当 x 和 y 都是分类变量时，Marimekko/Mosaic 图是一个很好的默认图。使用 ggplot 创建这些的最佳方法是什么？
我能找到的唯一参考是这篇 4yo博客文章，但这似乎有点过时了。现在有更好或更简单的实现吗？GGally 包有一个功能ggally_ratio，但这会产生一些完全不同的东西：
问问题
24750 次
7 回答
28
前段时间我自己做了，使用 just geom_bar，我把它变成了一个通用函数，所以它应该适用于任何两个factors。
ggMMplot <- function(var1, var2){
require(ggplot2)
levVar1 <- length(levels(var1))
levVar2 <- length(levels(var2))
jointTable <- prop.table(table(var1, var2))
plotData <- as.data.frame(jointTable)
plotData$marginVar1 <- prop.table(table(var1))
plotData$var2Height <- plotData$Freq / plotData$marginVar1
plotData$var1Center <- c(0, cumsum(plotData$marginVar1)[1:levVar1 -1]) +
plotData$marginVar1 / 2
ggplot(plotData, aes(var1Center, var2Height)) +
geom_bar(stat = "identity", aes(width = marginVar1, fill = var2), col = "Black") +
geom_text(aes(label = as.character(var1), x = var1Center, y = 1.05))
}
ggMMplot(diamonds$cut, diamonds$clarity)
于 2013-10-08T20:46:20.820 回答
23
前段时间我在一个项目中遇到了同样的问题。我的解决方案是与选项geom_bar一起使用以适应不同的条宽：scales="free_x", space="free_x"facet_grid
# using diamonds dataset for illustration
df <- diamonds %>%
group_by(cut, clarity) %>%
summarise(count = n()) %>%
mutate(cut.count = sum(count),
prop = count/sum(count)) %>%
ungroup()
ggplot(df,
aes(x = cut, y = prop, width = cut.count, fill = clarity)) +
geom_bar(stat = "identity", position = "fill", colour = "black") +
# geom_text(aes(label = scales::percent(prop)), position = position_stack(vjust = 0.5)) + # if labels are desired
facet_grid(~cut, scales = "free_x", space = "free_x") +
scale_fill_brewer(palette = "RdYlGn") +
# theme(panel.spacing.x = unit(0, "npc")) + # if no spacing preferred between bars
theme_void()
于 2017-08-15T07:12:54.040 回答
13
第一次尝试。我不确定如何将因子标签放在轴上。
makeplot_mosaic <- function(data, x, y, ...){
xvar <- deparse(substitute(x))
yvar <- deparse(substitute(y))
mydata <- data[c(xvar, yvar)];
mytable <- table(mydata);
widths <- c(0, cumsum(apply(mytable, 1, sum)));
heights <- apply(mytable, 1, function(x){c(0, cumsum(x/sum(x)))});
alldata <- data.frame();
allnames <- data.frame();
for(i in 1:nrow(mytable)){
for(j in 1:ncol(mytable)){
alldata <- rbind(alldata, c(widths[i], widths[i+1], heights[j, i], heights[j+1, i]));
}
}
colnames(alldata) <- c("xmin", "xmax", "ymin", "ymax")
alldata[[xvar]] <- rep(dimnames(mytable)[[1]],rep(ncol(mytable), nrow(mytable)));
alldata[[yvar]] <- rep(dimnames(mytable)[[2]],nrow(mytable));
ggplot(alldata, aes(xmin=xmin, xmax=xmax, ymin=ymin, ymax=ymax)) +
geom_rect(color="black", aes_string(fill=yvar)) +
xlab(paste(xvar, "(count)")) + ylab(paste(yvar, "(proportion)"));
}
例子：
makeplot_mosaic(mtcars, vs, gear)
于 2013-10-08T15:47:52.617 回答
9
您可以使用名为“ggmosaic”（https://github.com/haleyjeppson/ggmosaic）的 ggplot2 扩展包。
https://cran.r-project.org/web/packages/ggmosaic/vignettes/ggmosaic.html提供了带有示例代码和视觉结果的广泛教程。
于 2016-12-27T16:23:10.273 回答
5
Plotluck是一个基于ggplot2的库，旨在根据 1-3 个变量的特征自动选择绘图类型。它包含马赛克图的功能。例子：
plotluck(mtcars,vs,gear)
于 2015-04-17T06:03:54.233 回答
4
按照user2030503的建议，这里有一个使用ggmosaic. （请注意ggplot 3.0 破坏了一些 ggmosaic，因此您需要最新版本。）
