1

I had following 

     RMSE      Model fold
1  0.4164260    lm    1
2  0.5412474    lm    2
3  0.3496156    lm    3
4  0.4431751    lm    4
5  0.4082882    lm    5
6  0.4398985    lm    6
7  0.4506528    lm    7
8  0.4236024    lm    8
9  0.4346417    lm    9
10 0.4982427    lm   10
11 0.3993842  MARS    1
12 0.5684007  MARS    2
13 0.3487462  MARS    3
14 0.4174607  MARS    4
15 0.3930033  MARS    5
16 0.4456201  MARS    6
17 0.4373339  MARS    7
18 0.4110089  MARS    8
19 0.4249311  MARS    9
20 0.4295155  MARS   10

I am plotting as follows: 

ggplot(df, aes(x=factor(fold), y=RMSE, group=Model, colour=Model)) + geom_line()

I want to do mean line for both models. How can I achieve that?

add-semi-colons
  • 18,094
  • 55
  • 145
  • 232

1 Answers1

4

One way is to create a separate data frame containing the means:

library(ggplot2)
library(plyr)
df_means <- ddply(df, "Model", summarise, mean_RMSE = mean(RMSE))
#   Model mean_RMSE
# 1    lm 0.4405790
# 2  MARS 0.4275405

ggplot(df, aes(x=factor(fold), y=RMSE, group=Model, colour=Model)) + 
  geom_line() +
  geom_hline(data=df_means, aes(yintercept=mean_RMSE, color=Model), 
             linetype="dashed")

Plot

You can also use: geom_line() with a "stat" to plot means by the group aesthetic (albeit, you will have less control):

ggplot(df, aes(x=factor(fold), y=RMSE, group=Model, colour=Model)) + 
  geom_line() +
  geom_line(stat="hline", yintercept="mean", linetype="dashed")

This keeps the line from "extending":

Plot 2

JasonAizkalns
  • 20,243
  • 8
  • 57
  • 116
  • I added a smooth and it created a really strange plot is there away to add smoothing to multiple plots? – add-semi-colons Apr 06 '15 at 15:26
  • @Null-Hypothesis Try to keep each Q&A isolated to one issue/question. If you have a "new" question, just post it with the problem. That having been said, there's probably nothing wrong with the smoother - it's likely smoothing by each group and has very wide confidence intervals which will make the plot look "strange". – JasonAizkalns Apr 06 '15 at 15:27