Gnuplot is really good for doing quality plots (Good enough for scientific papers). The output terminals include screen, LaTEX, png, SVG, EPS and PDF.
The dreaded Wackypedia has a reasonable amount of info (I would especially recommend the IBM tutorial - it really got me started with GNUplot) http://en.wikipedia.org/wiki/Gnuplot If you need a non-Windows GUI there is a Java GUI for GNUplot (I haven't tried it so YMMV) see http://sourceforge.net/projects/jgp/
Hope that this helps, Mark
On 16/06/13 17:54, mark fernandes (IFR) wrote:
Gnuplot is really good for doing quality plots (Good enough for scientific papers). The output terminals include screen, LaTEX, png, SVG, EPS and PDF.
The dreaded Wackypedia has a reasonable amount of info (I would especially recommend the IBM tutorial - it really got me started with GNUplot) http://en.wikipedia.org/wiki/Gnuplot If you need a non-Windows GUI there is a Java GUI for GNUplot (I haven't tried it so YMMV) see http://sourceforge.net/projects/jgp/
Hope that this helps,
Thanks - I'll compare it with LO Calc.
On 16-Jun-2013 17:30:52 Anthony Anson wrote:
On 16/06/13 17:54, mark fernandes (IFR) wrote:
Gnuplot is really good for doing quality plots (Good enough for scientific papers). The output terminals include screen, LaTEX, png, SVG, EPS and PDF.
The dreaded Wackypedia has a reasonable amount of info (I would especially recommend the IBM tutorial - it really got me started with GNUplot) http://en.wikipedia.org/wiki/Gnuplot If you need a non-Windows GUI there is a Java GUI for GNUplot (I haven't tried it so YMMV) see http://sourceforge.net/projects/jgp/
Hope that this helps,
Thanks - I'll compare it with LO Calc.
-- Tony http://www.girolle.co.uk/
If I had a typical dataset to work with, I would like to try implementing this project using groff (in particular the 'pic' preprocessor).
I gather that it would consist of 24 hours of blood pressure data, over a month, with a reading taken potentially every half hour on each day (so maximum of 31*24*2 = 1488 readings in a month), with the strong possibility that some (or many?) of them may be missing.
And is it just one BP reading at a time (e.g. diastolic) or would it be both diastolic and systolic for each reading?
One of the crucial aspects of this kind of thing is how one wants it to look, and how easily readable one wants it to be! This can be more readily customised using groff+pic than with many other data/plotting packages (and, in particular, setting up customised graphics using the generally excellent R[*] statistical package can involve many hours wandering disorientedly through a maze of help pages ... ).
Best wishes, Ted.
------------------------------------------------- E-Mail: (Ted Harding) Ted.Harding@wlandres.net Date: 16-Jun-2013 Time: 23:28:23 This message was sent by XFMail -------------------------------------------------
On 16/06/13 23:28, (Ted Harding) wrote:
On 16-Jun-2013 17:30:52 Anthony Anson wrote:
Thanks - I'll compare it with LO Calc.
If I had a typical dataset to work with, I would like to try implementing this project using groff (in particular the 'pic' preprocessor).
Thanks for the interest: I have readings of systolic and diastolic pressures, as well as pulse rate, all taken between the end of February and now. The readings tend to be taken hourly except when I'm sleeping, out, or the readings are worrying.
I gather that it would consist of 24 hours of blood pressure data, over a month, with a reading taken potentially every half hour on each day (so maximum of 31*24*2 = 1488 readings in a month), with the strong possibility that some (or many?) of them may be missing.
Well, I plan monthly graphs of sys/dia + pulse rates, all on the same time-lines, day by day.
It has occurred to me, late as usual, that I should have included current activity and perhaps what I have eaten and drunk, and when. (Including medications - these last ones seem to be making things worse...)
I say 'late', as I have appointments to have a 24 hr ECG fitted and unfitted.
And is it just one BP reading at a time (e.g. diastolic) or would it be both diastolic and systolic for each reading?
Both, and pulse-rate.
ATM they are recorded as numbers. These are somewhat higher than I would like, as my BP was typically low and my resting pulse around forty:
12.45 - 149/83 + 152 bpm
13.45 - 128/85 + 131 bpm
14.45 - 127/80 + 148 bpm
15.45 - 185/105 + 144 bpm
One of the crucial aspects of this kind of thing is how one wants it to look, and how easily readable one wants it to be! This can be more readily customised using groff+pic than with many other data/plotting packages (and, in particular, setting up customised graphics using the generally excellent R[*] statistical package can involve many hours wandering disorientedly through a maze of help pages ... ).
Yes - ideally it would just be three wiggly lines with time data and other comments - nothing fancy, and the idea would be to make a readable graph which is neither understated nor exaggerated.
On 17/06/13 00:40, Anthony Anson wrote:
12.45 - 149/83 + 152 bpm
13.45 - 128/85 + 131 bpm
14.45 - 127/80 + 148 bpm
15.45 - 185/105 + 144 bpm
I've been trying to work out how to do this with Gnuplot, as a mental exercise! I re-entered that data above into a file I called test3.dat
#Date Time High Low BPM 17/6/2013 12:45 149 83 152 17/6/2013 13:45 128 85 131 17/6/2013 14:45 127 80 148 17/6/2013 15:45 185 105 144
NB that's space separated.
