07-30-2015 01:38 PM
I was trying proc sgplot to generate a box plot. It was aborted due to memory issue. The data set has about 1.7 million records, but there's only 16 distinct values for the x-axis. There are also some other procedures in SAS that can do box plot. I was wondering which procedure can handle big data sets better. I would like to add a limit that if the number of rows exceeds the limit it will not do the box plot. But I'm not sure how to calculate the limit based on the memory I have.
07-30-2015 01:59 PM
You should show the code used in case you specified something that might be an issue. If you requested plots for multiple variables that might cause some issue.
I just generated a random data set with 1.7 million records and with one analysis variable and 16 x values proc boxplot completed the graph in under 1 second. So the file size is unlikely to be the main issue.
07-30-2015 03:37 PM
Thanks ballardw for pointing me to the right direction. I think it's the format statement that's causing the problem. After I removed the format. It completed in seconds.