How to do it...

1. Load ggplot2 and call for the geom_smooth() layer:

> library(ggplot2)
> scatter <- ggplot(data = iris, 
                    aes(x = Petal.Length, y = Petal.Width)) + 
    geom_point(alpha = .5, aes(colour = Species, shape = Species))
> scatter + 
    geom_smooth(method = 'lm', se = F, show.legend = F, 
                aes(group = Species))

Resulting illustration would look like the following figure 2.11:

Figure 2.11 - scatterplot with grouped regression lines.

1. For ggvis, call layer_model_predictions() along with the group_by() function:

> library(ggvis)
> ggvis( iris, x = ~Petal.Length, y = ~Petal.Width, opacity := .5) %>%
>  layer_points( shape = ~Species, fill = ~Species) %>% group_by(Species) %>%
>  layer_model_predictions( model = 'lm', stroke = ~Species)

1. Using plotly, combine add_lines() with the lm() function to achieve a similar result:

> library(plotly)
> plot_ly(iris, x = ~Petal.Length, y = ~Petal.Width, 
          showlegend = F, alpha = .5, color = ~Species) %>% 
    add_markers(showlegend = T, symbol = ~Species) %>%
    add_lines(data = iris %>% filter(Species == 'setosa'), 
              y = ~fitted(lm( Petal.Width ~ Petal.Length))) %>%
    add_lines(data = iris %>% filter(Species == 'versicolor'),
              y = ~fitted(lm( Petal.Width ~ Petal.Length))) %>%
    add_lines(data = iris %>% filter(Species == 'virginica'),
              y = ~fitted(lm( Petal.Width ~ Petal.Length)))

Following section holds some explanations.