Consider the case of a binary classification problem in which the response is sampled from a non-linear complex function, which of the following algorithms has the potential to perform the best?
a) Logistic Regression
b) K-nearest neighbors
c) Linear Discriminant Aanlysis
d) Quadratic Discriminant Analysis
I am not sure what "the response is sampled from a non-linear complex function" means..Could someone explain this for me? Thank you!
From my understanding, the author seems to try to suggest that it is not something of which the decision boundary is easily described.
For example, we should not be able to separate the points using a single straight line or a simple curve.
Some positive cluesters can be surrounded by a sea of negative data points and similarly some negative clusters can be surrounded by a sea of positive data points.