I obviously installed gnuplot, but then I installed gnuplot-qt. This caused gnuplot-nox to be uninstalled. I don't know if I needed to do this, but before I did it, I didn't know how to get output to appear on screen.
I moved to the directory that file test3.dat was in, & started GnuPlot, then typed
set xdata time set timefmt "%d/%m/%Y %H:%M" plot 'test3.dat' using 1:3 with linespoints title 'High', 'test3.dat' using 1:4 with linespoints title 'Low', 'test3.dat' using 1:5 with linespoints title 'BPM'
What this does is 1) sets the x axis to be showing dates & times
Sets the format of the time column(s) in the data file to be the format you are using. In this case days (0-31), months (0-12), years (4 digits), a space, then Hours (24 clock format - 00-23), Minutes (00-59).
Gnuplot considers date/time to take up 2 columns with the format above.
The plot line plots data in file "test3.dat" using columns 1 (date&time), and 3 - high bp, and labels it High. It then plots the data in file "test3.dat" (the same file) using columns 1 (date & time) and column 4 and labels it "low".
It then plots data in file using columns 1 and column 5 which it labels BP.
I'm sure it'll be possible to convince gnuplot to use different format data. Use help timefmt to look at possible values to put in the set timefmt command. If you have commas between columns, you can use set datafile sep ','
I think the thing is you'll need consistency - e.g. always have space or comma separated columns.
Actually, I just changed the data so that it was in the format #Date Time High Low BPM 17/6/2013 12:45 149 83 - 152 BPM ...
and changed the final plot to 'test3.dat' using 1:6 and it worked.
Obviously you'll have to ensure that if you're graphing over separate days, that you have the date in the data and not just the time.
Hope that helps!
Steve
On 17/06/13 15:32, steve-ALUG@hst.me.uk wrote:
I obviously installed gnuplot, but then I installed gnuplot-qt. This caused gnuplot-nox to be uninstalled. I don't know if I needed to do this, but before I did it, I didn't know how to get output to appear on screen.
Just checked and installing gnuplot-x11 will also get output to appear on my screen, and it uninstalls gnuplot-qt. gnuplot-nox outputs to printers and pdf files apparently.
Steve
On 17/06/13 15:32, steve-ALUG@hst.me.uk wrote:
On 17/06/13 00:40, Anthony Anson wrote:
12.45 - 149/83 + 152 bpm
13.45 - 128/85 + 131 bpm
14.45 - 127/80 + 148 bpm
15.45 - 185/105 + 144 bpm
I've been trying to work out how to do this with Gnuplot, as a mental exercise! I re-entered that data above into a file I called test3.dat
#Date Time High Low BPM 17/6/2013 12:45 149 83 152 17/6/2013 13:45 128 85 131 17/6/2013 14:45 127 80 148 17/6/2013 15:45 185 105 144
NB that's space separated.
I obviously installed gnuplot, but then I installed gnuplot-qt. This caused gnuplot-nox to be uninstalled. I don't know if I needed to do this, but before I did it, I didn't know how to get output to appear on screen.
I moved to the directory that file test3.dat was in, & started GnuPlot, then typed
set xdata time set timefmt "%d/%m/%Y %H:%M" plot 'test3.dat' using 1:3 with linespoints title 'High', 'test3.dat' using 1:4 with linespoints title 'Low', 'test3.dat' using 1:5 with linespoints title 'BPM'
What this does is 1) sets the x axis to be showing dates & times
Sets the format of the time column(s) in the data file to be the format you are using. In this case days (0-31), months (0-12), years (4 digits), a space, then Hours (24 clock format - 00-23), Minutes (00-59).
Gnuplot considers date/time to take up 2 columns with the format above.
The plot line plots data in file "test3.dat" using columns 1 (date&time), and 3 - high bp, and labels it High. It then plots the data in file "test3.dat" (the same file) using columns 1 (date & time) and column 4 and labels it "low".
It then plots data in file using columns 1 and column 5 which it labels BP.
I'm sure it'll be possible to convince gnuplot to use different format data. Use help timefmt to look at possible values to put in the set timefmt command. If you have commas between columns, you can use set datafile sep ','
I think the thing is you'll need consistency - e.g. always have space or comma separated columns.
Actually, I just changed the data so that it was in the format #Date Time High Low BPM 17/6/2013 12:45 149 83 - 152 BPM ...
and changed the final plot to 'test3.dat' using 1:6 and it worked.
Obviously you'll have to ensure that if you're graphing over separate days, that you have the date in the data and not just the time.
Hope that helps!
If I can get my head round it, probably, thanks...
The complete documents (monhly) are written in html (4 Transitional), and have day, month, year included, as well as a few comments.
-
Anthony Anson -
mark fernandes (IFR) -
steve-ALUG＠hst.me.uk -
Ted.Harding＠wlandres.net