library(tidyverse)
library(ggmosaic)
# Data copied from linked blog post
df <- data.frame(
segment = LETTERS[1:4],
segpct = c(40, 30, 20, 10),
Alpha = c(60, 40, 30, 25),
Beta = c(25, 30, 30, 25),
Gamma = c(10, 20, 20, 25),
Delta = c(5, 10, 20, 25)
)
# Convert to "long" for plotting
df_long <- gather(df, key = "greek_letter", value = "pct",
-c("segment", "segpct")) %>%
mutate(
greek_letter = factor(
greek_letter,
levels = c("Alpha", "Beta", "Gamma", "Delta")
),
weight = (segpct * pct) / 10000
)
# Plot
ggplot(df_long) +
geom_mosaic(aes(x = product(greek_letter, segment), fill = greek_letter,
weight = weight))
于 2020-01-22T18:54:00.260 回答
2
感谢所有创建此条目的人，这确实帮助了我，因为 ggmosaic 没有做我想要的（并且没有正确标记轴）。Z.Lin 的 nice 函数在https://github.com/tidyverse/ggplot2/issues/3142中给出了一种解释的警告，这似乎是在说，在技术上其内容不真实的警告实际上是在警告我们， ggplotocracy，祝福并感谢他们，觉得 geom_bar 不应该真的有可变宽度。我想我明白了这一点，所以我从 Jake Fisher 那里得到了这个功能，并根据我自己的需要对其进行了调整。如果它对其他人有用，这里是：
makeplot_mosaic2 <- function(data, x, y, statDigits = 1, residDigits = 1, pDigits = 3, ...){
### from https://stackoverflow.com/questions/19233365/how-to-create-a-marimekko-mosaic-plot-in-ggplot2,
### this from Jake Fisher (I think)
xvar <- deparse(substitute(x))
yvar <- deparse(substitute(y))
mydata <- data[c(xvar, yvar)]
mytable <- table(mydata)
widths <- c(0, cumsum(apply(mytable, 1, sum)))
heights <- apply(mytable, 1, function(x){c(0, cumsum(x/sum(x)))})
alldata <- data.frame()
allnames <- data.frame()
for(i in 1:nrow(mytable)){
for(j in 1:ncol(mytable)){
alldata <- rbind(alldata, c(widths[i], widths[i+1], heights[j, i], heights[j+1, i]))
}
}
colnames(alldata) <- c("xmin", "xmax", "ymin", "ymax")
alldata[[xvar]] <- rep(dimnames(mytable)[[1]],rep(ncol(mytable), nrow(mytable)))
alldata[[yvar]] <- rep(dimnames(mytable)[[2]],nrow(mytable))
chisq <- chisq.test(mytable)
df <- chisq$parameter
pval <- chisq$p.value
chisqval <- chisq$statistic
# stdResids <- chisq$stdres
alldata$xcent <- (alldata$xmin + alldata$xmax)/2
alldata$ycent <- (alldata$ymin + alldata$ymax)/2
alldata$stdres <- round(as.vector(t(chisq$stdres)), residDigits)
# print(chisq$stdres)
# print(alldata)
titleTxt1 <- paste0("Mosaic plot of ",
yvar,
" against ",
xvar,
", ")
titleTxt2 <- paste0("chisq(",
df,
") = ",
round(chisqval, statDigits),
", p = ",
format.pval(pval, digits = pDigits))
titleTxt <- paste0(titleTxt1, titleTxt2)
subTitleTxt <- "Cell labels are standardised residuals"
ggplot(data = alldata,
aes(xmin=xmin, xmax=xmax, ymin=ymin, ymax=ymax)) +
geom_rect(color="black", aes_string(fill=yvar)) +
geom_text(aes(x = xcent, y = ycent, label = stdres)) +
xlab(paste0("Count of '",
xvar,
"', total = ",
max(alldata$xmax))) + # tweaked by CE
ylab(paste0("Proportion of '",
yvar,
"' per level of '",
xvar,
"'")) +
ggtitle(titleTxt,
subtitle = subTitleTxt) +
theme_bw() +
theme(plot.title = element_text(hjust = .5),
plot.subtitle = element_text(hjust = .5))
}
makeplot_mosaic2(mtcars, vs, gear)
makeplot_mosaic2(diamonds, cut, clarity)
于 2020-09-20T09:10:49.387 回答