Is there a way to determine the number of distance calculations performed during the kNN computation? The brute-force approach requires evaluating the distances between every query and corpus point, resulting in $n_q \times n_c$ calculations. However, methods like KDTree or BallTree can potentially reduce this number, especially in low-dimensional spaces.
Does the package provide a mechanism to retrieve the actual number of distance evaluations during the kNN computation?
Thank you!
Is there a way to determine the number of distance calculations performed during the kNN computation? The brute-force approach requires evaluating the distances between every query and corpus point, resulting in$n_q \times n_c$ calculations. However, methods like
KDTreeorBallTreecan potentially reduce this number, especially in low-dimensional spaces.Does the package provide a mechanism to retrieve the actual number of distance evaluations during the kNN computation?
Thank you!