Size of Rdata file compared to csv

Question

The size of my .Rdata file is 92 MB.

However, the original csv-file is around 3 GB. I included it with the usual read.csv()

How can that be?

Answer 1

The comments already hinted at what is going on. But this is so straightforward, let us do an example:

R> X <- 1:1e5   # data, no repeats
R> save(X, file="/tmp/foo.RData")
R> write.csv(X, file="/tmp/foo.csv")
R> system("ls -l /tmp/foo*")
-rw-r--r-- 1 x y 1377797 Jun  4 09:11 /tmp/foo.csv
-rw-r--r-- 1 x y  212397 Jun  4 09:11 /tmp/foo.RData

Now with data that repeats:

R> X <- rep(1,1e5)   # data, lots of repeats
R> write.csv(X, file="/tmp/bar.csv")
R> save(X, file="/tmp/bar.RData")
R> system("ls -lh /tmp/bar*")
-rw-r--r-- 1 x y 966K Jun  4 09:12 /tmp/bar.csv
-rw-r--r-- 1 x y 1.3K Jun  4 09:12 /tmp/bar.RData
R> 

So we are getting ratios of 6.5 to 743 depending on how well this compresses. And that is before we make the csv more "expensive" by forcing several decimals to be printed...