包括NA的气泡图的频率

Question

我有一个看起来像这样的数据框：

Data<- data.frame(item1=c(1, 2, 3, 4, 5, 1, 2, 3, 4, 5, NA, 5, NA, NA), 
                  item2=c(1, 2, 2, 4, 1, 1, 2, 3, 5, 5, NA, NA, NA, NA),
                  item3=c(1, 2, 2, 4, 1, 1, 2, 3, 5, 5, NA, NA, NA, NA),
                  item4=c(1, 2, 2, 4, 1, 1, 4, 3, 1, 5, NA, 3, NA, NA),
                  item5=c(1, 5, 2, 4, 2, 1, 2, 3, 5, 5, NA, NA, 1, NA))

我已经定义了一个函数，它提取列频率并在没有NA的情况下绘制它

frequencies <- function(x,K=5)
{
  p <- length(x) # items
  n <- nrow(x) # observations
  r <- (5, NA) # values
  myf <- function(y) # extract frequencies
  {
    y <- y[!is.na(y)]
    y <- as.factor(y)
    aux <- summary(y)
    res <- rep(0, r)
    res[1:r %in% names(aux)] <- aux
    100 * res / sum(res)
  }

  freqs <- apply(x, 2, FUN = myf) #  apply myf by columns
  df2 <- expand.grid(vals = 1:r, item = 1:p) # all possible combinations
  df2$freq <- as.numeric(freqs) # add frequencies

  # graph
  plot(df2$item,df2$vals,type="n",xlim=c(1,p),ylim=c(1,r),xaxt = "n",
       xlab="", ylab="", ann=FALSE)


  axis(1, labels=FALSE)
  labs <- paste(names(x))   ##labels=c("v1", "v2", ...)
  text(1:p,  srt = 60, adj=0.5, pos=1, las=2,
       labels = labs, xpd = TRUE, par("usr")[1], cex.main=0.8, offset=1)



  points(df2$item,df2$vals,pch=22,col="black", bg="gray", cex=(df2$freq/n)*K)
}

我希望NA被标记为＆＃34;值＆＃34; （在y坐标系列中）所以我的情节看起来与那个情节相似（用编辑器编辑，没有R）：

提前谢谢你，

Angulo

Answer 1

另一种可能是您将数据melt用于长格式，然后使用table计算exclude = NULL以计算NA。如果要使频率与面积成比例而不是正方形的宽度，请检查scale_size_area。

library(reshape2)
library(ggplot2)

Data2 <- melt(Data)
Data3 <- with(Data2, as.data.frame(table(variable, value, exclude = NULL)))
Data3 <- Data3[!is.na(Data3$variable), ]

ggplot(data = Data3, aes(x = variable, y = value, size = Freq)) +
  geom_point(shape = 0)

Answer 2

尝试这样的事情：

#u Useful packages:
library(plyr)
library(ggplot2)

# Loop over variables getting the counts of each value
counts <- lapply(Data, count)

# Combine the list of counts into a single data frame
all_counts <- do.call(rbind, counts)

# A bit of fixing.  Make x into a factor, and get the variable name
all_counts <- within(
  all_counts,
  {
    Value <- factor(x)
    Variable <- rep(names(counts), vapply(counts, nrow, integer(1)))
  }
)

# Remove NAs (it isn't very clear from the question whether you want NAs or not)
all_counts <- subset(all_counts, !is.na(x))

# Draw the plot.  sqrt is to scale area by freq rather than width by freq
(p <- ggplot(all_counts, aes(var, x, size = sqrt(freq))) + 
  geom_point(shape = 15) # shape 15 is a square. See ?points.
)