Chapter 7 Multinomial Regression

7.1 Fitting the Model

library(nnet) m1 <-multinom( y ~ x, data = )

• Replace y with your categorical response variable.
• Replace x with your explanatory variable.
• data = your data set

7.2 Other Useful Codes

summary(m1): To get the estimates and standard errors.

coefficients(m1): To get the estimates.

AIC(m1): To get the AIC.

7.3 Making Predictions

probabilities <- predict(Model, type = "response")

predicted.Y <- levels(data$Y)[which.max(probabilities)]

• The first line creates predicted probability that \(Y_i\) equals each of the K choices of Y for each row \(i\).
• The second row assign all rows with the outcome level associated with the higest probability.
• Replace “model” with the name of your model.
• Replace data with the name of your data set.
• Replace Y with your response variable.