Question about parameters in a logistic regression

juanvg1972 — Sat, 27 May 2017 16:23:23 GMT

Hi,

I have two questions about the procedure proc logistic:

1) When is recommended to use crossing effects?, I refer to the case of including the product of two input vars( var1*var2) in the model. Example:

proc logistic data=test outmodel=modelo2 plots(only)=roc;
class cod_posicion nivel_sal rango_edad rango_ant rango_eval;
model baja = rango_edad rango_ant nivel_sal rango_eval cod_posicion
			rango_edad*rango_ant rango_edad*nivel_sal rango_edad*rango_ant;
quit;

I think is recommended when there is a correlation between the varis, isn't it?

2) Haw can I do a regularizarion in logistoc regression usind proc logistic. I mean regularizaton methods like

Ridge, Lasso y Elastic Net.

Any advice will be greatly appreciated

Thanks in advance

Re: Question about parameters in a logistic regression

Ksharp — Sun, 28 May 2017 03:48:19 GMT

1) when the combination of these two variable are significant .

E.X.

sex=F is significant for Y

age=20 is significant for Y

sex=F and age=20 is also significant for Y

2) PROC HPGENSELECT

topic Question about parameters in a logistic regression in Statistical Procedures

Question about parameters in a logistic regression

Re: Question about parameters in a logistic regression