A Novel Clustering Algorithm via the Support and K-Nearest Neighbors of Data.

K-Means is a widely used algorithm among many clustering algorithms. However, in the process of clustering by K-Means, there are problems, such as the difficulty of selecting K value, high parameter dependence and sensitive outliers. Thus, we propose a novel clustering algorithm via the support and K-Nearest Neighbors (KNN) of data to solve these problems in this paper. Firstly, we introduce the concept of data support and calculate and make the supports of pair data sparse. Secondly, we calculate the sum of the supports for each data point with other points and select and consider the data point with the maximum sum of the support as the cluster center among the data set. At the same time, we use the KNN algorithm to determine the K value selection, and the initial data set is divided into M sub-clusters (M k ). Then, we combine the M sub-clusters into the final clustering by using the aggregation function as the similarity measure. When processing data, the algorithm combines density and distance and tests it on the UCI data set. Comparing with other typical clustering algorithms, the hybrid algorithm based on distance and density has a better clustering effect.