At first: I have individuals represented by vectors with four entries/properties:
Individual1:
Height Age Blood Gender
171 24 A w
Individual2:
Height Age Blood Gender
179 21 B m
Individual3:
Height Age Blood Gender
181 33 B w
(..)
It's just an arbitrary example but assume I want to measure similarities between two individuals across the whole pool. This means comparing Individual1 with Individual2, Individual3 with Individual4 and so on. Therefore I want to use the Mahalanobis distance to measure the distance between vectors:
The question is: Does the covariance matrix consist of the two individuals x and y or of the whole pool of individuals? I can't find a clear answer anyhwere..
