Question

我为heatmap.2绘制了一个图，用于在[0,1]上定义的两个变量。我现在想画一个abline(v=0.2)。这将是错误的＆＃34; （相对于我想要的，而不是R的内部logid，我推测）因为heatmap.2也将其他东西放在绘图区域中，比如颜色键（我猜，热图的轴是内部其他东西）我也只是重命名下面例子中的轴。）

我的问题：如何最方便地放置abline之类的东西，理想情况是根据实际热图的坐标而不是整个绘图区域。

rm(list=ls())
library(rpart)
library(gplots)
library(RColorBrewer)

N <- 150

# just some data
X1 <- runif(N)
X2 <- runif(N)
y <- as.factor((X1<.2+rnorm(N,sd=0.2))|(X2>.7+rnorm(N,sd=0.1)))   

theData <- data.frame(y,X1,X2)

X1grid <- seq(0,1,by=.02)
X2grid <- seq(0,1,by=.02)
newData <- expand.grid(X2=X2grid,X1=X1grid)

FitSingle <- rpart(y~.,data=theData)
PredictionsSingle <- matrix(as.logical(predict(FitSingle, newdata=newData, type="class"))+0,nrow=length(X2grid))

names <- rep("", length(X2grid)) 
names[seq(1,length(X2grid), length(X2grid)/10)] <- seq(0, .9, by=.1) # not fully checked, but should be OK
rownames(PredictionsSingle) <- names
colnames(PredictionsSingle) <- names

rc <- colorRampPalette(c("darkgreen", "white", "purple3"))(n = nrow(X2grid))
heatmapplotSingle <- heatmap.2(PredictionsSingle, Rowv=NA, Colv=NA, col=rc,density.info="none",trace="none",revC=T)
abline(v=0.2) # should be at 0.2, not at roughly .6. should also stay in the heatmap

Answer 1

如评论中所述，add.expr是要走的路：

heatmapplotSingle <- heatmap.2(PredictionsSingle, Rowv = NA, Colv = NA, col = rc, 
                           density.info = "none", trace = "none", revC = T,
                           main = "Classification Tree full sample", 
                           add.expr  = { segments(0.2*ncol(MeanPrediction), 0, 0.2*ncol(MeanPrediction),
                                                  .7*nrow(MeanPrediction), lwd = 3, col="darkslateblue");
                                         segments(0.2*ncol(MeanPrediction), .7*nrow(MeanPrediction), 1*ncol(MeanPrediction), 
                                                  .7*nrow(MeanPrediction), lwd = 3, col = "darkslateblue")}
                           , srtCol = 0)

R中的访问坐标heatmap.2

1 个答案: