Density Peak Clustering Algorithm Based On Adaptive Cluster Center

With the development of the era of big data,the novelty of data is gradually improving,so how to classify data is particulary important.Clustering technology,one of the methods of data classification,plays an important role in these pratical applications.The main study in this thesis is the density peak clustering algorithm in clustering technology.Firstly,the density peak clustering algorithm can not adaptively select the clustering center and the number of clusters,a new density peak clustering algorithm is proposed,secondly,aiming at the problem that the density peak clustering algorithm is not accurate enough for complex data sets with many features and the manual selection of clustering centers,a new density peak clustering algorithm is proposed.Aiming at the problems that the density peak clustering method in the density clustering algorithm cannot adaptively select the number of clusters,and manually select the cluster center,etc,a clustering algorithm combining the idea of small class merging and fuzzy c-means algorithm is proposed(CFDPC).Firstly,the algorithm adaptively obtains the number of initial clusters with the idea of subclass merging,secondly,the number of initial clusters and the initial clustering center obtained manually by the density peak clustering algorithm are used as the input conditions of the fuzzy c-means clustering algorithm,after a series of operations,the cluster center point set and the number of clusters are obtained adaptively,realizing the division of the whole data set.The experimental results show that the algorithm is compared with the comparison algorithm in the experiment,it is found that the clustering effect of this algorithm is better,which proves the effectiveness of the proposed algorithm.In order to effectively deal with problems,including the clustering of complex data with more features by the density peak clustering algorithm is not accurate enough,and selcecting the clustering center manually,thus,a new clustering algorithm is formed by combining attribute weight Markov distance and fuzzy Cmeans algorithm(WNDPC).The algorithm uses Markov distance of attribute weight as the distance measure,the calculation of Euclidean distance is standardized,avoiding ignoring the different importance of multiple features,secondly,the initial clustering center obtained by the density peak clustering algorithm is used as the input condition of the improved fuzzy c-means clustering algorithm,so as to adaptively obtain the cluster center point set,realizing the division of the whole data set.The experimental results show that the algorithm is compared with the comparision algorithm in the experiment,and finally achieves the purpose of optimizing the density peak clustering algorithm,improves the shortcomings,and has a better clustering effect.