I have data from some sensors which are installed at different locations. For analysis, I would like to see in which direction and at which speed the analysed material propagates in the room.

The following diagram shows some sensor data as an example of four different sensors:

Sensor Data

I shifted the location of the data for the sensors to keep all lines as visible as possible. The time labels are not useful in this diagram, I just quickly created this example - it is actually a time axis with a one-hour resolution.

As you can see, the readings are very similar, with very small differences.

Now I struggle to find a good visualization for this data, where this time differences are made visible.

The shown data is available here in CSV format.

My questions are:

  • How can I visualize the small time differences between the readings?
  • How can I deal with the large reading range to keep everything visible?

1 Answer 1


Thanks for including the data!

Those are tough questions without perfect answers, but here are some common techniques.

  • Smooth the data with the goal of making easier to see trends over individual values, which presumably have some noise.
  • Show only a few days at a time so there is better time resolution.
  • Use a scale transformation, such as log or square root, that exaggerates small values.
  • Plot derived values that highlight what you're interested in. For instance, plot the difference of each series compare to the mean of all series.

Below is an example of smoothing each value with its closest two neighbors and using a square root transformation on the y axis. It's paneled by week to increase the time resolution.

enter image description here

  • $\begingroup$ This already looks much better than my try. :-) Which software did you use to produce your example? $\endgroup$
    – Flovdis
    Commented Mar 17, 2019 at 20:36
  • 1
    $\begingroup$ Great -- I'm using JMP, but the techniques should be applicable to any tool. $\endgroup$
    – xan
    Commented Mar 18, 2019 at 0:58

Your Answer

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

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