00:01
Yes, so logistic regression and discriminant analysis are both used for classification problems, but there are situations where one might be more suitable than the other.
00:09
So for logistic, this would be in the assumption of distribution.
00:24
So logistic regression does not require assumptions about the distributions of the classes in the feature space.
00:31
So use it when you cannot assume normal distributions for your independent variables or when you know your predictors are not normally distributed.
00:43
And then robustness, logistic regression is more robust to outliers in the response variable and does not assume equal variance, covariance matrices across the groups.
00:56
Some advantages would be flexibility.
01:03
It can easily include interaction in nonlinear terms.
01:08
And then again, robustness, so less sensitive to departures from assumptions...