зацикливание функции ggscatter для уникальных комбинаций переменных x, y из столбцов данных

Question

Мне нужно нарисовать диаграммы рассеяния для оценки линейной регрессии. Я хочу использовать функцию "ggscatter", но мне нужно писать циклы.

Я проверил много примеров на этой платформе и других сайтах, но не смог найти подходящего решения.

Данные необходимо ввести как 45 уникальных комбинаций переменных x, y, соответствующих отдельным столбцам. Таким образом, будет 45 графиков, и эти графики могут быть нарисованы как субплоты на многоплановой фигуре с 4 столбцами и 4 строками. Итак, на выходе будет 3 файла рисунков.

Например, уникальными комбинациями x, y являются A, B; А, С; ОБЪЯВЛЕНИЕ; ... ДО Н.Э; В, D; БЫТЬ; ... ПРИВЕТ; Н, J; и я, J.

x и y изменяются в виде пар, вместо того, чтобы одна из них была постоянной.

Размеры кадра данных равны ncol = 10, nrow = 28

    A   B   C   D   E   F   G   H   I   J
0.99    0.13    0.88    0.07    7.83    3.63    6.78    14.5    76.7    880.3
0.29    0.84    0.75    0.93    2.69    7.71    8.23    84.7    12.9    476.8
0.54    0.22    0.99    0.60    8.57    8.53    4.56    79.1    21.3    686.9
0.84    0.89    0.82    0.50    0.48    8.40    2.96    24.5    35.5    300.1
0.58    0.98    0.86    0.41    3.05    6.48    2.36    75.5    43.5    993.4
0.21    0.78    0.64    0.90    3.32    3.03    0.02    89.2    18.1    358.0
0.07    0.46    0.66    0.42    2.97    9.35    0.42    68.2    59.2    576.5
0.19    0.01    0.89    0.69    6.59    6.93    4.23    46.6    74.1    299.7
0.33    0.05    0.60    0.58    9.86    1.37    5.20    79.2    28.9    494.1
0.25    0.50    0.01    0.29    7.64    6.23    4.80    48.4    96.5    762.2
0.30    0.96    0.71    0.85    6.89    5.08    0.48    50.3    80.8    396.7
0.03    0.18    0.25    0.56    0.57    3.83    7.91    58.5    43.6    201.7
0.59    0.62    0.74    0.62    0.59    0.00    2.62    63.3    51.6    416.0
0.53    0.18    0.21    0.96    9.68    1.80    1.71    69.5    28.4    584.1
0.25    0.05    0.46    0.92    9.16    2.64    8.15    24.5    80.7    699.0
0.33    0.88    0.99    0.70    1.60    5.38    0.26    64.4    60.0    265.3
0.06    0.57    0.75    0.69    5.55    0.02    0.77    47.5    82.4    729.1
0.90    0.55    0.26    0.26    1.38    6.24    0.31    28.4    61.5    181.5
0.77    0.74    0.42    0.94    3.35    7.75    3.46    22.2    78.5    347.8
0.61    0.16    0.98    0.51    8.81    8.27    3.86    42.4    67.8    303.1
0.26    0.58    0.71    0.24    1.97    9.14    5.67    65.6    63.9    967.4
0.16    0.35    0.26    0.23    4.25    3.98    9.07    99.2    83.6    817.9
0.71    0.87    0.31    0.94    0.55    5.72    0.15    86.3    41.8    579.1
0.23    0.45    0.19    0.45    0.10    0.52    5.53    46.8    47.7    788.5
0.57    0.82    0.68    0.55    2.23    6.83    4.93    67.2    94.4    482.7
0.93    0.26    0.89    0.48    6.85    3.95    6.82    76.2    88.9    551.9
0.70    0.72    0.02    0.99    1.07    0.62    8.03    19.4    88.9    355.9
0.06    0.24    0.46    0.74    5.18    1.15    8.69    18.4    100.0   676.7

R скрипт выглядит следующим образом:

setwd("working_directory")
data <- read.table("data.txt", header=TRUE)
attach(data)

str(data)

### It was possible to draw histograms as below
### Multiple histogram plot (uses single x variable with plot function) 

library(rcompanion)

histplot <- function (df) {
    plots <- vector('list', ncol(df))

    for (i in seq_along(df)) {
        plots[[i]] <- local({
            i <- i
            p1 <- plotNormalHistogram(df[[i]], 
                prob=FALSE,
                main=paste("Histogram of",colnames(df[i]),"\n with Std.Norm. Curve"))
        })
    }
}

dev.new()
par(mfrow=c(3,4))
histplot(data)

