我可以在 r 中相互绘制子集的值吗？

Question

I have a data for a neurocognitive study. We measure outcome by three slightly different surveys with the same range of possible points a participant can obtain. I have my data in long format – ie I have three rows for every participant and variables points and outcome . Variable outcome indicates what type of the survey was used in a given row ( scd_gb , scd_rb or scd_ab ) for measuring points.

    id outcome points
    1  scd_gb   20
    1  scd_rb   15
    1  scd_ab   3
    2  scd_gb   6
    2  scd_rb   18
    2  scd_ab   15

I would like to create a scatter plot where I have scd_gb on the x axis and scd_gb & scd_rb on y axis, each with a different color.

So I have two questions: First, can I plot subsets against each other or have I transform the data into wide format? Second (in general), can I plot one variable against two others?

I tried the following code that returns an error.

    library(ggplot2)
    ggplot(SCD_long , aes(x = points(subset(SCD_long, outcome %in% c("scd_gb"))), 
    y = points(subset(SCD_long, outcome %in% c("scd_rb" , "scd_ab"))))) +
        geom_point(aes(color = outcome), alpha = .5)   

    Error: Aesthetics must be either length 1 or the same as the data (606): colour, x, y
In addition: Warning messages:
1: In data.matrix(x) : NAs introduced by coercion
2: In data.matrix(x) : NAs introduced by coercion
3: In data.matrix(x) : NAs introduced by coercion
4: In data.matrix(x) : NAs introduced by coercion

I think that both questions can be solved by data wrangling. I wonder if I can receive the required plot without changing the format of my data.

Answer 1

Yes, I think making your data wide would be a good approach. Here's one way to do that, which I've applied to a long format of a subset of the iris data frame.

like_your_data <- structure(list(points = c(5.1, 4.9, 4.7, 4.6, 7, 6.4, 6.9, 5.5, 
6.3, 5.8, 7.1, 6.3), outcome = structure(c(1L, 1L, 1L, 1L, 2L, 
2L, 2L, 2L, 3L, 3L, 3L, 3L), .Label = c("setosa", "versicolor", 
"virginica"), class = "factor"), participant = c(1L, 2L, 3L, 
4L, 1L, 2L, 3L, 4L, 1L, 2L, 3L, 4L)), class = c("tbl_df", "tbl", 
"data.frame"), row.names = c(NA, -12L), .Names = c("points", 
"outcome", "participant"))

First I make a version that is just setosa (equivalent to your scd_gb ). Then I join that to a version that excludes setosa. This has the effect of adding the values for the others in one column, and their survey type in another. This will work well with ggplot, as we can map the survey type to color.

like_your_data %>%
  filter(outcome == "setosa") %>% # Equiv to scd_gb
  left_join(like_your_data %>% 
              filter(outcome != "setosa"), by = "participant") %>%
  ## Output at this point:
  # A tibble: 8 x 5
  #   points.x outcome.x participant points.y outcome.y 
  # <dbl> <fct>           <int>    <dbl> <fct>     
  # 1      5.1 setosa              1      7   versicolor
  # 2      5.1 setosa              1      6.3 virginica 
  # 3      4.9 setosa              2      6.4 versicolor
  ggplot(aes(points.x, points.y, color = outcome.y)) + geom_point()