Continue line function not enabled for stacked area chart

[brendan]

It is apparently impossible to continue the line for data gaps in a stacked area chart, but fine for normal - presumably an omission. Anyone know if this will be fixed in the next release, (or if a patch available - couldnt find one)?

[jrkrideau]

It is apparently impossible to continue the line for data gaps in a stacked area chart, but fine for normal - presumably an omission.

I am unable to get a continuous line in either a normal or stacked line chart with missing data. In fact the normal chart seems to be dropping some data given my little test data set. Are you sure the other plot that you refer to is not an xy-graph?

The other question is; why would a line chart give a continuous line when plotting categorical data? Remember a line chart is simply a version of a bar chart. There is no intrinsic order to the data on the x-axis.

In a line chart the x-axis is categorical and yours could read x =[580, 60, -20, 19] just as easily as it could read x=[A, B, C, D] or x=[-20, 19 60, 580]. It's all the same thing to the computer and to the reader. A line chart gives the illusion of a continuity or of a function that is extremely misleading.

[brendan]

no, definitly line charts - stacked/unstacked:

[brendan]

I am having difficulty with your point about continuous v categorical data in line graphs. As i understand it, a classic line graph might be to plot something against time. That is, again as i understand it, continuous interval data and not categorical. Classic line graph shapes like exponential, sigmoid or standard deviation are also presumably based on continuous rather than categorical, otherwise the shape would be irrelevant. The same with my graph which charts temperature as the interval. I wish to stack the two data sets as the two individual data sets (heat transfer coefficients) sum to form the overall coefficient. But i cannot get a continuous line from either a stacked line or stacked area graph (but can on unstacked). I cant see the logic of it allowing continuous line for unstacked but not stacked so presume it must be a bug rather than an attribute of the type of graph.

[jrkrideau]

I am having difficulty with your point about continuous v categorical data in line graphs. As i understand it, a classic line graph might be to plot something against time. That is, again as i understand it, continuous interval data and not categorical. Classic line graph shapes like exponential, sigmoid or standard deviation are also presumably based on continuous rather than categorical, otherwise the shape would be irrelevant.

Okay, I think I see the problem: Terminology! You are right about a classic line graph but that is not what OOo is doing in its version of a line chart. The term line chart is a bit of a bastard term. What you are calling a classic line graph Calc calls an xy-graph Spreadsheet graphics are basiclly pretty crappy for several reasons and bad nomenclature is one of them.

Let's take the idea of graphing something say children's height against time where the child's height is measured in months but not every month—say the parents' forget some months so you have something making up some data heights measured at months = (12, 13, 17, 18 20, 21, 22, 30) and heights = (24, 25,30, 31, 33, 33.5, 34, 41). All data just made up on the spot. In the attached file I have plotted the data with months as continuous (xy-plot) in Figure 1a and categorical (line chart) in Figure 1b.

What OOo is doing when you use a line chart is treating the spaces between each value as being equal. So in the chart, the difference between 12 & 13 months equals the difference between 22 and 30 months.

Translated into your plot, in a line chart Calc would give the the difference between -280C and -20 and -20 and 19 the same spacing.

Currently in the graph you have posted it looks like you are actually using an xy-graph since the spacing does not appear to be the same. This might explain why there is no stacked line graph option. Unless you are somehow rigging the x-axis layout? I clearly was not paying enough attention to that plot earlier.

The same with my graph which charts temperature as the interval. I wish to stack the two data sets as the two individual data sets (heat transfer coefficients) sum to form the overall coefficient. But i cannot get a continuous line from either a stacked line or stacked area graph (but can on unstacked). I cant see the logic of it allowing continuous line for unstacked but not stacked so presume it must be a bug rather than an attribute of the type of graph.

I see what you mean. I don't know either and I am getting a different result -- I cannot get an acceptable line when there is missing data. I may have a problem with Calc.

What I don't understand is what the stacked graph is supposed to show. Are you trying to show HTC radiated + HTC conducted?

If so, it would probably be as good as anything to simply add another column to your data, add them together and graph a third line on the graph or, probably better, do a second graph. If I understand what you are doing, there are some problems in using something like a stacked or area type charts due to how how the human eye reads the difference of lines on some slopes and two graphs are likely to be better in many cases.

