【译】顺序聚类 | idouba

# 【译】顺序聚类

@todo 翻译下文
Sequential Clustering

Sequential algortihms are pretty fast and straightforward methods, which tend to produce compact clusters. All the feature vectors are presented to the algorithm only once or just a few times, and the final clustering is usually strongly dependent on the presenting order of the data.

#### Basic Sequential Algorithmic Scheme – BSAS

In this algortihm each data vector is considered only once and the number of cluster in not known a priori. However, the maximum number of clusters, q, is decided beforehand. Let d(x, C) denote the distance between data vector x and a cluster C. Furthermore, q is the allowed threshold of dissimilarity.

#### Modified Basic Sequential Algorithmic Scheme – MBSAS

MBSAS algorithm consists of two phases. The clusters are determined in the first phase via assignment of some of the vectors of X to them. The unassigned vectors are assigned to the appropriate clusters in the second phase.