Relative error, absolute error, percentage error

[Nikoleto]

Hi all..I have a question to the visitors of that sweet site...
it is known that the relative error=absolute error/exact value
and so my question is if that exact value is equal to 0...is it possible to compute the relative error?
Thanks all!!!

[krtxmrtz]

Hi all..I have a question to the visitors of that sweet site...
it is known that the relative error=absolute error/exact value
and so my question is if that exact value is equal to 0...is it possible to compute the relative error?
Thanks all!!!

Simply put, the relative error tends to infinity but it's meaningless or rather useless.

It's a matter of interpretation: if you are computing the distance between 2 places, whcih is 100 km and your measurement yields 101 km, then you're off by 1%, relative error = (101-100)/100 = 0.01

If both places are the same so that the distance is 0, then any measurement producing a result other than 0 will be infinitely inaccurate.

[Nikoleto]

my question is what if my measurement yields 0 km not 101....shall i really put the relative error to tend to infinity?

[krtxmrtz]

my question is what if my measurement yields 0 km not 101....shall i really put the relative error to tend to infinity?

If your measurement is 0 over an actual distance of 100 km, then you're off by 100% (absolute error 100 km). If you measure 0 when the actual value is 0, then, again, the relative error has no meaning, it's computed as 0/0 which can be anything: 0/0 = x and this is satisfied by all real numbers x.

[System_Error]

This is a problem with relative error, but the likelyhood of encountering a situation where the exact value is zero, is almost zero...