Create custom legend in ggplot2

Question

So, here is my issue. I have data similar to the minimal reproducible example below that I want to plot with ggplot2 , where a data frame contains one column that I want to use as the x-values for every geom and several columns that I want as different y-values. As you can see from the example below, I want to combine geom_line and geom_point with different shapes for the geom_point values.

My question now is, how do I insert a legend that tells me what columns which shape represents. From what I have learned about ggplot2 so far, a legend is usually generated, if I map some factor (might not be the correct term) to color = or group = , I am wrong? So how can I still get a legend without that prerequisite?

Help is much appreciated!

library(tidyverse)

df <- structure(list(rep = c(1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
1, 1), Y_1 = c(0.0198, 0.0108, 0, 0.0117, 0.00931, 0.0089, 0.0115, 
0.00509, 0.00831, 0.0158, 0.0437, 0.0953, 0.267, 0.677, 1.81), 
    Y_2 = c(0.025, 0.00249, 0.00303, 0.00268, 0.0102, 0.0112, 
    0.0231, 0.0326, 0.0575, 0.0852, 0.143, 0.219, 0.384, 0.687, 
    1.01), X = c(0.1, 0.164, 0.268, 0.439, 0.72, 1.18, 1.93, 
    3.16, 5.18, 8.48, 13.9, 22.8, 37.3, 61.1, 100)), row.names = c(NA, 
15L), class = "data.frame")

df_plot <- ggplot(data = df) +
  geom_line(mapping = aes(x = X, y = Y_1)) +
  geom_point(mapping = aes(x = X, y = Y_1), shape = 15) +
  geom_line(mapping = aes(x = X, y = Y_2)) +
  geom_point(mapping = aes(x = X, y = Y_2), shape = 0) +
  scale_x_log10() +
  scale_y_log10() +
  theme_classic()

df_plot

Answer 1

The way to create a legend is to map the different levels of a variable to an aesthetic scale (in your case, a shape scale). The most idiomatic way to do this in ggplot is reshape your data into long format, by making Y_1 and Y_2 into a single column of y values, with a new column that labels each y value according to the original column it came from. This means you only need a single call to geom_line and a single call to geom_point :

ggplot(data = tidyr::pivot_longer(df, c("Y_1", "Y_2"))) +
  geom_line(mapping = aes(x = X, y = value, group = name)) +
  geom_point(mapping = aes(x = X, y = value, shape = name)) +
  scale_shape_manual(values = c(0, 15)) +
  scale_x_log10() +
  scale_y_log10() +
  labs(shape = "variable") +
  theme_classic()