Pandas pd.eval for 64.1*1e3

Asked Mar 04 '20 at 22:46

Active Mar 04 '20 at 23:50

Viewed 27 times

I'm currently scrubbing some numbers and converting numbers if they are 66.1K to be 66100 using

df['number'].replace({'[kK]': '*1e3', '[mM]': '*1e6', '[bB]': '*1e9', },regex=True).map(pd.eval)

can replicate by just doing a simple print(pd.eval('64.1*1e3'))

However the return for 64.1K is 64099.99999999999.

Does anyone know why this happens? How would you use pandas to get 64100.0?

I have tested with 65.1K and it is the same result 65099.99999999999 but if I do 66.1K it equals to 66100.0.

edited Mar 04 '20 at 23:50

asked Mar 04 '20 at 22:46

I eat toast

well kinda but I know I can get it to 64100 by just using the round function. I'm really more curious as to why Pandas is behaving in the way it is. – I eat toast Mar 04 '20 at 23:30
Because of ```float``` representation in python – Grzegorz Skibinski Mar 04 '20 at 23:32
Could you expand on that? I don't really get it. – I eat toast Mar 05 '20 at 15:23
ah - this has an straight forward answer. https://floating-point-gui.de/ but thanks Grzegorz Skibinski – I eat toast Mar 05 '20 at 20:24

0 Answers0