Outliers for classification; imbalanced data for regression? When I hear of the terms "outliers" and "imbalanced data" it's usually in the context of regression and classification respectively.
With "outliers" meaning the continuous response falling substantially outside of the rest of the data and "imbalanced data" meaning the discrete response is primarily of one class.
Is there a such thing as outliers in classification (I'm not talking about leverage here, I'm strictly talking about outliers in the response)? Is there a such thing as imbalanced data in regression?
 A: 
Is there a such thing as outliers in classification (I'm not talking
about leverage here, I'm strictly talking about outliers in the

You can have outliers in the features, but you are asking about the response. In classification, the response is a multi-dimensional array of zeroes and ones (it has a single dimension for binary classification and multiple for multi-class and multioutput classification). In such a case, there cannot be outliers ("strange values"), since the values can be only zeros and ones. If they aren't, it's not an outlier but rather a problem with the data (invalid coding, bug in the preprocessing code).

Is there a such thing as imbalanced data in regression?

Yes, kind of. Let's say that your variable is human age. You gathered your sample by approaching random people on the street and giving them a survey to fill in. By accident, the place where you gathered your surveys is close to a secondary school, so while other age groups are represented quite well, teenagers are overrepresented in your sample. This kind of imbalance can be much more subtle and harder to diagnose than the one in classification.
Moreover, keep in mind that you can have multivariate outliers, where each of the variables individually looks okay-ish, but their combination is an anomaly (e.g. very high preschool kid). Same with data imbalance (e.g. underrepresented 70+ black men). In such a case, it applies to both regression and classification.
