How to deal with frequency weight??

[bokchy]

Hi,

I'm trying to run simple OLS, Y = b0 + b1*X.
And what I want to ask is about treating input data with "frequency" weight.

My data is as follows.

x y _freq_
1 10 1
5 20 3
2 35 2
3 15 1


Above data seems to have 4 observations.
However it has actually 7 observations, considering the frequency.

...like follows.

x y _freq_
1 10
5 20
5 20
5 20
2 35
2 35
3 15

So, is there any option to use under the 'frequency' weight ?


Thanks.

[EViews Glenn]

Least squares doesn't support frequency weights, but GLM does. You can estimate an equivalent model using GLM and the default normal family and identity link .

equation eq1.glm(fwgt=_freq_) y x c