1

I would like to plot means and confidence interval with Gnuplot. My data has several samples with the same x value

21  27  4   12  20  100 50000   false   0.1 "DFSA"  2   205068  31.986
24  27  4   12  20  100 50000   false   0.1 "DFSA"  2   207798  32.49448
22  27  4   12  20  100 50000   false   0.1 "DFSA"  2   207499  32.60746
23  27  4   12  20  100 50000   false   0.1 "DFSA"  2   214065  33.17844
25  27  4   12  20  100 50000   false   0.1 "DFSA"  2   198386  31.4488
26  27  4   12  20  100 50000   false   0.1 "DFSA"  2   208307  32.77412
28  27  4   12  20  100 50000   false   0.1 "DFSA"  2   208669  32.77474
27  27  4   12  20  100 50000   false   0.1 "DFSA"  2   214324  33.16506
30  27  4   12  20  100 50000   false   0.1 "DFSA"  2   209924  32.77186
29  27  4   12  20  100 50000   false   0.1 "DFSA"  2   212959  33.3218
32  27  4   12  20  100 50000   false   0.2 "DFSA"  2   486241  64.3479
31  27  4   12  20  100 50000   false   0.2 "DFSA"  2   487487  64.65076
33  27  4   12  20  100 50000   false   0.2 "DFSA"  2   494703  65.06718
34  27  4   12  20  100 50000   false   0.2 "DFSA"  2   488164  64.77828
36  27  4   12  20  100 50000   false   0.2 "DFSA"  2   476513  63.3158
35  27  4   12  20  100 50000   false   0.2 "DFSA"  2   491005  65.02426
38  27  4   12  20  100 50000   false   0.2 "DFSA"  2   487454  64.44962
37  27  4   12  20  100 50000   false   0.2 "DFSA"  2   490494  65.06572
39  27  4   12  20  100 50000   false   0.2 "DFSA"  2   472081  63.31234
40  27  4   12  20  100 50000   false   0.2 "DFSA"  2   498294  65.02114

It has ten samples with column 9 = 0.1 and the same with value 0.2. I want to plot the mean of column 13 (last column) with the confidence interval and use Gnuplot yerrorlines, which need data in four columns x mean min max. The formula of min and max values uses the STATS_mean, STATS_ssd, and STATS_records of column 13 with equal value in column 9.

I have tried to write the values x mean min max in a file with this

system("touch lbt.dat")
stats "data.txt" u ($9==0.1?$13:1/0) name "lbt01"
system("(0.1 lbt01_mean (lbt01_mean - 2.262*lbt01_ssd/sqrt(lbt_records)) (lbt01_mean + 2.262*lbt01_ssd/sqrt(lbt_records))) >> lbt.dat")

but I get error in the last system command to save the values in the lbt.dat file. I have also tried

system("printf '%f\t\%f\t\%f\t\%f' 0.1 lbt01_mean (lbt01_mean - 2.262*lbt01_ssd/sqrt(lbt_records)) (lbt01_mean + 2.262*lbt01_ssd/sqrt(lbt_records))) >> lbt.dat ")

(2.262 is the value of the t-student pdf for 95% confidence interval and 9 degrees of freedom).

It looks like the results of stasts, e.g. lbt01_mean cannot be written in a file using system.

Besides, since the data file has samples for 0.1 to 1.7 in steps of 0.1 (column 9), I would like to ask if is there a way to short the Gnuplot script.

Regards

1 Answer 1

3

An easy way to print to a file from gnuplot is with the print command, after setting the file with set print. Try

set print "lbt.dat"
print 0.1, lbt01_mean, \
  (lbt01_mean - 2.262*lbt01_ssd/sqrt(lbt01_records)),\
  (lbt01_mean + 2.262*lbt01_ssd/sqrt(lbt01_records))

In order to loop through all the values 0.1 to 1.7, you can use a for loop going from 1 to 17, then dividing by 10.

set print "lbt.dat"
do for [i=1:17] {
  v = i/10.
  stats "data.txt" u ($9==v?$13:1/0) name "A" nooutput
  print v, A_mean,   (A_mean - 2.262*A_ssd/sqrt(A_records)),\
    (A_mean + 2.262*A_ssd/sqrt(A_records))
}

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .