在 R 中绘制具有颜色和频率的混淆矩阵答案

【问题标题】：Plot A Confusion Matrix with Color and Frequency in R在 R 中绘制具有颜色和频率的混淆矩阵
【发布时间】：2019-05-23 21:24:41
【问题描述】：

我想绘制一个混淆矩阵，但我不想只使用热图，因为我认为它们的数值分辨率很差。相反，我还想在正方形中间绘制频率。例如，我喜欢这样的输出：

library(mlearning);
data("Glass", package = "mlbench")
Glass$Type <- as.factor(paste("Glass", Glass$Type))

summary(glassLvq <- mlLvq(Type ~ ., data = Glass));
(glassConf <- confusion(predict(glassLvq, Glass, type = "class"), Glass$Type))

plot(glassConf) # Image by default

但是，1.) 我不明白“01、02 等”是指沿每个轴。我们怎样才能摆脱它？ 2.）我希望“预测”作为“y”维度的标签，“实际”作为“x”维度的标签 3.) 我想用频率/概率替换绝对计数。

或者，是否有其他软件包可以做到这一点？

本质上，我希望在 R 中这样：

http://www.mathworks.com/help/releases/R2013b/nnet/gs/gettingstarted_nprtool_07.gif

或者：

http://c431376.r76.cf2.rackcdn.com/8805/fnhum-05-00189-HTML/image_m/fnhum-05-00189-g009.jpg

【问题讨论】：

标签： r plot confusion-matrix

【解决方案1】：

mlearning 包在绘制混淆矩阵方面似乎相当不灵活。

从您的 glassConf 对象开始，您可能想要执行以下操作：

prior(glassConf) <- 100 
# The above rescales the confusion matrix such that columns sum to 100.
opar <- par(mar=c(5.1, 6.1, 2, 2))
x <- x.orig <- unclass(glassConf)
x <- log(x + 0.5) * 2.33
x[x < 0] <- NA
x[x > 10] <- 10
diag(x) <- -diag(x)
image(1:ncol(x), 1:ncol(x),
      -(x[, nrow(x):1]), xlab='Actual', ylab='',
      col=colorRampPalette(c(hsv(h = 0, s = 0.9, v = 0.9, alpha = 1), 
                             hsv(h = 0, s = 0, v = 0.9, alpha = 1), 
                             hsv(h = 2/6, s = 0.9, v = 0.9, alpha = 1)))(41), 
      xaxt='n', yaxt='n', zlim=c(-10, 10))
axis(1, at=1:ncol(x), labels=colnames(x), cex.axis=0.8)
axis(2, at=ncol(x):1, labels=colnames(x), las=1, cex.axis=0.8)
title(ylab='Predicted', line=4.5)
abline(h = 0:ncol(x) + 0.5, col = 'gray')
abline(v = 0:ncol(x) + 0.5, col = 'gray')
text(1:6, rep(6:1, each=6), 
     labels = sub('^0$', '', round(c(x.orig), 0)))
box(lwd=2)
par(opar) # reset par

上面的代码使用了plot.confusion调用的confusionImage函数的一些片段。

【讨论】：

【解决方案2】：

这是我从 jbaums 优秀答案开发的用于绘制混淆矩阵的函数。
它很相似，但看起来更好一些（IMO），并且不会转置您提供的混淆矩阵，这可能会有所帮助。

### Function for plotting confusion matrices
confMatPlot = function(confMat, titleMy, shouldPlot = T) {
  #' Function for plotting confusion matrice
  #' 
  #' @param confMat: confusion matrix with counts, ie integers. 
  #' Fractions won't work
  #' @param titleMy: String containing plot title
  #' @return Nothing: It only plots

  ## Prepare data
  x.orig = confMat; rm(confMat)  # Lazy conversion to function internal variable name
  n = nrow(x.orig)  # conf mat is square by definition, so nrow(x) == ncol(x)
  opar <- par(mar = c(5.1, 8, 3, 2))
  x <- x.orig
  x <- log(x + 0.5)  # x<1 -> x<0 ,  x>=1 -> x>0
  x[x < 0] <- NA
  diag(x) <- -diag(x)  # change sign to give diagonal different color
  ## Plot confusion matrix
  image(1:n, 1:n,  # grid of coloured boxes
        # matrix giving color values for the boxes
        # t() and [,ncol(x):1] since image puts [1,1] in bottom left by default
        -t(x)[, n:1],  
        # ylab added later to avoid overlap with tick labels
        xlab = 'Actual', ylab = '',
        col = colorRampPalette(c("darkorange3", "white", "steelblue"), 
                               bias = 1.65)(100),
        xaxt = 'n', yaxt = 'n'
        )
  # Plot counts
  text(rep(1:n, each = n), rep(n:1, times = n), 
       labels = sub('^0$', '', round(c(x.orig), 0)))
  # Axis ticks but no lables
  axis(1, at = 1:n, labels = rep("", n), cex.axis = 0.8)
  axis(2, at = n:1, labels = rep("", n), cex.axis = 0.8)
  # Tilted axis lables
  text(cex = 0.8, x = (1:n), y = -0.1, colnames(x), xpd = T, srt = 30, adj = 1)
  text(cex = 0.8, y = (n:1), x = +0.1, colnames(x), xpd = T, srt = 30, adj = 1)
  title(main = titleMy)
  title(ylab = 'Predicted', line = 6)
  # Grid and box
  abline(h = 0:n + 0.5, col = 'gray')
  abline(v = 0:n + 0.5, col = 'gray')
  box(lwd = 1, col = 'gray')
  par(opar)
}

输出示例：

【讨论】：