gpt4 book ai didi

r - 生成具有交替轴和不同比例的堆叠多面板图

转载 作者:行者123 更新时间:2023-12-05 01:00:50 25 4
gpt4 key购买 nike

想象一下我们有这样的数据:

    dat <- structure(list(variable = c("a1", "a1", "a1", "a1", "a1", "a1", 
"a2", "a2", "a2", "a2", "a2", "a2", "a3", "a3", "a3", "a3", "a3",
"a3", "a4", "a4", "a4", "a4", "a4", "a4"), value = c(9.17804065427195,
-0.477515191225569, 0.189943035684685, -6.06095979017212, -10.4173631972868,
-6.119330192816, -14.3820530117637, 13.9823789620469, 15.6437973890843,
0.754856919261315, -0.887052526388938, 7.4096244573169, 0.61043977214679,
28.4639357142541, 15.4511442682744, 15.8118136384483, 6.65940292893,
0.467862281678766, 482.791905769932, 493.606761379037, 491.254828253119,
504.323684433231, 499.323576709646, 492.625278087471)), .Names = c("variable",
"value"), row.names = c(NA, -24L), class = "data.frame")

我要作图 value对比 value每个 variable ,这样我就有了以下格式的 6 个面板,其中字母代表轴所在的位置和一个 p显示面板所在的位置。
a2   p
a3 p p
a4 p p p
a1 a2 a3

我知道我可以绘制每个并安排数据是否很长......,例如
par(.....)
plot(a1 ~ a2, data=longdat)
plot(a1 ~ a3, data=longdat)
plot(a1 ~ a4, data=longdat)
......

如果这就是我所能做的,那么有一种快速的方法吗?
最好我想知道是否已经有办法做到这一点,比如 facet_wrapfacet_gridggplot2 Lattice似乎有我想要的那种形状的图(见下文),但我只能看到如何为 using two axes split per factor 做到这一点.这里的直方图不是必需的,这只是一个例子。

enter image description here

格子可以做类似的事情,但不是我想要的..
xyplot(value~value|variable, 
data = a,
scales=list(alternating=FALSE,relation="same"),
layout=c(2,2))

通过重新排序数据,我可以使这项工作正常进行,但是当您更改时 relation"free"所以你在轴上为每个变量获得不同的比例,然后将面板分解成单独的面板。

最佳答案

编辑:使用 GGally (v1.0.1)

使用 ggpairs() 更容易来自 GGally 的函数包裹。让 ggpairs()绘制和定位散点图,然后从结果图中删除不需要的元素。首先,以宽格式转换数据。

# Packages
library(GGally)
library(ggplot2)
library(tidyr)

# Data
dat <- structure(list(variable = c("a1", "a1", "a1", "a1", "a1", "a1",
"a2", "a2", "a2", "a2", "a2", "a2", "a3", "a3", "a3", "a3", "a3",
"a3", "a4", "a4", "a4", "a4", "a4", "a4"),
value = c(9.17804065427195,
-0.477515191225569, 0.189943035684685, -6.06095979017212, -10.4173631972868,
-6.119330192816, -14.3820530117637, 13.9823789620469, 15.6437973890843,
0.754856919261315, -0.887052526388938, 7.4096244573169, 0.61043977214679,
28.4639357142541, 15.4511442682744, 15.8118136384483, 6.65940292893,
0.467862281678766, 482.791905769932, 493.606761379037, 491.254828253119,
504.323684433231, 499.323576709646, 492.625278087471)), .Names = c("variable",
"value"), row.names = c(NA, -24L), class = "data.frame")

# Get the data in its wide format
dat$id <- sequence(rle(as.character(dat$variable))$lengths)
dat2 = spread(data = dat, key = variable, value = value)


# Base plot
gg = ggpairs(dat2,
columns = 2:5,
lower = list(continuous = "points"),
diag = list(continuous = "blankDiag"),
upper = list(continuous = "blank"))

使用来自 here 的代码修剪掉未命名的元素
# Trim off the diagonal spaces
n <- gg$nrow
gg$nrow <- gg$ncol <- n-1
v <- 1:n^2
gg$plots <- gg$plots[v > n & v%%n != 0]

# Trim off the last x axis label
# and the first y axis label
gg$xAxisLabels <- gg$xAxisLabels[-n]
gg$yAxisLabels <- gg$yAxisLabels[-1]

# Draw the plot
gg = gg +
theme_bw() +
theme(panel.grid = element_blank())
gg

enter image description here

原装
pairs()函数让您接近,但如果您只需要布局矩阵中显示的六个面板,那么您可能必须手动构建它。您可以使用 grid 构建图表, 或 ggplotgtable .这是一个 ggplot / gtable版本。

