The first step is data preparation

shafi765@gmail.

Form a cluster tree: After the iteration, DIANA generates a hierarchical tree, which is the hierarchical structure of clusters, called a cluster tree. The tree reflects the splitting process, with each node representing a cluster. The DIANA algorithm uses a top-down recursive splitting strategy, not a bottom-up merging strategy like AGNES. Because it chooses to split at each stage and gradually refines the clustering structure, it is also called a "divide and conquer" clustering algorithm. From the algorithm characteristics, we can find that DIANA is suitable for relatively


small data sets, because the dissimilarity between all cluster pairs needs to be calculated Malaysia Phone Number Data at each step. If the data scale is larger, the computational complexity will become higher and the efficiency will be lower. will decrease. What problems can hierarchical clustering algorithm solve? What I have learned on paper is ultimately shallow, and I know that I have to do it in detail. What role can hierarchical clustering algorithms play when faced with practical problems? Next, we will take a look at hypothetical cases in application scenarios based on the principles of





the AGNES algorithm and the DIANA algorithm respectively, and learn more about what problems they can solve from the cases. [AGNES Algorithm-Document Classification] The AGNES algorithm can solve the problem of document classification, such as classifying similar documents into one category. Suppose there is a large amount of document data in the company's file management system, and we need to complete the document classification work in the shortest possible time. When there are many types of documents and a large number of documents, we can use the AGNES algorithm to solve the problem. .