Peter, this is extremely helpful. It is my first time using model.matrix()
and I was not very clear about it.Thank you for clarifying it!
It also helped me figure out how to do out-of-sample prediction. I add it
here in case someone else needs help with this.
oos.df<-data.frame(x=c(0,1), z=factor
Right on! I was using the expression for binomial logit. Thank you so much!
On Fri, Nov 20, 2015 at 12:23 PM, peter dalgaard wrote:
>
> > On 20 Nov 2015, at 17:17 , Damir Cosic wrote:
> >
> >
> > To do the same with matrix multiplication, I use the expression
> 1/(1+exp(-xb)):
> >
> > head(1/(
> On 20 Nov 2015, at 17:17 , Damir Cosic wrote:
>
>
> To do the same with matrix multiplication, I use the expression
> 1/(1+exp(-xb)):
>
> head(1/(1+exp(-(cbind(c(1), mm) %*% coef(m)[2,]
>
> which should produce the third column above, but it doesn't:
>
I'm rusty on this, but I suspec
On 20 Nov 2015, at 10:07 , peter dalgaard wrote:
>>
>> On 20 Nov 2015, at 04:53 , Damir Cosic wrote:
>>
>> Hello,
>>
>> I am having problems with predict() after a multinomial logit regression by
>> multinom(). I generate a design matrix with model.matrix() and use it to
>> estimate the mode
> On 20 Nov 2015, at 04:53 , Damir Cosic wrote:
>
> Hello,
>
> I am having problems with predict() after a multinomial logit regression by
> multinom(). I generate a design matrix with model.matrix() and use it to
> estimate the model. Then, if I pass the entire design matrix to predict(),
> it
Hello,
I am having problems with predict() after a multinomial logit regression by
multinom(). I generate a design matrix with model.matrix() and use it to
estimate the model. Then, if I pass the entire design matrix to predict(),
it returns the same output as fitted(), which is expected. But if I
6 matches
Mail list logo