Why do I get infs when I divide by 0 in pandas?

Question

All

I have two columns and I divide one by the other with colb being my denominator. When I divide one by the other however I get infs when there is 0 present in my colb e.g

df[cola]/df[colb]

Mathematically this is correct right? Just wondering if this is expected with pandas . Fir me it makes no sense to have these infs so I will probably replace with 0 but wanted to check

This is the expected result. `anything/0` is `inf`, except that `0/0` is `NaN`. — Barmar, Nov 07 '20 at 09:46
Does this answer your question? [How to deal with "divide by zero" with pandas dataframes when manipulating columns?](https://stackoverflow.com/questions/38886512/how-to-deal-with-divide-by-zero-with-pandas-dataframes-when-manipulating-colum) — Alex Shirley, Nov 07 '20 at 09:49
Actually anything divided by 0 is undefined mathematically, it is not infinite — Wasif, Nov 07 '20 at 09:53
`pandas` outputs `inf` because `positive/0` **approaches** `inf` — DeepSpace, Nov 07 '20 at 09:59

score 2 · Accepted Answer · edited Nov 07 '20 at 10:43

For x≠0 , and x∈R , the limits of x/0 as we approach from the left and right hand sides of 0 differ. Hence, it's not defined.

Consider approaching the limit from the right hand side. As the denominator gets smaller and smaller, the value will approach ±∞ depending upon the sign of x . If x>0 , the limit will approach ∞ and if x<0 , the limit approaches −∞ .

If we approach the limit from the left, if x>0 , the limit approaches −∞ and similarly ∞ when x<0 .

In both cases, there is a discontinuity when the denominator approaches 0 and hence x/0 is undefined.

Hence, even in pandas this should yield. However, it is a convention that x/0 is inf if x>0, -inf if x<0

Why do I get infs when I divide by 0 in pandas?

1 Answers1