Curve fitting data in Linux

[gilksjm]

I am currently lookuing for some function smoothing and plotting software.

Have checked the usual suspects but none seem to do quite what I need.

Plotting can be handled easily by gnuplot, grace or xmgr,
however I need to smooth the data before plotting.
By preference I would like to get a line of best fit for the data and then plot both the data and the line.

Really the problem is the curve fitting.

Does anyone have any suggestions, an easy gui with a batch mode as well would be nice but shell scripts or perl to match the curve will be fine.

Thanks in advance.

Love
Mike.

[hidders]

I\'ve never used it myself, but did you look at \'fityk\'? It\'s available as Mandrake rpm: fityk-0.3.2-1mdk



-- Jan H.

[mbody]

Most people would put the data in a spread sheet and let it plot and smooth (select the data set on the plot with your mouse and ask for regression). But I\'ve noticed that Open Office does not actually give the coefficients which are generated



Octave (which is available in the download area) can do this and MUCH more with little trouble.



For example, cut and paste the last bit of this message to a text file and name it \"fit.m\". Then start Octave from the terminal by typing \"octave\". At the prompt type \"fit\" and it will run provided you are in the same directory as the fit.m file.



If you run several scripts (or the same one over again) it would be good to clear all of the old variables by simply typing \"clear\" (of course you could put this at the begining of your script).



% data fitting



% your data

x = [1 2 3 3.5 4 5 6];

y = [1.1 2.2 3.1 3.4 4.5 4.6 6.3];



% the fit is

p = polyfit (x, y, 1);



% 1 is the order of the polynomial least squares fit (a line)

% p are the 2 coefficients required for the a line



% so you can now generate points for the line as



my_line_x = [min(x) max(x)];

my_line_y = polyval(p,my_line_x);



% and plot them as... use \"help plot\" to learn more about formatting

plot (x, y, \"@12\", my_line_x, my_line_y)





have fun fitting

[manouchk]

I though you could smooth curve with

plot \'data\' using 1:2:0.0000001 ..... smooth acsplines ?





(use a small values for 3rd number of using)
_________________
peace and love

[gilksjm]

Okay, thanks guys. I will give octave a look.



May be down to finding curve fitting algorithms for c or c++ and write a small program.



Hopefully be able to use someone elses expertise first though.



Thanks again.



Love

Mike.

[Butchu]

Grace and gnuplot can do curve fitting as well. I do not know about smoothing, but if you just want to fit a simple formula to some dataset there is no need to get into c or c++...



Sebastien

[shengchieh]

I think GNUplot probably is best altough I

only used for xy plots. (but I know spline, etc.

are there). Take a look at



http://www.gnuplot.info/

www.cs.uni.edu/Help/gnuplot

www.ucc.ie/gnuplot/gnuplot.html

http://www.duke.edu/~hpgavin/gnuplot.html



Sheng-Chieh