几个简单的线性回归图

Question

I use pima dataset to visualise the linear relationship between the predictor variable triceps in the x axis, and each of the three response variables: pregnant , glucose , pressure in the y axis in separate plots.

Here is my code:

library(pdp)
library(ggplot2)

pima=na.omit(pima)
head(pima)

lin1=lm(triceps~pregnant, data=pima)
coef=coef(lin1)

#plot1
p1=ggplot(data = pima, aes(pima$pregnant, pima$triceps))+
  geom_point()+
  geom_abline(col="blue",intercept = coef[1], slope = coef[2], size=1)+
  labs(title="Linear Regression", x="Pregnant", y="Triceps")+
  theme(plot.title = element_text(hjust = 0.5))

lin2=lm(triceps~glucose, data=pima)
coef=coef(lin2)

#plot2
p2=ggplot(data = pima, aes(pima$glucose, pima$triceps))+
  geom_point()+
  geom_abline(col="blue",intercept = coef[1], slope = coef[2], size=1)+
  labs(title="Linear Regression", x="Glucose", y="Triceps")+
  theme(plot.title = element_text(hjust = 0.5))

lin3=lm(triceps~pressure, data=pima)
coef=coef(lin3)

#plot3
p3=ggplot(data = pima, aes(pima$pressure, pima$triceps))+
  geom_point()+
  geom_abline(col="blue",intercept = coef[1], slope = coef[2], size=1)+
  labs(title="Linear Regression", x="Pressure", y="Triceps")+
  theme(plot.title = element_text(hjust = 0.5))

grid.arrange(p1, p2, p3, ncol=2)

Unfortunately the code is repeated three times with different features. If I want to plot all features against x ( triceps ) I would have to copy and paste all the time the same code and just change the features. Is there any simpler way to do this?

Here are the plots:

Answer 1

You sure can. Use facet_wrap . The tricky part is to get the data you want from wide to long format. For your case, I used gather .

library(pdp)
library(ggplot2)
library(tidyr)

pima$id <- 1:nrow(pima)

xy <- pima[, c("triceps", "pregnant", "glucose", "pressure", "id")]

xy <- gather(xy, key = state, value = value, glucose, pregnant, pressure, -id, -triceps)

ggplot(xy, aes(x = value, y = triceps)) +
  theme_bw() +
  geom_point() +
  geom_smooth(method = "lm", se = FALSE) +
  facet_wrap(~ state, scales = "free_x", ncol = 2)

Answer 2

I think you mix up between variables. In the content you use triceps in the x axis, but in your plots, you put it in the y axis. If you would like to plot triceps in the x axis, and pregnant, glucose, pressure in the y axis, this code will help you:

library(pdp)
library(ggplot2)
pima=na.omit(pima)
head(pima)

pima %>%
 select(pregnant, glucose, pressure, triceps) %>%
 melt(id.vars = "triceps") %>%
 ggplot(aes(x = triceps, y = value)) + geom_point() + 
 geom_smooth(method = "lm") +
 facet_grid(~variable, scales = "free_x")