I may be wrong but I think that you can do everything you want fairly easily using xy-graphs which are your classical line graphs.

I hope this is of some help, or at least that I have managed to confuse you a bit more

[brendan]

You have confused me a bit. The only chart i can see calling itself an xy chart is the scatter chart, which is not the same as my 'classic' line graph and i dont think is appropriate. It also changes the x axis so it isnt the temperatures i provided, and drops one of the data sets, so it shows just a single vertical line.

I have manipulated the spacing so that there are lots of empty points in the data set, then use the 'continue line' function to ignore these abscences.
I would think that a scatter graph would be better in your example graph if there were several children being measured, rather for a line where there is missed a couple of months so you have gappy data and so uneven intervals (which is equivalent to my situation).

I am thinking of sulkily ditching OO for graphs. It is just too full of niggling little problems of basic functionality that i am wasting too much time on. I will try it in excel or try and grapple with a proper graphing package. Seems wot you get is wot you pays for!
Thanks for your support.

[Cambirder]

The only chart i can see calling itself an xy chart is the scatter chart, which is not the same as my 'classic' line graph and i dont think is appropriate. It also changes the x axis so it isn't the temperatures i provided, and drops one of the data sets, so it shows just a single vertical line.

The reason you are loosing one of your data series is that that you have the "Use first column as label" option ticked, un-tick that and you will get both columns back.

You can't do a stacked X-y chart so you will need to create a column of data set 1 + dataset 2 to get the same effect, which adds a bit of work but on the plus side you can get rid of all those blank rows as you don't need them in x-y chart.

I am thinking of sulkily ditching OO for graphs. It is just too full of niggling little problems of basic functionality that i am wasting too much time on. I will try it in excel or try and grapple with a proper graphing package.

I would not bother with Excel, it defaults to returning blank cells as zero, and in Excel 2007 I can't see a way to change that so its worse than Calc.

[jrkrideau]

You have confused me a bit. The only chart i can see calling itself an xy chart is the scatter chart, which is not the same as my 'classic' line graph and i dont think is appropriate.

Actually it is. Sorry, as I said before the terminology in spreadsheets is a bastard at times.

It also changes the x axis so it isnt the temperatures i provided, and drops one of the data sets, so it shows just a single vertical line.

I think Cambirder has answered this one.

I have manipulated the spacing so that there are lots of empty points in the data set, then use the 'continue line' function to ignore these abscences.

That is really going to mess up the data presentation.

I would think that a scatter graph would be better in your example graph if there were several children being measured, rather for a line where there is missed a couple of months so you have gappy data and so uneven intervals (which is equivalent to my situation).

I really don't understand this. I was under the impression that you were trying to graph one two data series (HTC radiation and HTC conducted against a Kelvin temperature range. My children's height/weight graph is analogous to graphing, say HTC radiation against temperature.


We could expand my children's weight graph by saying that we were going to split my children into males and females and plotting height/weight by sex—giving us the rough equivalent of your two HTC variables.

When you say that you have gappy data, I don't quite understand this. Do you mean that you have missing data or just that you only have readings at irregular temperature intervals?

I am thinking of sulkily ditching OO for graphs. It is just too full of niggling little problems of basic functionality that i am wasting too much time on.

Well Calc does not have a lot of graphics functionality but I really think that it's a terminology problem-- you are using the more or less standard terms for graphs and spreadsheets are not.

I will try it in excel or try and grapple with a proper graphing package. Seems wot you get is wot you pays for!
Thanks for your support.

As Cambirder says, Excel is unlikely to be any better. In the long run, if you are going to be doing this type of thing a lot then a decent graphing package is worth while. Whether it is for a one off project is another matter. The only ones that I know have a fairly nasty learning curve at the start.

R
gnuplot

[brendan]

The graph in xy does indeed look much better without ticking the 'use 1st column as label'.

I have stacked in the way suggested by Cambirder.

I looked at excel and see that you are correct. I need to get into my thick skull that what i have hitherto thought of as a line graph is actually a scatter graph. How have i got past 30 with this misapprehension?! When i say gappy data you'll see what i mean on the attached spreadsheet - i wont present the actual tables like this, i only do this to get the graph shape i want.

