Display data values on geom_line

Question

I have created a plot using ggplot and geom_line and would like to have the individual data values (numbers) displayed on the plot.

This is my code so far:

ggplot(errors_prob_3) +

geom_point(aes(x = ID, y = train_mean, colour = 'red')) +
geom_line(aes(x = ID, y = train_mean, colour = 'red')) +

geom_point(aes(x = ID, y = train_sd, colour = 'blue')) +
geom_line(aes(x = ID, y = train_sd, colour = 'blue')) +

geom_point(aes(x = ID, y = test_mean, colour = 'green')) +
geom_line(aes(x = ID, y = test_mean, colour = 'green')) +

geom_point(aes(x = ID, y = test_sd, colour = 'purple') )+
geom_line(aes(x = ID, y = test_sd, colour = 'purple')) +

labs(x = "Number of predictors", y = "Mean/sd value", 
    title = "Plot of the training and test cv error") +
scale_color_discrete(name = "Legend", labels = c("Train mean", "Train sd", "Test mean", "Test sd")) 

and I get an error:

Aesthetics must be either length 1 or the same as the data (6): label.

errors_prob_3 is a data frame of 6 obs and 5 variables.

My data is as follows:

structure(list(ID = 1:6, train_mean = c(9.62056569110574, 9.16711345988648, 7.6586555534977, 7.62016661214069, 7.56713995100505,7.30900730515059),
train_sd = c(0.986375934334889, 1.00056316593354, 0.710802251192404, 0.691415430116925, 0.687478712185106, 0.726304511632279), 
test_mean = c(10.3738135132565, 10.0369765520266, 8.14168965211194, 8.40469077569391, 8.23953878513202, 8.0334726272832), 
test_sd = c(3.78543823754458, 3.6662904556102, 2.89196938350374, 2.74691446446284, 2.76568576521599, 2.98708614111826)), 
row.names = c(NA, -6L), class = "data.frame")

my current plot

any help would be appreciated. thank you!

Solution

The main trick is to reshape your data from wide to long format. Then everything becomes much simpler and even intuitive.

library(tidyverse)
library(ggplot2)

errors_prob_3 %>%
  gather(key = "variable", value = "value", -ID) %>%
  ggplot(aes(x = ID, y = value, colour = variable)) +
  geom_line() +
  geom_point() +
  geom_text(aes(label = round(value, 1)),
            vjust = "inward", hjust = "inward",
            show.legend = FALSE) +
  labs(x = "Number of predictors", y = "Mean/sd value", 
       title = "Plot of the training and test cv error") +
  scale_color_discrete(name = "Legend", labels = c("Train mean", "Train sd", "Test mean", "Test sd"))

---

Edit

To display only the first or the last text labels, subset the data. In this case I will display the last label per line.

errors_prob_3 %>%
  gather(key = "variable", value = "value", -ID) %>%
  ggplot(aes(x = ID, y = value, colour = variable)) +
  geom_line() +
  geom_point() +
  geom_text(
    data = . %>% filter(ID == max(ID)),
    aes(label = round(value, 1)),
    vjust = "inward", hjust = "inward",
    show.legend = FALSE) +
  labs(x = "Number of predictors", y = "Mean/sd value", 
       title = "Plot of the training and test cv error") +
  scale_color_discrete(name = "Legend", labels = c("Train mean", "Train sd", "Test mean", "Test sd"))

Comments

[Rui Barradas]

Can you post sample data? Please edit the question with the output of dput(errors_prob_3). Or, if it is too big with the output of dput(head(errors_prob_3, 20)). And take a look at geom_point.

[MY.BK]

i edited the question by adding my data set and adding geom_point

[Rui Barradas]

1) Please post data in dput format, we cannot copy&paste images. 2) Aren't the individual data points now displayed with geom_point? I am no longer understanding the question, could you please make it more clear? 3) There are simpler ways of plotting that graph, but I would really need the data.

[MY.BK]

Hi there, i have included my data in dput format. to answer 2), i am looking to have the individual data values (numbers) displayed beside each individual plotted point. I apologise for the confusion

[Bob]

How may I play with your code to just add the label(s) for the first and/or the last value to the graph? EDIT: I have done it by creating empty values for the first observations.

[Bob]

Thanks! Much more elegant than my attempt

[Rui Barradas]

@Bob No, wait! It's better now, there's no need to create an explicit long format df.