在 R 中的预测 glm model 中使用指数分布

Question

I am using a glm model to predict my depending variable. For that I need to choose a family of the distribution of my variable. unfortunately the exponential distribution is not part of the available objects of the argument "family". Now I don't know on how I can proceed with my research. This is my model. Does anyone have an idea, what I can do?

model<-train(duration~., data = data, method='glm', family= ???, trControl =trainControl(method = "repeatedcv", repeats = 10)

Answer 1

The exponential distribution is in the gamma family with dispersion parameter fixed at 1, so family = Gamma(link="log") should provide you with what you need. When interpreting significance or standard error of the fitted coefficients assuming exponential, you specify the dispersion.

Since your example wasn't reproducible, an example using glm and summary is:

mdl <- glm(formula = duration ~.,  family = Gamma(link="log"), data = your_data)

summary(mdl, dispersion = 1)