### Multiple scatterplot (uses two variable (x,y) with plot function)

### There are 45 unique combinations for 10 different data columns and I need to write a loop for this plot, but I couldn't write it 

library(ggplot2)
library(ggpubr)

### multiplot function for defining subplot layouts with ggplot functions 

multiplot <- function(..., plotlist=NULL, file, cols=1, layout=NULL) {
  require(grid)

  plots <- c(list(...), plotlist)

  numPlots = length(plots)

  if (is.null(layout)) {
    layout <- matrix(seq(1, cols * ceiling(numPlots/cols)),
                    ncol = cols, nrow = ceiling(numPlots/cols))
  }

 if (numPlots==1) {
    print(plots[[1]])

  } else {
    grid.newpage()
    pushViewport(viewport(layout = grid.layout(nrow(layout), ncol(layout))))

    for (i in 1:numPlots) {
      matchidx <- as.data.frame(which(layout == i, arr.ind = TRUE))

      print(plots[[i]], vp = viewport(layout.pos.row = matchidx$row,
                                      layout.pos.col = matchidx$col))
    }
  }
}

### scatterplot function (I need a loop after here but I couldn't write it)

dev.new()

p1 <- ggscatter(data, x="A",y="B", add="reg.line", conf.int=TRUE, cor.coef=TRUE, cor.method="pearson", xlab="A", ylab="B")
p2 <- ggscatter(data, x="A",y="C", add="reg.line", conf.int=TRUE, cor.coef=TRUE, cor.method="pearson", xlab="A", ylab="C")
p3 <- ggscatter(data, x="A",y="D", add="reg.line", conf.int=TRUE, cor.coef=TRUE, cor.method="pearson", xlab="A", ylab="D")
p4 <- ggscatter(data, x="A",y="E", add="reg.line", conf.int=TRUE, cor.coef=TRUE, cor.method="pearson", xlab="A", ylab="E")
p5 <- ggscatter(data, x="A",y="F", add="reg.line", conf.int=TRUE, cor.coef=TRUE, cor.method="pearson", xlab="A", ylab="F")
p6 <- ggscatter(data, x="A",y="G", add="reg.line", conf.int=TRUE, cor.coef=TRUE, cor.method="pearson", xlab="A", ylab="G")
p7 <- ggscatter(data, x="A",y="H", add="reg.line", conf.int=TRUE, cor.coef=TRUE, cor.method="pearson", xlab="A", ylab="H")
p8 <- ggscatter(data, x="A",y="I", add="reg.line", conf.int=TRUE, cor.coef=TRUE, cor.method="pearson", xlab="A", ylab="I")
p9 <- ggscatter(data, x="A",y="J", add="reg.line", conf.int=TRUE, cor.coef=TRUE, cor.method="pearson", xlab="A", ylab="J")
p10 <- ggscatter(data, x="B",y="C", add="reg.line", conf.int=TRUE, cor.coef=TRUE, cor.method="pearson", xlab="B", ylab="C")
p11 <- ggscatter(data, x="B",y="D", add="reg.line", conf.int=TRUE, cor.coef=TRUE, cor.method="pearson", xlab="B", ylab="D")
p12 <- ggscatter(data, x="B",y="E", add="reg.line", conf.int=TRUE, cor.coef=TRUE, cor.method="pearson", xlab="B", ylab="E")
p13 <- ggscatter(data, x="B",y="F", add="reg.line", conf.int=TRUE, cor.coef=TRUE, cor.method="pearson", xlab="B", ylab="F")
p14 <- ggscatter(data, x="B",y="G", add="reg.line", conf.int=TRUE, cor.coef=TRUE, cor.method="pearson", xlab="B", ylab="G")
p15 <- ggscatter(data, x="B",y="H", add="reg.line", conf.int=TRUE, cor.coef=TRUE, cor.method="pearson", xlab="B", ylab="H")
p16 <- ggscatter(data, x="B",y="I", add="reg.line", conf.int=TRUE, cor.coef=TRUE, cor.method="pearson", xlab="B", ylab="I")

multiplot(plotlist = list(p1,p2,p3,p4,p5,p6,p7,p8,p9,p10,p11,p12,p13,p14,p15,p16), cols = 4)

dev.new()

p17 <- ...
.
.
p32 <- ...

multiplot(...)

dev.new()

p33 <- ...
.
.
p45 <- ...

multiplot(...)

