Hi Shuo,
Following is my rudimentary comment until more qualified answer appear:
As Poisson is a probability distribution function, the Y range is [0, 1], and as it is for
the occurrence probability, it a discrete function which takes 0, 1, 2,... as probability
variable. In both ways, your data doesn't match. As Poisson has only one parameter "r"
which determines the shape of the curve. if your purpose to determine this r by fitting,
you probably need to force scaling in both directions. For example, you can modify Poisson to: y = u * nlf_Poisson(x*t, 0, r)
where u and t are scaling parameters in addition to r (and y offset was fixed to 0).
("nlf_" is a suffix to refer to a Origin's built-in function.) See my test run below.
Hope this suggests a few things useful.
--Hideo Fujii
OriginLab
