Plot more than one regression line in a scatterplot using facet() and add the slope coefficient to every line

Question

I want to plot the regression lines for each city in a scatterplot.

The dataframe looks like:

     df
            City      str    testscr
19     Los Angeles 22.70402  619.80
31       San Diego 20.60697  624.55
33     Los Angeles 21.53581  625.30
35  San Bernardino 21.19407  626.10
36     Los Angeles 21.86535  626.80
45       Riverside 19.26697  628.75
46     Los Angeles 23.30189  629.80
63          Orange 21.94756  633.15
67     Los Angeles 20.68242  634.05
69       San Diego 21.78650  634.10
72     Los Angeles 21.15289  634.40
76  San Bernardino 18.98373  634.95
86  San Bernardino 19.30676  636.60
87       Riverside 20.89231  636.70
105 San Bernardino 19.75422  639.35
114         Orange 19.62662  640.75
118      San Diego 20.08452  641.45
126      Riverside 22.81818  643.20
128    Los Angeles 21.37363  643.40
146      San Diego 19.79654  645.55
156         Orange 21.04869  646.70
157         Orange 20.17544  646.90
160      San Diego 20.29137  647.25
168      San Diego 17.15328  648.70
169 San Bernardino 22.34977  648.95
170         Orange 22.17007  649.15
191         Orange 23.01438  652.10
200      Riverside 21.03721  653.40

My Approach was:

ggplot(data=df,aes(x=str,y=testscr))+
  geom_point()+
  geom_smooth(method="lm",se=FALSE)+
  facet_grid(. ~City)

Is there a smarter or better way? And how can I add the slope coefficient to every Regression line?

Answer 1

Let's deal with groups first, then answer the second part about adding labels.

If you want to plot by group, there are basically two options. The first is to facet, as you have. The second is to group the points, either explicitly using aes(group = City), or by another aesthetic such as aes(color = City).

If the second approach generates a messy plot, for example with lots of overlapping lines, then it's best to go with facets.

A couple of examples using the iris dataset.

First, grouping by color:

library(ggplot2)
iris %>% 
  ggplot(aes(Petal.Length, Sepal.Length)) + 
  geom_point(aes(color = Species)) + 
  geom_smooth(method = "lm", 
              aes(color = Species), 
              se = FALSE)

Group by group:

iris %>% 
  ggplot(aes(Petal.Length, Sepal.Length)) + 
  geom_point(aes(group = Species)) + 
  geom_smooth(method = "lm", 
              aes(color = Species), 
              se = FALSE)

Use facets:

iris %>% 
  ggplot(aes(Petal.Length, Sepal.Length)) + 
  geom_point() + 
  geom_smooth(method = "lm", 
              se = FALSE) +
  facet_wrap(~Species)

For adding labels such as the coefficients, look at the ggpmisc package. Here is one way to add the coefficients using stat_fit_tb:

iris %>% 
  ggplot(aes(Petal.Length, Sepal.Length)) + 
  geom_point() + 
  geom_smooth(method = "lm", 
              se = FALSE) + 
  facet_wrap(~Species) + 
  stat_fit_tb(method = "lm", 
              tb.type = "fit.coefs")