You are running a logistic regression in order to classify observations in a dataset into two classes. When discussing this with your colleagues, three of them raise objections:
- Colleague A: For this dataset, we don't believe that the probability of an observation falling in the ?rst class is well-modeled as a linear function of the predictors, so logistic regression is not suitable.
- Colleague B: We don't believe that the true decision boundary between the classes is linear, so logistic regression is not suitable.
- Colleague C: The observations that we have here are perfectly separated by a hyperplane, so logistic regression is not suitable.
Which of your colleagues (if any) have raised valid objections? Respond to any invalid objections.