[Retracted] RFID Data Analysis and Evaluation Based on Big Data and Data Clustering

Table 2

Types of clustering methods.

Type	Advantage	Insufficient
Based on partition method	Wide application, fast convergence, incremental clustering, and suitability for large-scale data	It is necessary to determine the NCA, which is sensitive to initial values and outliers, so as to find circular clusters
Hierarchy-based method	It does not need to determine the NCA and can find clusters of any shape, which is suitable for data of any attribute and has strong clustering ability	No backtracking, no exchange of data objects between classes, no full processing of large-scale data, and no incremental clustering
Density-based method	It does not need to determine the NCA, can find clusters of different shapes, can detect outliers, and has good adaptability to large datasets	It is very sensitive to parameters. For datasets with uneven density distribution, the quality of clustering results is not high