n*********e 发帖数: 318 | 1 I am trying to fit two curves to my data below -
Input csv data:
AGE, GROUP_A_Rate, GROUP_B_Rate
18,0.4,0.25
22,0.35,0.15
23,0.22,0.12
25,0.12,0.09
26,0.07,0.03
27,0.05,0.01
30,0.02,0.005
32,0.01,0.002
…
Explanation of data:
We are sending direct mails to people at different age levels. People are
also divided into two groups - A and B.
Then we recorded response rate in each group at each age level.
--------------------------------------------------------
I have fitted two curves to data points above.
But the problem is – I am treating each data point equal while fitting
curve. So I am ignoring "weights" of each data point, which could be defined
as how many subjects within that data point or how many responses took
place within that data point.
So how should I approach this issue? Should I say that within each data
point, there might be a distribution of response behavior (?) shall I give
a prior distribution to within each data point? How shall I take into
account the dispersion within each data point?
Thanks for your help in advance! | s******h 发帖数: 539 | 2 This is an interesting question. Can you be more specific on "treating each
data point equal while fitting curve", and would the curve fitting (btw,
what do you mean by curve fitting, regression or curve smoothing?) go
through if you don't ignore the "weights" of each data point? |
|