It would be interesting to know if they have any particular justification for picking the natural logarithm or if it was simply a matter of using what first came to mind. Other likely candidates seem to be:
Yeah, as a software engineer, I can estimate log base 2 in my head, so that would be much more natural for me. But I could also see an argument for decibels since it’s a unit that’s already in use for other things.
Helpfully, log scales are off from each other by a constant factor, so whatever techniques they use ln for can be used just as well for log base 2. Just update any constants to the correct (much prettier) values.
Interesting Log-Odds paper by Brian Lee and Jacob Sanders, November 2011.
It would be interesting to know if they have any particular justification for picking the natural logarithm or if it was simply a matter of using what first came to mind. Other likely candidates seem to be:
log base 2 (a.k.a. bits)
10 * log base 10 (a.k.a. decibels)
log base 10
Roll on the log odds standardisation effort—by the sound of it!
Yeah, as a software engineer, I can estimate log base 2 in my head, so that would be much more natural for me. But I could also see an argument for decibels since it’s a unit that’s already in use for other things.
Helpfully, log scales are off from each other by a constant factor, so whatever techniques they use ln for can be used just as well for log base 2. Just update any constants to the correct (much prettier) values.
Much like how inches and centimeters are off by a constant factor. Different log bases are analogous to different units.