ggplot2 heatmaps:为类别使用不同的渐变

Question

这篇Learning R博客文章展示了如何使用ggplot2制作篮球统计数据的热图.完成的热图如下所示:

我的问题(受到杰克评论学习R博客文章的启发)是:是否可以针对不同类别的统计数据(攻击性,防御性,其他)使用不同的渐变颜色？

Answer 1

首先,从帖子重新创建图表,为更新的(0.9.2.1)版本更新它,该版本ggplot2具有不同的主题系统并附加更少的包:

nba <- read.csv("http://datasets.flowingdata.com/ppg2008.csv")
nba$Name <- with(nba, reorder(Name, PTS))

library("ggplot2")
library("plyr")
library("reshape2")
library("scales")

nba.m <- melt(nba)
nba.s <- ddply(nba.m, .(variable), transform,
               rescale = scale(value))

ggplot(nba.s, aes(variable, Name)) + 
  geom_tile(aes(fill = rescale), colour = "white") + 
  scale_fill_gradient(low = "white", high = "steelblue") + 
  scale_x_discrete("", expand = c(0, 0)) + 
  scale_y_discrete("", expand = c(0, 0)) + 
  theme_grey(base_size = 9) + 
  theme(legend.position = "none",
        axis.ticks = element_blank(), 
        axis.text.x = element_text(angle = 330, hjust = 0))

对不同类别使用不同的渐变颜色并不是那么简单.概念方法,映射fill到interaction(rescale, Category)(其中Category是攻击性/防御性/其他;见下文)不起作用,因为交互因子和连续变量给出了fill无法映射到的离散变量.

解决这个问题的方法是人为地进行这种交互,映射rescale到不同值的非重叠范围Category,然后用于scale_fill_gradientn将这些区域中的每一个映射到不同的颜色梯度.

首先创建类别.我认为这些映射到评论中的那些,但我不确定; 改变哪个变量在哪个类别中很容易.

nba.s$Category <- nba.s$variable
levels(nba.s$Category) <- 
  list("Offensive" = c("PTS", "FGM", "FGA", "X3PM", "X3PA", "AST"),
       "Defensive" = c("DRB", "ORB", "STL"),
       "Other" = c("G", "MIN", "FGP", "FTM", "FTA", "FTP", "X3PP", 
                   "TRB", "BLK", "TO", "PF"))

由于rescale在0的几个(3或4)范围内,不同的类别可以偏移一百来保持它们分开.同时,根据重新调整的值和颜色,确定每个颜色渐变的端点应该在哪里.

nba.s$rescaleoffset <- nba.s$rescale + 100*(as.numeric(nba.s$Category)-1)
scalerange <- range(nba.s$rescale)
gradientends <- scalerange + rep(c(0,100,200), each=2)
colorends <- c("white", "red", "white", "green", "white", "blue")

现在替换fill变量rescaleoffset并更改fill要使用的比例scale_fill_gradientn(记住重新缩放值):

ggplot(nba.s, aes(variable, Name)) + 
  geom_tile(aes(fill = rescaleoffset), colour = "white") + 
  scale_fill_gradientn(colours = colorends, values = rescale(gradientends)) + 
  scale_x_discrete("", expand = c(0, 0)) + 
  scale_y_discrete("", expand = c(0, 0)) + 
  theme_grey(base_size = 9) + 
  theme(legend.position = "none",
        axis.ticks = element_blank(), 
        axis.text.x = element_text(angle = 330, hjust = 0))

重新排序以获取相关统计数据是该reorder函数对各种变量的另一个应用:

nba.s$variable2 <- reorder(nba.s$variable, as.numeric(nba.s$Category))

ggplot(nba.s, aes(variable2, Name)) + 
  geom_tile(aes(fill = rescaleoffset), colour = "white") + 
  scale_fill_gradientn(colours = colorends, values = rescale(gradientends)) + 
  scale_x_discrete("", expand = c(0, 0)) + 
  scale_y_discrete("", expand = c(0, 0)) + 
  theme_grey(base_size = 9) + 
  theme(legend.position = "none",
        axis.ticks = element_blank(), 
        axis.text.x = element_text(angle = 330, hjust = 0))

Answer 2

这是一个更简单的建议,使用ggplot2美学来映射渐变和颜色类别.只需使用alpha-aesthetic来生成渐变,以及类别的填充美学.

这是代码,重构Brian Diggs的回复:

nba <- read.csv("http://datasets.flowingdata.com/ppg2008.csv")
nba$Name <- with(nba, reorder(Name, PTS))

library("ggplot2")
library("plyr")
library("reshape2")
library("scales")

nba.m <- melt(nba)
nba.s <- ddply(nba.m, .(variable), transform,
           rescale = scale(value))

nba.s$Category <- nba.s$variable
levels(nba.s$Category) <- list("Offensive" = c("PTS", "FGM", "FGA", "X3PM", "X3PA", "AST"),
   "Defensive" = c("DRB", "ORB", "STL"),
   "Other" = c("G", "MIN", "FGP", "FTM", "FTA", "FTP", "X3PP", "TRB", "BLK", "TO", "PF"))

然后,将rescale变量标准化为0到1之间:

nba.s$rescale = (nba.s$rescale-min(nba.s$rescale))/(max(nba.s$rescale)-min(nba.s$rescale))

现在,做绘图:

ggplot(nba.s, aes(variable, Name)) + 
  geom_tile(aes(alpha = rescale, fill=Category), colour = "white") + 
  scale_alpha(range=c(0,1)) +
  scale_x_discrete("", expand = c(0, 0)) + 
  scale_y_discrete("", expand = c(0, 0)) + 
  theme_grey(base_size = 9) + 
  theme(legend.position = "none",
        axis.ticks = element_blank(), 
        axis.text.x = element_text(angle = 330, hjust = 0)) +
  theme(panel.grid.major = element_blank(), panel.grid.minor = element_blank())

注意使用,alpha=rescale然后使用缩放alpha范围scale_alpha(range=c(0,1)),可以适应您的情节适当改变范围.