Research On Large Data Analysis Based On The On-line Fuzzy Ant Clustering

Under the background of big data,data is not in the traditional sense of the simple processing objects,but start to change into a kind of basic resource of the society.Big data has brought tremendous challenges and opportunities on the number of scale for data storage,as well as the storage of data management and data analysis.Big data imply enormous social,economic,and scientific research value.However,the traditional data analysis methods are based on the loading level of data for analysis,so the processing and analysis method on these huge amounts of data are changing massively.There are a lot of the methods based on sampling to clustering analysis for these large amounts of data,especially the sampling extension based on the fuzzy c-means algorithm,such as online fuzzy c-means algorithm.This kind of algorithm is widely used,on the one hand is because of its easiness to implement,algorithm idea is simple and easy to understand.On the other hand is because of the fuzzy c-means algorithm faster calculation speed,for the mass data processing speed is a very big advantage.However,this kind of fuzzy c-means algorithm has a very large defect,which is very sensitive to cluster center,so online fuzzy c-means algorithm must initialize the cluster center number and center position.Correcting the initialization of cluster number is very difficult;the gap between center of incorrect initial cluster number of clustering results and the actual cluster class will be very big.This paper presents a improved algorithm based on ant clustering class down to avoid this problem,the background of big data class ant clustering algorithm is a kind of adaptive algorithm,so it does not need the cluster number and cluster center position of the initial value,it can also compare the correct number of clusters and the results are obtained.Online class ant clustering algorithm is proposed in this paper,combined with the merits of two algorithms of various and under the background of big data presents some special processing mechanism of the organic combination of online class ant clustering algorithm can be a very good solve online fuzzy c-means algorithm is sensitive to the initial value problem.This article in detail in chapter 3 online ant clustering algorithm and algorithm steps,we in by doing a lot of experiments to test and contrast the on-line fuzzy c-means algorithm with online ant clustering algorithm of accuracy on the three data sets.Finally,in order to test and use the proposed online ant clustering algorithm and the on-line fuzzy c-means algorithm more conveniently,this article in the fourth chapter designs and implements to the two algorithms and debugging for the system.This can make good use of visualization technology to the performance of the contrast between the different algorithms;this also deepens the understanding of the results of the clustering algorithm.