[R] qqnorm & huge datasets

[Marc Schwartz]

On Dec 21, 2011, at 4:10 PM, Sam Steingold wrote:

> Hi,
> When qqnorm on a vector of length 10M+ I get a huge pdf file which
> cannot be loaded by acroread or evince.
> Any suggestions? (apart from sampling the data).
> Thanks.


Depending upon what your end product needs to be, generate a png file instead of pdf. With a pdf, every data point consumes 'space' in the file, therefore increasing the file size. Two data points that exactly overlap each other consume twice the space of a single data point in the pdf. 

That is not the case with a png file which only has to store information on each pixel in the image.

The key with the png is to create the file initially so that the resolution and size conform to your final product requirement without scaling, which tends to reduce the image quality.

HTH,

Marc Schwartz