j*******g
发帖数: 2140 | 1 I have two quantities, each of which has an estimated standard error. What
is the standard error of the ratio of these two quantities?
有人给出这两个答案:
1)To find an error, you need to see that is the maximum deviation from the
base value what both x and y change:
d(x/y) = dx*1/y - dy*x/y^2 = (dx/x)*(x/y) - (dy/y)*(x/y)
Note that dx/x and dy/y can be treated as the errors of your quantities.
Also, the maximum total error is a square root of the individual terms in
the above euqation (imagine right tira | j*******g
发帖数: 2140 | 2 很难吗? 看到有人留言了,然后又删了.
【在 j*******g 的大作中提到】
: I have two quantities, each of which has an estimated standard error. What
: is the standard error of the ratio of these two quantities?
: 有人给出这两个答案:
: 1)To find an error, you need to see that is the maximum deviation from the
: base value what both x and y change:
: d(x/y) = dx*1/y - dy*x/y^2 = (dx/x)*(x/y) - (dy/y)*(x/y)
: Note that dx/x and dy/y can be treated as the errors of your quantities.
: Also, the maximum total error is a square root of the individual terms in
: the above euqation (imagine right tira
| x***x
发帖数: 3401 | 3 I dont think variance of the ratio of two RVs has simple formula as that,
unless your X and Y has some special distribution - I might be wrong.
The usual way to go is to use delta method to approximate.
I dont quite understand 1). what is that about? |
|
