The name of the parameter 'affinity' is misleading, since it does not correspond to similarity/affinity functions, but to distance functions.

Besides, the documentation for the parameter 'affinity' seems wrong.

Metric used to compute the linkage. Can be “euclidean”, “l1”, “l2”, “manhattan”, “cosine”, or “precomputed”. If linkage is “ward”, only “euclidean” is accepted. If “precomputed”, a distance matrix (instead of a similarity matrix) is needed as input for the fit method.

If affinity="precomputed" a distance matrix, instead of a feature matrix (not a similarity matrix), is needed as input for the fit method ?