how to find the prediction cut off point in r -

i training glm model class attribute 'adverse_effects' factor containing 0 , 1

ctrl <- traincontrol(method = "cv", number = 5)  model_logreg <- train(adverse_effects ~.,family=binomial(link='logit'),data=trainsplit_logreg, method = "glm", trcontrol = ctrl)   predictors <- names(trainsplit_logreg)[names(trainsplit_logreg) != 'adverse_effects'] pred_logreg <- predict(model_logreg$finalmodel, testsplit_logreg[,predictors]) 

this summary of predictions

summary(pred_logreg)     min.  1st qu.   median     mean  3rd qu.     max.  -14.5600  -2.1220  -1.8700  -1.9890  -1.7090  -0.9459  

how know cut off point of prediction? how can map results of prediction 0s , 1s?

p.s got auc of 0.6144

predict.glm takes type argument determines scale on make predictions. think want type="response", try

predict(model_logreg$finalmodel, testsplit_logreg[,predictors], type="response")  

the natural cutoff 0.5, select outcome. unless have special situation (e.g. false negative worse false positive) stick 0.5 cutoff.