# pgfplots conditional data filtering

I'm wondering if there's a way to filter data points within the \addplot arguments. I have a data set (see MWE below) whose values correspond to one of three conditions (C). I'd like to be able to read the complete data set using pgfplots with three separate \addplot commands (one for each value of C) each of which filters out the lines that don't correspond to that C value, plotting each condition with separate markers etc. Is this possible or not?

I've looked at the skip coords between index and x filter/filter point options, but as these are arguments to the axis function (and hence alter the entire plot), these don't seem to do what I want.

Here's a MWE that shows what I want to achieve:

\documentclass{standalone}

\usepackage{pgfplots}

\usepackage{filecontents}

\begin{filecontents}
C   P1      P2      P3  sigP3
1   1.12E-4 0.06189 0.1865  0.0080
1   6.03E-3 0.00000 0.2517  0.0046
1   2.64E-2 0.00000 0.2247  0.0165
2   6.49E-5 0.00000 0.1906  0.0043
2   1.27E-5 0.00296 0.2120  0.0039
2   3.34E-4 0.00865 0.1709  0.0050
3   1.59E-2 0.01857 0.1596  0.0216
3   7.10E-4 0.01857 0.2547  0.0316
3   3.23E-5 0.00526 0.1557  0.0051
3   2.33E-4 0.01857 0.2008  0.0136
3   5.80E-4 0.01857 0.2389  0.0172
\end{filecontents}

\begin{document}

\begin{tikzpicture}
\pgfplotsset{legend cell align=right,legend style={font=\footnotesize,legend pos=outer north east}};
\begin{semilogxaxis}[
enlargelimits=false,axis on top,
width=12cm,height=8cm,
xlabel={$\Pi_1$},
ylabel={$\Pi_3$},
ymin=0.1,ymax=0.35,
xmin=1E-5,xmax=1E-1,
]

%% PLOT ONLY IF C = 1
\addplot+[only marks,error bars/.cd,y dir=both,y explicit] coordinates {
(1.12E-4,0.1865) +- (0,0.0080)
(6.03E-3,0.2517) +- (0,0.0046)
(2.64E-2,0.2247) +- (0,0.0165)
};

%% PLOT ONLY IF C = 2
\addplot+[only marks,error bars/.cd,y dir=both,y explicit] coordinates {
(6.49E-5,0.1906) +- (0,0.0043)
(1.27E-5,0.2120) +- (0,0.0039)
(3.34E-4,0.1709) +- (0,0.0050)
};

%% PLOT ONLY IF C = 3
\addplot+[only marks,error bars/.cd,y dir=both,y explicit] coordinates {
(1.59E-2,0.1596) +- (0,0.0216)
(7.10E-4,0.2547) +- (0,0.0316)
(3.23E-5,0.1557) +- (0,0.0051)
(2.33E-4,0.2008) +- (0,0.0136)
(5.80E-4,0.2389) +- (0,0.0172)
};

\end{semilogxaxis}
\end{tikzpicture}

\end{document}


You can use restrict expr to domain:

  \begin{filecontents*}{mydata.dat}
C   P1      P2      P3  sigP3
1   1.12E-4 0.06189 0.1865  0.0080
1   6.03E-3 0.00000 0.2517  0.0046
1   2.64E-2 0.00000 0.2247  0.0165
2   6.49E-5 0.00000 0.1906  0.0043
2   1.27E-5 0.00296 0.2120  0.0039
2   3.34E-4 0.00865 0.1709  0.0050
3   1.59E-2 0.01857 0.1596  0.0216
3   7.10E-4 0.01857 0.2547  0.0316
3   3.23E-5 0.00526 0.1557  0.0051
3   2.33E-4 0.01857 0.2008  0.0136
3   5.80E-4 0.01857 0.2389  0.0172
\end{filecontents*}

\documentclass[margin=10pt]{standalone}
\usepackage{pgfplots}
\pgfplotsset{compat=1.12}
\begin{document}
\begin{tikzpicture}
\pgfplotsset{
legend cell align=right,
legend style={font=\footnotesize,legend pos=outer north east}}
\begin{semilogxaxis}[
enlargelimits=false,axis on top,
width=12cm,height=8cm,
xlabel={$\Pi_1$},
ylabel={$\Pi_3$},
ymin=0.1,ymax=0.35,
xmin=1E-5,xmax=1E-1,
log basis x=10,
]
table[x=P1,y=P3,y error=sigP3,
restrict expr to domain={\thisrow{C}}{1:1}
]{mydata.dat};
table[x=P1,y=P3,y error=sigP3,
restrict expr to domain={\thisrow{C}}{2:2}
]{mydata.dat};
table[x=P1,y=P3,y error=sigP3,
restrict expr to domain={\thisrow{C}}{3:3}
]{mydata.dat};
\end{semilogxaxis}
\end{tikzpicture}
\end{document}


Or with \pgfplotsinvokeforeach

  \documentclass[margin=10pt]{standalone}
\usepackage{pgfplots}
\pgfplotsset{compat=1.12}
\begin{document}
\begin{tikzpicture}
\pgfplotsset{
legend cell align=right,
legend style={font=\footnotesize,legend pos=outer north east}}
\begin{semilogxaxis}[
enlargelimits=false,axis on top,
width=12cm,height=8cm,
xlabel={$\Pi_1$},
ylabel={$\Pi_3$},
ymin=0.1,ymax=0.35,
xmin=1E-5,xmax=1E-1,
log basis x=10,
]
\pgfplotsinvokeforeach{1,2,3}{