You want to detect outliers on a data set of $500$ examples and $4$ regular numeric attributes with different ranges. You are unsure if these attributes are truly independent of one another so you must assume they are not. In this case you should $($Select ANY correct answer$)$

A$.$ normalize the data first, and then use an outlier detection operator.

B$.$ use an outlier detection operator first, then normalize the data.

C$.$ use HBOS as it allows for attribute dependencies.

D$.$ uselocalouther factor as it allows for attribute dependencies