Is there a way to use a 32-bit float instead of a 64-bit in R-data frames?

Question

Memory issue. I do financial calculations on the amount in dollars, and currently my floating point accuracy is 64 bits. I would like to reduce the accuracy to 32 bits, but have not yet found a way to specify this in R. Ideally, this just applies to a data frame that has several columns, some of which are int and some are floats.

Answer 1

No, no, at least not in the "base R", which has only one type integer and numeric (floating point), and their sizes are fixed.

You can check them (and much more) through .Machine - see help(".Machine") .

Now, for your dollar amounts, you could certainly resort to cents, and in this case you could integer , which is usually half the size of numeric .