CommencerCommencer gratuitement

Remove the intercept

In the last exercise, we forgot to exclude the intercept. When we estimate a model with the intercept, it estimates parameters for each of the alternatives. For this conjoint data, the alternative number just indicates which chocolate bar was presented on the left, middle or right in the survey question. It doesn't make much sense to assume that the position would affect the choice.

Cet exercice fait partie du cours

Choice Modeling for Marketing in R

Afficher le cours

Instructions

  • The choc_m1 object is still loaded. Use summary(choc_m1) to view the output. Find the parameters for the intercepts. They are close to zero and non-significant, so we can remove them.
  • Modify the call to mlogit() to exclude the intercept by adding 0 + just after the ~ in the formula. Assign the output to choc_m2. You can also remove the print.level = 3 option.
  • Call summary(choc_m2) to confirm that the intercepts are removed. Did the other parameters change much?

Exercice interactif pratique

Essayez cet exercice en complétant cet exemple de code.

# summarize the choc_m1 model that includes the intercepts


# modify the call to mlogit to exclude the intercept
choc_m2 <- mlogit(Selection ~ Brand + Type + Price, data = chocolate, print.level=3)

# summarize the choc_m2 model
Modifier et exécuter le code