Kdtyjb

I'm trying to create multiple plots from one pgfplotstable (data below). I can create the plots by manually editing the data (dataA and dataB below) but this is error-prone. Essentially, I need help:

• 
  partitioning the table based on column cat
  


• 
  duplicating the last row (except xmin is xmax from the previous row)


• 
  creating a loop to create multiple tikzpictures [this I can probably figure out myself!]

documentclass{article}

usepackage{pgfplotstable}



pgfplotsset{compat=1.15}



begin{document}

pgfplotstableread{

x   xmin xmax par1    y cat par2

0.5    0    1  yes  0.3   1   no

1.5    1    2  yes  0.6   1   no

2.5    2    3  yes  0.7   1   no

0.5    0    1  yes  0.4   2   no

1.5    1    2  yes  0.5   2   no

2.5    2    3  yes  0.9   2   no

}data



pgfplotstableread{

x   xmin xmax par1    y cat par2

0.5    0    1  yes  0.3   1   no

1.5    1    2  yes  0.6   1   no

2.5    2    3  yes  0.7   1   no

2.5    3    3  yes  0.7   1   no

}dataA



pgfplotstableread{

x   xmin xmax par1    y cat par2

0.5    0    1  yes  0.4   2   no

1.5    1    2  yes  0.5   2   no

2.5    2    3  yes  0.9   2   no

2.5    3    3  yes  0.9   2   no

}dataB



begin{tikzpicture}

  begin{axis}[ybar, ymin=0, ymax=1, xmin=-0.5, xmax=3.5]

    addplot[black, ybar interval] table[x=xmin,y=y] {dataA}closedcycle;

  end{axis}

end{tikzpicture}



begin{tikzpicture}

  begin{axis}[ybar, ymin=0, ymax=1, xmin=-0.5, xmax=3.5]

    addplot[black, ybar interval] table[x=xmin,y=y] {dataB}closedcycle;

  end{axis}

end{tikzpicture}



end{document}

Regarding partitioning the data, I have been able to print the partitioned table, but then how can I use it as input for addplot?

newcommand{filtertable}[2]{

pgfplotstabletypeset[

  columns/par1/.style={string type},

  columns/par2/.style={string type},

  row predicate/.code={%

  pgfplotstablegetelem{##1}{cat}of{#1}

  ifnumpgfplotsretval=#2relax

  elsepgfplotstableuserowfalsefi}

]{#1}

}



filtertable{data}{1}

filtertable{data}{2}

Regarding duplicating the last row, the only relevant value in the last rows is xmin. According to the PGF manual with ybar interval "The last y value will be ignored." I have unsuccessfully tried to create the same plot without the additional row. Note the distance between xmin and xmax must not always equal 1.

One way is to install a filter like

unbounded coords=discard,
x filter/.expression={thisrow{cat}==1 ? x : nan}

MWE:

documentclass{article}

usepackage{pgfplotstable}



pgfplotsset{compat=1.15}



begin{document}

pgfplotstableread{

x   xmin xmax par1    y cat par2

0.5    0    1  yes  0.3   1   no

1.5    1    2  yes  0.6   1   no

2.5    2    3  yes  0.7   1   no

0.5    0    1  yes  0.4   2   no

1.5    1    2  yes  0.5   2   no

2.5    2    3  yes  0.9   2   no

}data



pgfplotstableread{

x   xmin xmax par1    y cat par2

0.5    0    1  yes  0.3   1   no

1.5    1    2  yes  0.6   1   no

2.5    2    3  yes  0.7   1   no

2.5    3    3  yes  0.7   1   no

}dataA



pgfplotstableread{

x   xmin xmax par1    y cat par2

0.5    0    1  yes  0.4   2   no

1.5    1    2  yes  0.5   2   no

2.5    2    3  yes  0.9   2   no

2.5    3    3  yes  0.9   2   no

}dataB



begin{tikzpicture}

  begin{axis}[ybar, ymin=0, ymax=1, xmin=-0.5, xmax=3.5]

    addplot[black, ybar interval,unbounded coords=discard,

    x filter/.expression={thisrow{cat}==1 ? x : nan}] table[x=xmin,y=y] {data}closedcycle;

  end{axis}

end{tikzpicture}



begin{tikzpicture}

  begin{axis}[ybar, ymin=0, ymax=1, xmin=-0.5, xmax=3.5]

    addplot[black, ybar interval,unbounded coords=discard,

    x filter/.expression={thisrow{cat}==2 ? x : nan}] table[x=xmin,y=y] {data}closedcycle;

  end{axis}

end{tikzpicture}



end{document}