Research on Data Stream Clustering Based on FCM Algorithm1

Based on traditional FCM clustering algorithm, this paper proposed the data stream clustering algorithm based on FCM. When dealing with data stream, inheriting the membership degree matrix can make traditional FCM more effective for data stream. The conclusion is that the result of data stream FCM is basically same as traditional FCM. But some points with similar distances from different centers of the clusters are easily misclassified. In computing time, data stream FCM is more sensitive to the number of parts the data divided. But it isn't so sensitive to the number of data arrives per time.