Correct Answer - Option 3 : K - nearest neighbor method
The correct answer is "option 3".
CONCEPT:
Pattern recognition system uses samples from known categories to form decision rules for unknown patterns.
The unknown patterns then get assigned to one of the categories, according to the decision rules.
Clustering method is used in pattern recognition to group similar patterns into clusters whose elements are similar to each other.
EXPLANATION:
option1: K-Mean method is vector quantization method & used for cluster analysis in data mining.
option2: Self Organizing feature map method is a data visualization technique & used for clustering.
option3: K - nearest neighbor method is used for regression & classification but not for clustering.
option4: Agglomerative method uses the bottom-up approach in which each cluster can further divide into sub-clusters i.e. it builds a hierarchy of clusters.
Hence, the K - nearest neighbor method is not a clustering method.
