CODEWITHSUNDEEP
rggplot2mean
YnkDK
YnkDK

Reputation: 741

Values of stat summary fun mean

I am trying to create a plot to compare running time of different algorithms. By running the below R code, I get the following plot, which I am generally statisfied with. However: It can be hard to read-off values from this graph, is there a way to get the plotted mean valeus for each DBMS for each instance? For example for gplus-combined, the value of CacheDBMS is around 50, while for BranchDBMS it is around 200.

ggplot(dt, aes(reorder(instance, V9), V9)) + 
  geom_point(aes(group=V2, colour=V2), stat='summary', fun.y='mean') +
  geom_line(aes(group=V2, colour=V2), stat='summary', fun.y='mean') +
  scale_y_log10() +
  ylab("Mean wall time") +
  xlab("") +
  ggtitle("Comparison of Database Management Systems") +
  theme_bw() +
  theme(axis.text.x = element_text(angle=45, vjust = 1, hjust = 1)) +
  guides(color=guide_legend(title="DBMS"))

Output

I want the y-values for each point. Preferably as a table, e.g.

BranchDBMS    gplus-combined    213.21
CacheDBMS     gplus-combined     48.68

EDIT

A small snippet (out of 10000-ish lines) of input data. I have removed unused columns, so the V* is not correct. But V2 is the first column here, V9 is the second and instance is the last.

BranchDBMS;      0.163352;  facebook-combined   
BranchDBMS;      0.169043;  facebook-combined   
BranchDBMS;      0.162545;  facebook-combined   
BranchDBMS;      0.159489;  facebook-combined   
BranchDBMS;      0.168414;  facebook-combined 
CacheDBMS ;      0.038515;  facebook-combined   
CacheDBMS ;      0.037179;  facebook-combined   
CacheDBMS ;      0.037385;  facebook-combined   
CacheDBMS ;      0.036514;  facebook-combined   
BranchDBMS;    281.149423;  gplus-combined    
BranchDBMS;    261.093502;  gplus-combined   
BranchDBMS;    258.816546;  gplus-combined     
CacheDBMS ;     22.442501;  gplus-combined    
CacheDBMS ;     22.377717;  gplus-combined   
CacheDBMS ;     22.469739;  gplus-combined   
CacheDBMS ;     22.451922;  gplus-combined

Upvotes: 1

Views: 2663

Answers (1)

eipi10
eipi10

Reputation: 93761

Here's an example of how to add the value labels directly to graph, using the built-in iris data frame:

p1 = ggplot(iris, aes(Sepal.Width, Sepal.Length, colour=Species)) +
  stat_summary(fun.y=mean, geom="line", alpha=0.5) +
  stat_summary(fun.y=mean, geom="text", aes(label=sprintf("%1.1f", ..y..)), 
               size=3, show.legend=FALSE) +
  guides(colour=guide_legend(override.aes = list(alpha=1, lwd=1)))

..y.. are the internally calculated means at each value of Sepal.Width for each Species. Because we used alpha=0.5 for the line geom, override.aes allows us to have bolder lines in the legend.

enter image description here

One way to add a table of data values would be as follows:

library(gridExtra)
library(dplyr)

# Change default fontsize for the data table
mytheme <- ttheme_default(
  core = list(fg_params=list(cex = 0.7)),
  colhead = list(fg_params=list(cex = 0.75)),
  rowhead = list(fg_params=list(cex = 0.75)))

# Create table (in this case I just show the first three values for each species)
tab = tableGrob(iris %>% group_by(Species, Sepal.Width) %>% 
             summarise(`Mean Sepal Length`=sprintf("%1.1f", mean(Sepal.Length))) %>%
               slice(1:3), theme=mytheme, rows=NULL)

# Lay out graph and table
grid.arrange(p1, tab, ncol=1)

enter image description here

Upvotes: 6

Related Questions