

Reset to default
6

What about simply graying out the line where the data doesn't exist / is unreliable?

plot

That way you can pick back up when you're in-range again, but there's an indication that there was a problem.

EDIT: Assuming your "out of ranges" are always trackable right up to the top/bottom of the range, the end-result would look more like this:

plot2

EDIT 2:For small out-of-range/unreliable data sectors (e.g. single data points) or out-of-range against the X axis, you could make it even more obvious with an icon:

plot3

Obviously, the ability to do this depends on your display technology, as well.

Improve this answer
5
  • 1
    I was about to propose a similar idea, but this risks getting confusing with 16 curves on the same graph. Still, I can't come up with anything significantly different at the moment! ;) Commented Nov 2, 2011 at 18:07
  • Excellent suggestion for how to show when data is missing/unreliable. However, I'm not sure how well it will work for out-of-range data. It would look something like this since the limit is a y max/min and data will likely be plotted right up to when it goes out of range: i.imgur.com/OD0l3.png Commented Nov 2, 2011 at 18:23
  • I see you updated your answer with the second plot while I was entering my comment. I have the same concern as msanford about using that method, but I think I will prototype it to see how well it works. Thanks. Commented Nov 2, 2011 at 18:26
  • @Series8217 Yes, I'm cheating my edge on the second plot. I added another idea to call attention to the issues you and msanford raised. Commented Nov 2, 2011 at 18:31
  • 1
    Brilliant! I really like that idea. It could easily be extended to multiple curves as well, by having the icons expand into infoboxes to indicate what exactly is going on (which channels are out of range) in that region or at that point, without clutter. Commented Nov 2, 2011 at 19:30

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.