0
$\begingroup$

How can I write a prediction formula if I can't use the glm's predict function ? If I have a model that uses the family=binomial(link="logic") I would write the model like this: model_1<-glm(IsUp ~ var1+var2,data-train ,family=binomial(link="logic"))

Then I would run the model summary, get the coefficients, including the intercept, and create the prediction formula:for example:

intercept = 0.78, var1 coeff is 0.28 and var2 coeff is 0.5

train$glm_pred1<-1/(1+exp(-(0.78+(train$var1* 0.28 + train$var2 *0.5))))

But how can I create a prediction function when my model is based on Gaussian like in this model:model_2<-glm(IsUp ~ var1+var2,data=train)

The summary of the model indicates that the intercept is 0.66, var1 coeff is 0.3 and var2 coeff is 0.8?

Improve this question
$\endgroup$
2
  • 2
    $\begingroup$ If I understand you correctly, the answer you seek is just 0.66 + 0.3 var1 + 0.8 var2. i.e. like your other example, but without inverting the link. It's odd that the question is phrased in terms of R syntax, because no R issues are involved here and because the model formula is usually explained at the outset before there is any question of using software. $\endgroup$
    – Nick Cox
    Nov 15, 2015 at 14:41
  • $\begingroup$ The title and content don't match. You are not asking about how to write a logistic regression model down. You already know that. $\endgroup$
    – Nick Cox
    Nov 15, 2015 at 14:43

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.