gnuplot：如何在箱线图上显示时间序列？

Question

我想可视化某些数据随时间的变化。为此，我想将两个块（交叉/长）的时间进程显示为通过参考数据框上方的线连接的点。相邻的地块不应重叠。那么我该如何定义或构建时间尺度呢？

reset session

files          = "data.dat reference.dat"

# File "reference.dat" contains only index / block "norm"
ref_blocks     = "norm"

$REF<<EOD
# norm
RH   LH   RA   LA
12019.1 444.6   16159.2 62355.9
8256.7  171.9   18879.5 67602.1
7560.4  192.7   15405.6 59350.9
13985.5 542.3   18920   56517.7
5401.5  364.9   13164.1 48907.4
3856.3  375.8   15093   73012.8
9087.7  532.8   15255.4 52057.8
8948.1  313.4   13690.2 53685.8
6511    275.3   11455.2 45066
10404.1 264.6   15223.6 60208.2
EOD


# File "data.dat" contains indices / blocks "cross long"
blocks         = "cross long"

$DAT<<EOD
# cross
date      RH        LH        RA        LA
20200203  9500.8    164.3     13004.2   54644.1
20200522  9940.9    203.6     20142.7   49401.3
20210521  11226.3   222.7     14085.7   54759
20220527  13605.5   380.1     19472.8   46403.6
20220701  13206.5   376.3     12784     53552.9
20220706  13148.9   272.6     13455.7   51796.4
20221006  13832.9   311.6     13006.1   52337.4
20230309  13389.8   310.5     12677     54125.9
20230907  13806.3   334.8     13381.2   54700.4
20240919  15376.2   492.4     13050.4   53682.1


# long
date      RH        LH        RA        LA
20200203  9070.1    236.9     5505.8    53849.7
20200522  9489.2    258.5     6505.8    54003.9
20220527  13466     441.1     14128.5   50221.7
20220701  12934.3   400.4     14421.3   52927
20220706  12921.9   386.5     14539.9   50843.7
20221006  13440.1   351.2     15709.2   50322.8
20230309  13258.9   362.8     14416.2   52378.4
20210521  10771.3   294.8     14199.1   53795.7
20230907  13369.6   376.7     14974.9   53158.1
20240919  15216     478.7     14213     51990.8


EOD

# Both files contain columns with header names / labels "RH LH RA LA"
labels         = "RH LH RA LA"
myXtic(n)      = sprintf( "%s", word( labels, n ) )

set style fill solid 0.25 border -1
set style boxplot outliers pointtype 7
set style data boxplot
set boxwidth 0.7 absolute

set title "dat -> ref" font "Arial,14"
set xtics 1, 1, words(labels) scale 0
#set datafile separator '\t'

plot for [i=1:words(labels)] $REF index "norm" using (i):( column( word( labels, i ) ) ) notitle

#replot for [label in labels] $DAT index "long" using (i):( column( label ) ) notitle

编辑：是的，每个盒子应该有一个时间线。所有框应具有相同的颜色，并且所有交叉值应使用一种颜色，所有长值应使用一种颜色。

Answer 1

这是一个建议。一旦您通过

stats

知道了时间范围，您就需要缩放时间数据并将其移动到每个箱线图的位置和宽度。

• 不确定，为什么

  xtic()

  似乎不能与

  boxplot

  一起使用，所以你必须通过

  keyentry

  “手动”完成（检查

  help keyentry

  ）
• 对于超过两个数量级的数据，

  logscale y

  可能会有所帮助

检查以下最小化示例作为进一步优化的起点。

脚本：

### overlay boxplots with timedata
reset session

$REF<<EOD
# norm
RH   LH   RA   LA
12019.1 444.6   16159.2 62355.9
8256.7  171.9   18879.5 67602.1
7560.4  192.7   15405.6 59350.9
13985.5 542.3   18920   56517.7
5401.5  364.9   13164.1 48907.4
3856.3  375.8   15093   73012.8
9087.7  532.8   15255.4 52057.8
8948.1  313.4   13690.2 53685.8
6511    275.3   11455.2 45066
10404.1 264.6   15223.6 60208.2
EOD

$DAT<<EOD
# cross
date      RH        LH        RA        LA
20200203  9500.8    164.3     13004.2   54644.1
20200522  9940.9    203.6     20142.7   49401.3
20210521  11226.3   222.7     14085.7   54759
20220527  13605.5   380.1     19472.8   46403.6
20220701  13206.5   376.3     12784     53552.9
20220706  13148.9   272.6     13455.7   51796.4
20221006  13832.9   311.6     13006.1   52337.4
20230309  13389.8   310.5     12677     54125.9
20230907  13806.3   334.8     13381.2   54700.4
20240919  15376.2   492.4     13050.4   53682.1


# long
date      RH        LH        RA        LA
20200203  9070.1    236.9     5505.8    53849.7
20200522  9489.2    258.5     6505.8    54003.9
20220527  13466     441.1     14128.5   50221.7
20220701  12934.3   400.4     14421.3   52927
20220706  12921.9   386.5     14539.9   50843.7
20221006  13440.1   351.2     15709.2   50322.8
20230309  13258.9   362.8     14416.2   52378.4
20210521  10771.3   294.8     14199.1   53795.7
20230907  13369.6   376.7     14974.9   53158.1
20240919  15216     478.7     14213     51990.8
EOD

myHeaders     = "LH RH LA RA"
H             = words(myHeaders)
dataHeader(i) = column(word(myHeaders, i))
myBlocks      = "cross long"
B             = words(myBlocks)
myBlock(i)    = word(myBlocks, i)
myBoxwidth    =  0.8
myFmt         = "%Y%m%d"
t(col,i)      = ((timecolumn(col,myFmt)-t0)/(t1-t0) - 0.5)*myBoxwidth + i

stats $DAT u (timecolumn(1,myFmt)) nooutput
t0 = STATS_min
t1 = STATS_max

set title "dat -> ref" font "Arial,14"
set style boxplot outliers pointtype 7
set style data boxplot
set xrange [0.5:H+0.5] noextend
set logscale y
set grid y,my
set style fill solid 1.0 border -1
set boxwidth myBoxwidth absolute
set key left top reverse noautotitle

set linetype 1 lc rgb 0xaaff0000
set linetype 2 lc rgb 0xaa00aa00
set linetype 3 lc rgb 0x77ffff00

plot for [i=1:H] $REF index "norm" u (i):(dataHeader(i)) lt 3, \
     for [b=1:B] for [i=1:H] $DAT index myBlock(b) u (t(1,i)):(dataHeader(i)) w lp pt 7 lt b, \
     for [b=1:B] keyentry w lp pt 7 lt b ti myBlock(b), \
     keyentry w boxes lt 3 ti "Ref", \
     for [i=1:H] '+' every ::::0 u (i):(NaN):xtic(word(myHeaders, i))
### end of script

结果：