Incidentally i think there are some errors in the actual data making the graphs look a bit wobbly. I will need to go over them and check. I have deleted some of the data to make the file small enough, so scroll through all the ghost tables.

[jrkrideau]

The graph in xy does indeed look much better without ticking the 'use 1st column as label'. I have stacked in the way suggested by Cambirder.

Excellent, it looks like you now have what you want.

I looked at excel and see that you are correct. I need to get into my thick skull that what i have hitherto thought of as a line graph is actually a scatter graph. How have i got past 30 with this misapprehension?!

Not your fault, spreadsheets have some strange naming conventions and some truly weird and useless graphing types.

When i say gappy data you'll see what i mean on the attached spreadsheet - i wont present the actual tables like this, i only do this to get the graph shape i want.

Now that you are using a scatterplot you don't need all that padding. You get exactly the same plot if you use the existing data. For example, I got the same plot you did using the table below. Names changed to save me some typing but the data comes from $A1$191:$CF$193 in the Sand.box.ods file.

Code: Select all

T1	Hc	Hr
-260	3.53	2.09
-20	3.53	4.18
0	3.53	4.63
	3.53	4.88
19	3.53	5.1
	3.53	5.4
40	3.53	5.69
60	3.53	6.35
300	3.53	18.4
580	3.53	47.8

Incidentally i think there are some errors in the actual data making the graphs look a bit wobbly. I will need to go over them and check.

That's always a possibility but it may also just be a bit of random noise in the measurements.

[brendan]

oh yeah!

[User030413]

Hi,

I am very new (for less then one hour) and probably one-time member and forgive me if I placed this query in a wrong place.
I also fail see what the OO's <http://forum.openoffice.org/en/forum/faq.php#f3r0> page tells me to look for:
"How do I post a topic in a forum?
To post a new topic in a forum, click the relevant button on either the forum or topic screens. You may need to register before you can post a message. A list of your permissions in each forum is available at the bottom of the forum and topic screens. Example: You can post new topics, You can vote in polls, etc."
Just no such button(s), nor permissions notice...

I have been using OpenOffice-Calc for years but in a very mathematically-primitive way.

Now I need to create a simple line-graph but with one apparently complicated (for me element:
Values represented on the Y-axis have been measured in non-equal times, expressed in decimals, such as "20.42" (meaning "8:25 PM"), "21.92", "22.58".

Anyway, on X-axis it could be any "non-linear" series, such as 2, 6, 13, 43, etc.

I would like these "time-distances" or generally "intervals" to be represented proportionally on my X-axis.

Is it possible (I haven't found any solution in OO tutorial[s])?

Thanks in advance for any advice and/or indication where to look for a solution.

[jrkrideau]

Hi,

I am very new (for less then one hour) and probably one-time member and forgive me if I placed this query in a wrong place.
I also fail see what the OO's <http://forum.openoffice.org/en/forum/faq.php#f3r0> page tells me to look for:
"How do I post a topic in a forum?
To post a new topic in a forum, click the relevant button on either the forum or topic screens. You may need to register before you can post a message. A list of your permissions in each forum is available at the bottom of the forum and topic screens. Example: You can post new topics, You can vote in polls, etc."
Just no such button(s), nor permissions notice...

I have been using OpenOffice-Calc for years but in a very mathematically-primitive way.

Now I need to create a simple line-graph but with one apparently complicated (for me element:
Values represented on the Y-axis have been measured in non-equal times, expressed in decimals, such as "20.42" (meaning "8:25 PM"), "21.92", "22.58".

Anyway, on X-axis it could be any "non-linear" series, such as 2, 6, 13, 43, etc.

I would like these "time-distances" or generally "intervals" to be represented proportionally on my X-axis.

Is it possible (I haven't found any solution in OO tutorial[s])?

Thanks in advance for any advice and/or indication where to look for a solution.

You probably should have started a new thread but this is also a good place since the discusion above sets the correct tone

As far as I can see from your description of the data all you need is the x-axis data in one column and the y-axis data in another -- use x and y in the top row to give the data names.

For your first quick and dirty graph
1 Select the two rows of data
2 Click on the Charting Icon
3 Select XY scatter graph
4 Click Finish
5 Done.

If this works then you can get rid of that graph and then more carefully work through the rest of the graphing options on your next try to make it look like a real graph with titles, etc.