"Uniform error"...what does it mean?

**AxiomOfChoice** · May22-09, 11:10 AM

I'm reading a paper right now that talks about an error term being uniform under certain conditions. But what does it *mean* for an error term to be "uniform"? I have no idea.

Also, I recall having read some things about "uniform error estimates." Is this a similar notion?

**D H** · May22-09, 11:15 AM

It means the error is drawn from a uniform distribution.

**AxiomOfChoice** · May22-09, 11:28 AM

Originally Posted by D H

It means the error is drawn from a uniform distribution.

Thanks for your response, but that really means very little to me. I'm familiar with the uniform distribution, but I can't immediately make sense of "drawing error" from it. Do you mean that all errors (within some range) are equally probable?

**Bob S** · May22-09, 12:24 PM

Uniform distribution may mean:
http://en.wikipedia.org/wiki/Uniform...on_(continuous)

**Enuma_Elish** · May22-09, 03:35 PM

Originally Posted by AxiomOfChoice

Thanks for your response, but that really means very little to me. I'm familiar with the uniform distribution, but I can't immediately make sense of "drawing error" from it. Do you mean that all errors (within some range) are equally probable?

Surely that would be my interpretation.