Multiple grouping variables in data frame in ggplot

Question

I have a data set of transactions regarding resale flats.

I used the pipeline function to group and summarise the data based on the flat_type and region

plotdata1<-data1 %>% 
  group_by(year, region,flat_type) %>% 
  summarize(mean_price = mean(resale_price))

Here's my data set

 year region  flat_type mean_price
  <int> <fct>   <fct>          <dbl>
1  2007 Central 3 ROOM       236452.
2  2007 Central 4 ROOM       367471.
3  2007 Central 5 ROOM       467264.
4  2007 East    3 ROOM       198682.
5  2007 East    4 ROOM       266645.
6  2007 East    5 ROOM       323110.

However, this does not allow me to plot the line graph correctly as there are 2 grouping variables. I intend to plot the average resale prices of flats based on their region including all 3 flat-types. Using shiny I will then be able to use the selectInputfunction to select between all flat-types or each individual flat-type. For example, the graph will be able to show the average resale price of 3 room flats in each region.

Here the code for the plot

ggplot(data=plotdata1,aes(x=year,y=mean_price))+
  geom_line(stat = 'identity',aes(colour=region,group=region))+
  geom_point()+
  xlim(c(2006,2018))+
  ylab("Average Price")+
  xlab('Year')

Which gives me the error

Error: Column region can't be modified because it's a grouping variable

Doing this instead works and plots them based on the region but then I will not be able to select between each flat-type from the input boxes.

plotdata1<-data1 %>% 
  group_by(year, region) %>% 
  summarize(mean_price = mean(resale_price))

Here's the intended look for the plot which works when using the pipeline code above this . But then again I will not be able to select between either all flat-types or each flat-type

kstew kstew · Accepted Answer · 2019-07-17T22:50:17

Hmm not sure why you are getting that error. Can you try the call putting aes() inside the ggplot() call instead of the geom? This works for me...

set.seed(123)
df <- data.frame(region=sample(letters[1:4],100,T),
                 type=sample(c('1 room','2 room','3 room'),100,T),
                 year=sample(2010:2015,100,T),
                 price=sample(1000:2000,100,T))

df %>% group_by(region,type,year) %>% 
  summarise(mean.price=mean(price)) %>% 
  ggplot(aes(year,mean.price,group=region,color=region)) + geom_point() + geom_line() +
  facet_wrap(~type,scales='free')

If you are using shiny, then you can remove the facet_wrap and add filter() before the ggplot call so that users can specify which type of room prices to plot.

Multiple grouping variables in data frame in ggplot

1 Answers