Hi,
You should use error propagation formula to calculate the error of the transformed y:
f(y)=ln(y)
df/dy = (d/dy)*ln(y)*delta_y
= (1/y)*delta_y
So if your original y and delta_y are in cols B and C, add two new columns D and E, and set them as follows:
col(d)=ln(col(b));
col(e)=(1/col(b))*col(c);
To verify this, you can transform backwards:
g(y)=exp(y)
dg/dy = (d/dy)exp(y) * delta_y
= exp(y) * delta_y
Add two more columns F and G, and set them as below:
col(f)=exp(col(d));
col(g)=exp(col(d)) * col(e);
and this should give you back your original y and dy
Easwar
OriginLab