Answer 1

Вы можете попробовать что-то подобное (я использовал plot_grid из cowplot вместо multiplot, чтобы сократить воспроизводимый пример)

library(ggplot2)
library(ggpubr)
#> Loading required package: magrittr
library(cowplot)
#> 
#> ********************************************************
#> Note: As of version 1.0.0, cowplot does not change the
#>   default ggplot2 theme anymore. To recover the previous
#>   behavior, execute:
#>   theme_set(theme_cowplot())
#> ********************************************************
#> 
#> Attaching package: 'cowplot'
#> The following object is masked from 'package:ggpubr':
#> 
#>     get_legend

data <- structure(list(A = c(0.99, 0.29, 0.54, 0.84, 0.58, 0.21, 0.07, 0.19, 0.33, 0.25, 0.3, 0.03, 0.59, 0.53, 0.25, 0.33, 0.06, 0.9, 0.77, 0.61, 0.26, 0.16, 0.71, 0.23, 0.57, 0.93, 0.7, 0.06), B = c(0.13, 0.84, 0.22, 0.89, 0.98, 0.78, 0.46, 0.01, 0.05, 0.5, 0.96, 0.18, 0.62, 0.18, 0.05, 0.88, 0.57, 0.55, 0.74, 0.16, 0.58, 0.35, 0.87, 0.45, 0.82, 0.26, 0.72, 0.24), C = c(0.88, 0.75, 0.99, 0.82, 0.86, 0.64, 0.66, 0.89, 0.6, 0.01, 0.71, 0.25, 0.74, 0.21, 0.46, 0.99, 0.75, 0.26, 0.42, 0.98, 0.71, 0.26, 0.31, 0.19, 0.68, 0.89, 0.02, 0.46), D = c(0.07, 0.93, 0.6, 0.5, 0.41, 0.9, 0.42, 0.69, 0.58, 0.29, 0.85, 0.56, 0.62, 0.96, 0.92, 0.7, 0.69, 0.26, 0.94,0.51, 0.24, 0.23, 0.94, 0.45, 0.55, 0.48, 0.99, 0.74), E = c(7.83, 2.69, 8.57, 0.48, 3.05, 3.32, 2.97, 6.59, 9.86, 7.64, 6.89, 0.57, 0.59, 9.68, 9.16, 1.6, 5.55, 1.38, 3.35, 8.81, 1.97, 4.25, 0.55, 0.1, 2.23, 6.85, 1.07, 5.18), F = c(3.63, 7.71, 8.53, 8.4, 6.48, 3.03, 9.35, 6.93, 1.37, 6.23, 5.08, 3.83, 0, 1.8, 2.64, 5.38, 0.02, 6.24, 7.75, 8.27, 9.14, 3.98, 5.72, 0.52, 6.83, 3.95, 0.62, 1.15), G = c(6.78, 8.23, 4.56, 2.96, 2.36, 0.02, 0.42, 4.23, 5.2, 4.8, 0.48, 7.91, 2.62, 1.71, 8.15, 0.26, 0.77, 0.31, 3.46, 3.86, 5.67, 9.07, 0.15, 5.53, 4.93, 6.82, 8.03, 8.69), H = c(14.5, 84.7, 79.1, 24.5, 75.5, 89.2, 68.2, 46.6, 79.2, 48.4, 50.3, 58.5, 63.3, 69.5, 24.5, 64.4, 47.5, 28.4, 22.2, 42.4, 65.6, 99.2, 86.3, 46.8, 67.2, 76.2, 19.4, 18.4), I = c(76.7, 12.9, 21.3, 35.5, 43.5, 18.1, 59.2, 74.1, 28.9, 96.5, 80.8, 43.6, 51.6, 28.4, 80.7, 60, 82.4, 61.5, 78.5, 67.8, 63.9, 83.6, 41.8, 47.7, 94.4, 88.9, 88.9, 100), J = c(880.3, 476.8, 686.9, 300.1, 993.4, 358, 576.5, 299.7, 494.1, 762.2, 396.7, 201.7, 416, 584.1, 699, 265.3, 729.1, 181.5, 347.8, 303.1, 967.4, 817.9, 579.1, 788.5, 482.7, 551.9, 355.9, 676.7)), class = "data.frame", row.names = c(NA, -28L))
plotComb <- combn(colnames(data), 2, simplify = FALSE)
scatterfn <- function(l, data=data){
  ggscatter(data, x=l[1], y=l[2], add="reg.line", conf.int=TRUE, cor.coef=TRUE, cor.method="pearson", xlab=l[1], ylab=l[2])
}
plots <- lapply(plotComb, scatterfn, data=data)
plots <- split(plots, ceiling(seq_along(plots)/16))
pdf(file="Rplot1.pdf", width=12, height=12)
lapply(plots, function(x) plot_grid(plotlist=x, cols=4))
dev.off()

^{Создано в 2020-04-18 представителем пакета (v0.3.0)}