该脚本适用于您的 dat数据文件(即长格式)。它构造了六个 ggplot 的列表散点图。 ggplots 转换为 grobs,并提取相关轴 - 将成为新图表中左侧和底部轴的轴。构建了 gtable 布局,并将散点图 grobs(仅限绘图面板)添加到布局中。修改布局以采用轴,然后再次修改布局以采用可变标签。最后,还有一点整理。
dat <- structure(list(variable = c("a1", "a1", "a1", "a1", "a1", "a1", 
"a2", "a2", "a2", "a2", "a2", "a2", "a3", "a3", "a3", "a3", "a3",
"a3", "a4", "a4", "a4", "a4", "a4", "a4"),
value = c(9.17804065427195,
-0.477515191225569, 0.189943035684685, -6.06095979017212, -10.4173631972868,
-6.119330192816, -14.3820530117637, 13.9823789620469, 15.6437973890843,
0.754856919261315, -0.887052526388938, 7.4096244573169, 0.61043977214679,
28.4639357142541, 15.4511442682744, 15.8118136384483, 6.65940292893,
0.467862281678766, 482.791905769932, 493.606761379037, 491.254828253119,
504.323684433231, 499.323576709646, 492.625278087471)), .Names = c("variable",
"value"), row.names = c(NA, -24L), class = "data.frame")

# Load packages
library("ggplot2")
library("plyr")
library("gtable")
library(grid)

# Number of items and item labels
item = unique(dat$variable)
n = length(item)

## List of scatterplots
scatter <- list()

for (i in 1:(n-1)) {
for (j in (i+1):n) {

# Data frame
df.point <- na.omit(data.frame(cbind(x = dat[dat$variable == item[i], 2], y = dat[dat$variable == item[j], 2])))

# Plot
p <- ggplot(df.point, aes(x, y)) +
geom_point(size = 1) +
theme_bw() +
theme(panel.grid = element_blank(),
axis.text = element_text(size = 6))

name <- paste0("Item", i, j)
scatter[[name]] <- p
} }

# Convert ggplots to grobs
scatterGrob <- llply(scatter, ggplotGrob)

# Extract the axes as grobs
# x axis
xaxes = subset(scatterGrob, grepl(paste0("^Item.", n), names(scatterGrob)))
xaxes = llply(xaxes, gtable_filter, "axis-b")

# y axis
yaxes = subset(scatterGrob, grepl("^Item1.*", names(scatterGrob)))
yaxes = llply(yaxes, gtable_filter, "axis-l")

# Tick marks and tick mark labels are easier to position if they are separated.
labelsb = list(); ticksb = list(); labelsl = list(); ticksl = list()
for(i in 1:(n-1)) {
x = xaxes[[i]][[1]][[1]]$children[[2]]
labelsb[[i]] = x$grobs[[2]]
ticksb[[i]] = x$grobs[[1]]

y = yaxes[[i]][[1]][[1]]$children[[2]]
labelsl[[i]] = y$grobs[[1]]
ticksl[[i]] = y$grobs[[2]]
}

## Extract the plot panels
scatterGrob <- llply(scatterGrob, gtable_filter, "panel")

## Set up initial gtable layout
gt <- gtable(unit(rep(1, n-1), "null"), unit(rep(1, n-1), "null"))

# Add scatterplots in the lower half of the matrix
k <- 1
for (i in 1:(n-1)) {
for (j in i:(n-1)) {
gt <- gtable_add_grob(gt, scatterGrob[[k]], t=j, l=i)
k <- k+1
} }

# Add rows and columns for axes
gt <- gtable_add_cols(gt, unit(0.25, "lines"), 0)
gt <- gtable_add_cols(gt, unit(1, "lines"), 0)
gt <- gtable_add_rows(gt, unit(0.25, "lines"), 2*(n-1))
gt <- gtable_add_rows(gt, unit(0.5, "lines"), 2*(n-1))

for (i in 1:(n-1)) {
gt <- gtable_add_grob(gt, ticksb[[i]], t=(n-1)+1, l=i+2)
gt <- gtable_add_grob(gt, labelsb[[i]], t=(n-1)+2, l=i+2)
gt <- gtable_add_grob(gt, ticksl[[i]], t=i, l=2)
gt <- gtable_add_grob(gt, labelsl[[i]], t=i, l=1)
}

# Add rows and columns for variable names
gt <- gtable_add_cols(gt, unit(1, "lines"), 0)
gt <- gtable_add_rows(gt, unit(1, "lines"), n+1)
for(i in 1:(n-1)) gt <- gtable_add_grob(gt,
textGrob(item[i], gp = gpar(fontsize = 8)), t=n+2, l=i+3)
for(i in 2:n) gt <- gtable_add_grob(gt,
textGrob(item[i], rot = 90, gp = gpar(fontsize = 8)), t=i-1, l=1)

# Add small gaps between the panels
for(i in (n-1):2) {
gt <- gtable_add_cols(gt, unit(0.4, "lines"), i+2)
gt <- gtable_add_rows(gt, unit(0.4, "lines"), i-1)
}

# Add margins to the whole plot
for(i in c(2*(n-1)+2, 0)) {
gt <- gtable_add_cols(gt, unit(.75, "lines"), i)
gt <- gtable_add_rows(gt, unit(.75, "lines"), i)
}

# Turn clipping off
gt$layout$clip = "off"

# Draw it
grid.newpage()
grid.draw(gt)

enter image description here

关于r - 生成具有交替轴和不同比例的堆叠多面板图,我们在Stack Overflow上找到一个类似的问题: https://stackoverflow.com/questions/28943553/

25 4 0
Copyright 2021 - 2024 cfsdn All Rights Reserved 蜀ICP备2022000587号
广告合作:1813099741@qq.com 6ren.com