define target variable in glm r

问题

I want to be able to define my target variable 'def_target' outside the regression code below:

model1 <- glm(def_target~., family=binomial(link='logit'),data=train70)

I have tried the following but get an error pop up

tv1 <- 'def_target' 
model1 <- glm(tv1~., family=binomial(link='logit'),data=train70)

If anyone could help me that would be great.

Thanks

回答1:

I think it might tip up cause you tell it in your model1: glm(tv1~., family=binomial(link='logit'),data=train70). train70 does not have a matching column. Try to assign your variable directly, i.e.

tv1 <- train70[['def_target']]

then

model1 <- glm(tv1~., family=binomial(link='logit'))

I must admit I am not familiar with the "." you have in there. But make sure to relate that to your original dataset like you did with tv1.

回答2:

glm() requires the first argument to be of class "forumla", and simply inserting a string (i.e., 'def_target') will not parse correctly. You need to turn the argument into a formula by using as.formula(), but the entire formula you want to use must be included. Here's code that works:

model1 <- glm(as.formula(paste(tv1," ~ .")), family=binomial(link='logit'), data=train70)

来源：https://stackoverflow.com/questions/37364122/define-target-variable-in-glm-r