我有一个包含多个箱线图的图，按不同条件分组（5×5），如图所示
每组 5 个箱线图基于单个文件中定义的数据。我想添加一条线来连接每组箱线图的中位数；像这样
这是我的数据集的一个虚拟子集，它更大。因此，我想依靠一种不需要我直接输入不同中位数的方法，而是根据用于绘制箱线图的文件来计算它们。有没有简单的方法可以做到这一点？如果我可以决定要连接多少个箱线图，例如，只连接每组五个中的前三个，而不是全部五个，那将是一个加分项。
这是用于生成第一个图的 MWE
% used PGFPlots v1.14
\begin{filecontents}{audio1.tsv}
37 22 79 3 100
9 60 113 1 97
29 63 65 0 99
12 21 83 0 99
7 28 78 1 97
29 -10 73 1 98
17 32 73 1 98
10 58 78 0 95
17 45 73 2 99
38 28 91 0 96
\end{filecontents}
\begin{filecontents}{speech4.tsv}
17 47 97 0 95
3 53 103 1 97
37 70 104 1 99
21 41 109 0 100
36 53 86 3 98
37 58 90 3 99
8 30 81 1 97
19 44 99 4 96
20 20 118 3 96
27 2 107 3 99
\end{filecontents}
\begin{filecontents}{music3.tsv}
26 -5 79 4 100
12 61 101 0 99
13 32 116 1 98
10 17 94 3 93
9 81 75 4 100
7 50 58 0 98
30 56 79 2 99
19 28 93 1 92
10 31 49 1 99
11 22 46 0 100
\end{filecontents}
\documentclass[border=3mm]{standalone}
\usepackage{pgfplots}
\usepgfplotslibrary{
groupplots,
statistics,
}
\pgfplotscreateplotcyclelist{mycolorlist}{%
{color=red},
{color=green},
{color=blue},
{color=magenta},
{color=black}
}
\pgfplotsset{
compat=1.10,
width=8cm,
height=5cm,
cycle list name=mycolorlist,
boxplot/draw direction=y,
}
\begin{document}
\begin{tikzpicture}
\begin{groupplot}[
group style={
group size=1 by 2,
},
xmin=0,
xmax=20,
xtick={3,10,17},
xticklabels={a,b,c},
xlabel={something-x},
ylabel={something-y},
]
\nextgroupplot
% add here another variable so we can use it for calculations
\foreach \filename/\j in {
audio1/1,
speech4/2,
music3/3%
} {
\foreach \i in {0,...,4} {
% draw each originally row, now column as boxplot
\addplot+ [
boxplot={
% now calculate the positions depending on how they
% should be organized
draw position=7*(\j-1) + \i+1,
}
] table [y index=\i] {\filename.tsv};
}
}
\end{groupplot}
\end{tikzpicture}
\end{document}
答案1
稍微迂回地做这件事，因此可能会有更优雅的解决方案。
我在中线中间创建命名坐标，并使用\draw之外的法线groupplot绘制中线之间的线。若要仅连接每组中的前三个，请更改{1,...,4}为{1,...,2}。
% used PGFPlots v1.14
\begin{filecontents}{audio1.tsv}
37 22 79 3 100
9 60 113 1 97
29 63 65 0 99
12 21 83 0 99
7 28 78 1 97
29 -10 73 1 98
17 32 73 1 98
10 58 78 0 95
17 45 73 2 99
38 28 91 0 96
\end{filecontents}
\begin{filecontents}{speech4.tsv}
17 47 97 0 95
3 53 103 1 97
37 70 104 1 99
21 41 109 0 100
36 53 86 3 98
37 58 90 3 99
8 30 81 1 97
19 44 99 4 96
20 20 118 3 96
27 2 107 3 99
\end{filecontents}
\begin{filecontents}{music3.tsv}
26 -5 79 4 100
12 61 101 0 99
13 32 116 1 98
10 17 94 3 93
9 81 75 4 100
7 50 58 0 98
30 56 79 2 99
19 28 93 1 92
10 31 49 1 99
11 22 46 0 100
\end{filecontents}
\documentclass[border=3mm]{standalone}
\usepackage{pgfplots}
%\usetikzlibrary{
\usepgfplotslibrary{
groupplots,
statistics,
}
\pgfplotscreateplotcyclelist{mycolorlist}{%
{color=red},
{color=green},
{color=blue},
{color=magenta},
{color=black}
}
\pgfplotsset{
compat=1.10,
width=8cm,
height=5cm,
cycle list name=mycolorlist,
boxplot/draw direction=y,
}
\begin{document}
\begin{tikzpicture}
\begin{groupplot}[
group style={
group size=1 by 2,
},
xmin=0,
xmax=20,
xtick={3,10,17},
xticklabels={a,b,c},
xlabel={something-x},
ylabel={something-y},
]
\nextgroupplot
% add here another variable so we can use it for calculations
\foreach[count=\j] \filename in {
audio1,
speech4,
music3%
} {
\foreach \i in {0,...,4} {
% draw each originally row, now column as boxplot
\addplot+ [
boxplot={
% now calculate the positions depending on how they
% should be organized
draw position=7*(\j-1) + \i+1,
}
] table [y index=\i] {\filename.tsv}
node (tmp) at (boxplot box cs:\boxplotvalue{median},0.5) {}
;
\edef\tmp{\noexpand\coordinate (m-\j-\i) at (tmp.center);}
\tmp
}
}
\end{groupplot}
\foreach \j in {1,2,3}
\draw [thick] (m-\j-0) foreach \i in {1,...,4} { -- (m-\j-\i) };
\end{tikzpicture}
\end{document}