循环遍历 dataframe 名称并在 R 中绘图

Question

I have a dataframe of 12 variables and I'd like to plot exactly one variable against others using ggplot's geom_point() . Wouldn't want to do it manually so i need to loop through the variables making plots. For example, I have a df like this (simplified to 4 variables for readability):

> head(df)
letters    value1    value2    value3
A          1         0         10
B          3         1         9
C          6         0         8
D          76        0         7
E          13        1         6
F          58        1         5

And I'd like to produce two plots where value1 is plotted over value2 and value3 . I've tried this:

plts <- vector()
for (i in names(df)) {
  p <- ggplot(df, aes(x=value1, y=i, fill=letters)) + geom_point())
  plts <- append(plts, p)
}

but it treats the values 2 & 3 different than the value 1 and produces something like this (eg, value1 over value3): Plot of value1 over value3

What should be done to improve this and achieve the goal of having the plots like this:

ggplot(df, aes(x=value1, y=value3, fill=letters)) + geom_point()

Produced without a loop

Answer 1

I think using aes_string() instead of aes will give you what you want. Your problem is caused by the tidyverse's use of non-standard evaluation ( NSE ).

lapply(
  names(df),
  function(y) {
    df %>% ggplot() + geom_point(aes_string(x="value1", y=y, colour="letters"))
  }
)

giving, for example

You can customise the first argument to lapply to select the variables you need.

That said, I think it would be easier and more robust to reformat your data frame to a more helpful layout and then create your plots...

For example,

df %>% 
  pivot_longer(
    cols=c("value2", "value3"), 
    names_to="Variable", 
    values_to="y"
  ) %>% 
  ggplot() + 
  geom_point(aes(x=value1, y=y, colour=letters)) + 
  facet_grid(rows=vars(Variable))

Giving

By the way, using colour=letters is probably more informative than fill=letters when using geom_point .