在ggduo散点图矩阵中显示相关指数(带有颜色) [英] Show correlation index in ggduo scatterplot matrix (with coloring)

问题描述

基于

要在@aosmith的答案中对齐图像中的相关标签，

来自帮助的

 #，但已修改PointsWithCor<-函数(数据，映射，...，方法=皮尔逊"){df<-data.frame(x = eval(映射$ x，数据)，y = eval(映射$ y，数据)，c = eval(映射$ color，数据))xPos =分钟(df $ x)yPos = max(df $ y)sumdf<-df％&％;％group_by(c)％&％;％总结(实验室=圆(cor(x，y)，3)，x = xPos，y = yPos * min(as.numeric(c))/max(as.numeric(df $ c)))ggally_points(数据，映射，...)+ggplot2 :: geom_label(数据= sumdf，映射= ggplot2 :: aes(x = x，y = y，标签= lab，颜色= c)，恰好= 0，恰好= 1，大小= 5，字体=粗体"，Inherit.aes = FALSE#不从...继承任何东西)} 

解决方案

这是一种方法.我发现关键是要按组估算标签值和轴位置.我使用了来自 dplyr 的帮助程序功能进行分组和汇总.

否则，这与您所做的相似，即使用情节中的 mapping .我将映射( x ， y ， colour )存储在data.frame中，以便进行汇总.

您可能需要处理轴的位置.您会看到min x和max y对于所有这些都不真正起作用.您可能决定以其他方式计算它们.

这是我做的功能:

 库(GGally)图书馆(dplyr)points_with_cor_color =函数(数据，映射，...，方法=皮尔逊"){dat = data.frame(x = data [，as.character(mapping $ x)]，y = data [，as.character(mapping $ y)]，color = data [，as.character(mapping $ colour)])sumdat = dat％&％;％group_by(颜色)％>％summarise(lab = round(cor(x，y，method = method)，3)，x = min(x，na.rm = TRUE)，y = max(y，na.rm = TRUE))ggally_points(数据，映射，...)+ggplot2 :: geom_label(数据=求和映射= ggplot2 :: aes(x = x，y = y，标签=实验室，颜色=颜色)，恰好= 0，恰好= 1，size = 5，fontface ="bold"，#不继承...Inherit.aes = FALSE)} 

这是带有 ggduo()的绘图代码.

  ggduo(df，columnsX = 1:2，columnY = 3:5，映射= ggplot2 :: aes(color = f)，类型=列表(连续= points_with_cor_color)) 

Based on this post, now that I have

library (GGally)

# from help
PointsWithCor <- function(data, mapping, ..., method = "pearson") {
  x <- eval(mapping$x, data)
  y <- eval(mapping$y, data)
  cor <- cor(x, y, method = method)
  ggally_points(data, mapping, ...) +
    ggplot2::geom_label(
      data = data.frame(
        x = min(x, na.rm = TRUE),
        y = max(y, na.rm = TRUE),
        lab = round(cor, digits = 3)
      ),
      mapping = ggplot2::aes(x = x, y = y, label = lab),
      hjust = 0, vjust = 1,
      size = 5, fontface = "bold",
      inherit.aes = FALSE # do not inherit anything from the ...
    )
}

# data frame
df = data.frame(runif(100),
                rnorm(100),
                rgamma(100,1,2),
                rt(100,1),
                rf(100,1,2),
                as.factor(round(runif(100,0,1))))
colnames(df) = c("a","b","c","d","e","f")

# points + cor, but only one cor index
ggduo(df,columnsX = 1:2, columnsY = 3:5,
      mapping = aes(colour = f),
      types = list(continuous = PointsWithCor))

but it produces a matrix of scatterplot with correlation in all x and all y. I'd like to show correlations colored by the same way to coloring the points in scatterplots.

I think it needs to modify the function to use the colour attribute in mapping, but not sure how to do it. Could anyone please give me a suggestion?

Edit:
To align the correlation labels in the image in the answer from @aosmith,

# from help but modified
PointsWithCor <- function(data, mapping, ..., method = "pearson") {
  df <- data.frame(x = eval(mapping$x, data), y = eval(mapping$y, data), c = eval(mapping$colour, data))

  xPos = min(df$x)
  yPos = max(df$y)

  sumdf <- df %>%
    group_by(c) %>%
    summarise(
      lab = round(cor(x, y),3),
      x = xPos,
      y = yPos*min(as.numeric(c))/max(as.numeric(df$c))
    )

  ggally_points(data, mapping, ...) +
    ggplot2::geom_label(
      data = sumdf,
      mapping = ggplot2::aes(x = x, y = y, label = lab, color = c),
      hjust = 0, vjust = 1,
      size = 5, fontface = "bold",
      inherit.aes = FALSE # do not inherit anything from the ...
    )
}

解决方案

Here is one approach. I found the key to be estimating the label values and the axis locations by group. I used helper functions from dplyr for the grouping and summarizing.

Otherwise this is similar to what you did, working with the mapping from the plot. I store the mappings (x, y, colour) in a data.frame so I can do the summarizing.

You'll likely want to work on the axis position placement. You'll see the min x and max y doesn't really work for all of these. You might decide to calculate them a different way.

Here is the function I made:

library(GGally)
library(dplyr)

points_with_cor_color = function(data, mapping, ..., method = "pearson") {
     dat = data.frame(x = data[, as.character(mapping$x)],
                      y = data[, as.character(mapping$y)],
                      color = data[, as.character(mapping$colour)])

     sumdat = dat %>%
          group_by(color) %>%
          summarise(lab = round(cor(x, y, method = method), 3),
                    x = min(x, na.rm = TRUE), 
                    y = max(y, na.rm = TRUE) )

     ggally_points(data, mapping, ...) +
          ggplot2::geom_label(
               data = sumdat,
               mapping = ggplot2::aes(x = x, y = y, label = lab, color = color),
               hjust = 0, vjust = 1,
               size = 5, fontface = "bold", # do not inherit anything from the ...
               inherit.aes = FALSE
          )
}

And here is the plotting code with ggduo().

ggduo(df,columnsX = 1:2, columnsY = 3:5,
      mapping = ggplot2::aes(color = f),
     types = list(continuous = points_with_cor_color))

这篇关于在ggduo散点图矩阵中显示相关指数(带有颜色)的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！