在GGPlot上绘制GLM回归方程和Rsquared

Question

I have a problem like this:

Sample data:

library(tidyverse)
library(ggpubr)
a <- mtcars
reorganized <- a %>% gather (-mpg, key = "var", value = "value")

g <- ggplot(reorganized, aes(y = value, x = var)) +
  stat_smooth(method="glm", se=TRUE, fill=NA, 
              method.args = list(family = "binomial"), fullrange = F) +
  geom_smooth(method="glm", fill='red',
              method.args = list(family = "binomial")) +
  stat_regline_equation(
    aes(label =  paste(..eq.label.., ..adj.rr.label.., sep = "~~~~"))
  )+
  geom_point(aes(), alpha=2/10, shape=21, fill="blue", colour="black", size=0.2) +
  facet_wrap(~var, nrow=1)+
  theme_bw()

The data is the reorganized dataframe, with 3 columns: mpg, var, value

Using this code, I want plot facet with logistic regression line, equation with scatter plot. However, the formula appears incorrect, they only has the form of y = b + ax , and sometimes the Radj is out of the plot

How can I draw the correct formula of logistic regression in this figure?

Answer 1

It seems you are trying to use the glm method for non-categorical variables. This won't work. Instead, you should use either "loess" or "lm" methods. Also, to insert correctly the equation inside each facet you can control the position and the size of the equation. The code would be the following (geom_smooth defaults to "loess"):

ggplot(reorganized, aes(y = value, x = mpg)) +
  geom_smooth(se=TRUE, fill=NA, method.args=list(family="gaussian"), fullrange = F) +
  stat_regline_equation( aes(label =  paste(..eq.label.., ..adj.rr.label.., 
                                             sep = "~~~~")),
                         label.x.npc = "left", label.y.npc = "top", size = 2)+
  geom_point(aes(), alpha=2/10, shape=21, fill="blue", colour="black", size=0.2) +
  facet_wrap(~var, nrow=2)+
  theme_bw()

It should look like this

By default, geom_smooth is set to the loess method with the formula y ~ x. That is why you only have y and x on the equations. If you want a different formula you should look for how to customize formulas in geom_smooth.