3 ms·
IMHO the first port of call when clustering. You can use kd trees to speed it up. Approx nearest neighbour queries for super speedups. You can transform your da
by tlarkworthy 13y ago
IMHO the first port of call when clustering. You can use kd trees to speed it up. Approx nearest neighbour queries for super speedups. You can transform your data before applying to get better results, eg pca You can use it to initialize "better" clusterers like Gaussian mixture models.
- msellout 13y agoI think you mean k-means is "the first port of call when clustering." K-nearest neighbors is a supervised learning algorithm (for regression or classification), not a clustering algorithm.