Research On Quality And Forecast Analysis Of Big Data In The Field Of E-commerce

Nowadays,the prevalence of information technology,especially Cyber-Physical Systems,Internet,cloud computing and social network,has largely fueled the popularity of big data.While creating huge value,big data also brings several severe challenges,and quality of big data is one of them.In addition,with the advent of the era of big data,the scale of the data set is getting bigger,the traditional data analysis methods can not solve the problem of large-scale data set.However the hidden information behind the big data,especially in the field of e-commerce,has became the key factor between competitive enterprises.For instance,both Alibaba and JD have employed the mainstream cloud computing platform to conduct big data analysis,which manifests that e-commerce companies are attaching increasing attention to big data mining analysis.Currently,big data quality appraisal system remains to be developed globally,we through the study of existing data quality,and based on the big data of the four characteristics of huge volume,massive variety,fast processing speed and low value density,this study contributes to the literature by suggesting five quality dimensions:availability,usability,reliability,dependency and appearance quality.For each dimension,one to three elements are examined and explained in detail.Finally,a big data quality appraisal system is built based on this big data quality system.After the evaluation of the data quality appraisal system,the quality data can be used to analyze and research and create value for the enterprises.According to the received big data quality evaluation system,assess and preprocess all the data,and then using a support vector machine(SVM)based on parallel computing method to analyze the data.This method starting from dividing training samples into several subsets by SOM,this method then trains SVM learning machine on each subset and combine the training results of each subset in order to reach the goal of processing massive data prediction and analysis efficiently.The big data quality standard and quality evaluation system proposed in this paper has good expansibility and adaptability,which can meet the needs of big data quality assessment.It has certain meanning of solving the shortage of quality assessment methods.Finally consider the parallel support vector machines(SVM)in data mining and analysis of the outstanding performance,we will put it into data analysis for e-commerce,the result shows the parallel support vector machine(SVM)is a good solution when dealing with large-scale data set with inefficient problem.