0

I have a continuous variable which I divide in 3 class. Now I have run the multinomial logistics model to predict the occurrence of any particular class. My question is can I convert the predicted probabilities of multinomial logistics model to calculated a continuous variables. If yes, what is the formula of the same. In the binomial distribution the expected values is calculated as np. Can I use the same logic here too? For example: I have a score which range from 0-2 in my historical series. I classify it in 3 classess (0-06, 0.6-1 and >1) and run the multinomial logistics model. If I will use the final model to score my testing data. I would get the output as a probability. Is there any way I can convert this probability into a numeric values which can represent my score?

0 Answers0