Disjoint chart ranges

English
,
#1

I have a chart that displays a set of measurements taken over a few months. There are long periods when no measurements were taken. I want to omit these in the chart. I’ve attached a simple example of what I’m trying to do
ChartRanges.ods (18.2 KB)

In this sample spreadsheet there are a set of measurements from 1-4 January, then from 28 January to 5 February. I want the x axis to display 1,2,3,4 January then 28,29… I don’t want 5-27 January when there were no measurements.

In this chart, I set the data range to $Sheet1.$A$1:$B$4,$Sheet1.$A$6:$B$10, hoping this might omit the missing dates, but it doesn’t.

Is there any way to display disjoint intervals along the x axis?

#2

Change the chart type from x-y scatter to line chart and the type of x-axis from date to text.

#4

I use an x-y scatter chart because line displays dates equidistantly. The example I gave was very simple. In the real data, there may be several readings in a day, then some the next day. I don’t want all these measurements displayed equidistantly.

#5

@Lupp Thanks for the updated chart. I didn’t explain enough about the data. Often there will be a set of measurements throughout a day, then one or two the next day. If I display all these using dates as equidistant text, they are not easy to interpret, particularly as what I want to show is a trend over time.

I agree. That’s why I don’t want a long slope between two sets of measurements a long time apart.

I’ll think about using a column chart, but there are many reasons why a line (or xy scatter) would be better. There are actually a few data series to display. I want to display trend lines on each. I was considering adding error bars. Trying to do all this on a column chart would be messy.

#6

It may be a case where “not easy” is much better than misleading.

This is exactly what I would strictly dissuade from if those series not are rightfully displayed as x-y-diagrams. All the math behind trendlines is approximation assuming a continuous x-axis divided using one fix unit. In addition trendlines only can make sense if you have an idea about an applicable model (or are trying to find a preferrable one experimentally). You obviously haven’t. Be sceptical with respect to all the silly “visualitzations” you get yourself presented and don’t join the caravan of those producing them,
See
RE2_chartRanges.ods (37.4 KB)
See also this xkcd-sketch.

#7

Thanks for the new chart, but it’s not what I meant. Each of my data series covers the whole time interval. When I applied a trend line to each there was a clear trend which was less obvious in the raw data. It helped me see what was happening.
I like the xkcd-sketch, but that’s not what I’m trying to emulate. I’m just trying to understand some data myself. I’m not in marketing!

#3

In principle you can do so by converting the dates which are provisionally used as category names to actual texts. The chart will then no longer try to scale the x-axis reasonably.
You will be fully responsible then. This includes, imo, that you don’t create a “line” graph where the subdivision of the x-axis can’t be equidistant (or in some cases logarithmic).
Students need to be trained to interpret the slope of a line as something showing real properties of the participating measurables. Please don’t spoil this. Use a column chart in such a case, and leave a gap.
RE_chartRanges.ods (33.6 KB)

#8

For what you describe, making an XY plot but leave out some ranges, you would need a feature to break the axis. LibreOffice does not have such feature, unfortunately.

#9

There is a kind of profession concerned with visualization. Sometimes they do a good job, sometimes they fail.
If they fail it’s (imo) due to putting impressive images before clear information.
Anyway we shouldn’t expect an all-purpose software to do better without our effort.
If we accept the need of some handiwork, and probably of developing some Calc formulas or even a decent amount of user code, we can do a lot to get visualizations presenting undistorted, relevant, and correct information. In any case, we must trust that the audience will also make the necessary effort to extract the presented information correctly.
A very raw example of what I tried in similar cases is attached.
disask89171TrendsComparingCharts.ods (30.9 KB)

Impressum (Legal Info) | Datenschutzerklärung (Privacy Policy)   Content on this site is licensed under a Creative Commons Attribution Share Alike 3.0 license.