Pgfplots: Error of Linear Regression

Question

Hope you don't mind me asking this, I'm only a starter at both gnuplot and LaTeX, but I've already searched the pgfplots manual and couldn't find an answer:

Is it possible to have pgfplots calculate and output the error of its linear regression for both parameters a and b (preferrably in the legend)? I know Origin does this automatically, and it shouldn't be so hard in pgfplots, but I haven't found anything yet.

Welcome to TeX.SX. Have a look at Linear regression - trend line with pgfplots — Johannes_B, Dec 05 '13 at 18:22
First, thank you for answering this fast. Yet I've seen this post and as far as I can see, it only explains how to do a linear regression, and doesn't solve my problem. — Denna, Dec 05 '13 at 18:29
Hi! It would be nice if you edited your question and added a minimal working example (MWE), with a plot that contains the regression and only misses the error values. That way, people can much easier help you ;) — yo', Dec 05 '13 at 18:37

score 7 · Accepted Answer · edited Apr 13 '17 at 12:35

You can adapt the approach from Adding values to pgfplot legend:

\documentclass[11pt]{article}
\usepackage{pgfplots}
\usepackage{pgfplotstable}
\usepackage{filecontents}

\begin{filecontents}{data.dat}
1 3
2 5
3 4
4 8
5 9
6 8
7 10
8 12
9 10
10 11
\end{filecontents}



\begin{document}

\begin{figure}[h!t]
\centering
\begin{tikzpicture}

\begin{axis}[
    legend pos=north west,
    axis x line*=bottom,
    axis y line*=left,
    tick label style={font=\small},
    grid=both,
    tick align=outside, 
    tickpos=left,
    xmin=0, ymin=0
    ]

\addplot[only marks, mark size=1.8, black] file {data.dat};
    % Now call gnuplot to fit this data
    % The key is the raw gnuplot option
    % which allows to write a gnuplot script file
    \addlegendentry[]{Experiment 1}

\addplot+[raw gnuplot, red, mark=none, smooth] gnuplot {
    f(x)=a*x+b;
    % let gnuplot fit, using column 1 and 2 of the data file
    % using the following initial guesses
    a=1;
    b=1;
    set fit errorvariables;
     fit f(x) 'data.dat' using 1:2 via a,b;
         % Next, plot the function and specify plot range
         % The range should be approx. the same as the test.dat x range
     plot [x=0:10] f(x);
    set print "parameters.dat"; % Open a file to save the parameters into
    print a, a_err; % Write the parameters to file
    print b, b_err;
       };       
\addlegendentry[]{\pgfplotstableread{parameters.dat}\parameters % Open the file Gnuplot wrote
    \pgfplotstablegetelem{0}{0}\of\parameters \pgfmathsetmacro\paramA{\pgfplotsretval} % Get first element, save into \paramA
    \pgfplotstablegetelem{1}{0}\of\parameters \pgfmathsetmacro\paramB{\pgfplotsretval}
     $\pgfmathprintnumber{\paramA}\times x + \pgfmathprintnumber{\paramB}$    
}

\end{axis}
\end{tikzpicture}

\end{figure}
\centering
\pgfplotstabletypeset[
    dec sep align,
    fixed,
    columns/0/.style={
        column name=Parameter Value
    },
    columns/1/.style={
        column name=Standard Error
    }
]{parameters.dat}

\end{document}

score 1 · Answer 2 · edited Apr 13 '17 at 12:35

EDIT: I missed the part where you requested the error printing. So this answer doesn't really answer the question.

You can do it without using gnuplot:

\documentclass{article}
\usepackage{pgfplots}
\usepackage{pgfplotstable}
\pgfplotsset{
    compat=1.9,
    legend style={font=\footnotesize}
}
\begin{document}
\pgfplotstableread{
    X Y
    1 1 
    2 4 
    3 3
    4 7
}{\datatable}
\begin{tikzpicture}
    \begin{axis}[
        xlabel={Some xlabel},
        ylabel={Some yabel},
        legend cell align=left,
        legend pos=north west]
        \addplot[only marks] table {\datatable};
        \addlegendentry{some legend}
        \addplot table [y={create col/linear regression={y=Y}}]
        {\datatable};
        \addlegendentry{%
        $\pgfmathprintnumber{\pgfplotstableregressiona} \cdot x
        \pgfmathprintnumber[print sign]{\pgfplotstableregressionb}$ lin. Regression} %
    \end{axis}
\end{tikzpicture}
\end{document}

The code is adapted from Linear regression - trend line with pgfplots but slightly changed.

Pgfplots: Error of Linear Regression

2 Answers2

Linked