Research On Fuzzy Clustering Algorithm Based On Density Function

Cluster analysis refers to the process of grouping similar data samples into the same cluster by analyzing the internal characteristics and attributes of data.Fuzzy clustering is widely used because the concept of fuzzy membership degree is compatible with the situation that the category of things is not either-or.However,fuzzy clustering algorithm also has many disadvantages,such as the clustering results are easily affected by the initial cluster center,easy to fall into the local optimal,and need to set the number of clustering categories in advance.Density Weighted Fuzzy C Means(DWFCM)algorithm based on Density Weighted Fuzzy C Means(DWFCM)is proposed to solve the problem that the initial clustering center randomly selected is prone to fall into local optimal and the clustering result is susceptible to isolated noise points.Firstly,the density and region partition length of all sample points in the space were calculated.C points with the highest density in unlabeled regions were selected as the initial cluster center,and the regional points centered on this point were marked.In the process of iteration,the distance between other sample points and the sample point is calculated and multiplied by the effective radius of the neighborhood density.The reciprocal sum of the product is taken as the density value of the sample point.The density value is normalized and the membership degree is weighted.The proposed DWFCM algorithm is applied to the customer analysis of air passenger transport,and the practical case application of the algorithm is explored and analyzed.Through experimental comparison,the initial clustering center selected based on density function improves the stability of FCM algorithm and the accuracy of results,and reduces the number of iterations of the algorithm.Compared with the traditional FCM,DFCM has higher accuracy in the data set containing solitary noise points,and DWFCM has the characteristics of low iteration times and high accuracy in the standard data set.However,the efficiency of the algorithm is not improved in the concentration of common data without solitary noise points,so it needs to be improved.