ValueError: Metric 'cosine' not valid. Use sorted(sklearn.neighbors.VALID_METRICS['ball_tree'])

Question

I'm trying to use cosine metric with Ball_tree algorithm since my data is of high dimensions. e.g.

nbrs = NearestNeighbors(n_neighbors=2,algorithm='ball_tree',metric='cosine').fit(data_seed)

However i get the following error:

ValueError: Metric 'cosine' not valid. Use sorted(sklearn.neighbors.VALID_METRICS['ball_tree']) to get valid options. Metric can also be a callable function.

I don't understand why i cannot use cosine similarity with ball tree?

see: https://stackoverflow.com/questions/46791191/valueerror-metric-cosine-not-valid-for-algorithm-ball-tree-when-using-sklea — A. K., May 16 '22 at 14:35
I'm asking why i cannot use cosine with balltree. I know it states it's incompatible — Maths12, May 16 '22 at 14:42

user16181 · Answer 1 · 2022-10-14T13:44:20.047

0

check valid matrics sklearn.neighbors.VALID_METRICS['ball_tree']) and use instead of cosine

edited Oct 14 '22 at 13:44

answered Oct 13 '22 at 17:59

user16181

126
4

ValueError: Metric 'cosine' not valid. Use sorted(sklearn.neighbors.VALID_METRICS['ball_tree'])

1 Answers1