The Application Research Of Telecom Operation Big Data In Bank's Personal Customer Credit Evaluation

With the advent of the Internet consumption and the online payment era,the personal net loan business of China's commercial Banks presents a boom trend,and the bank's assessment method for customer credit is transformed from the original central bank credit reference to the establishment of big data risk control and trust platform based on Internet behavior information.The level of modeling,automation and quantification of risk control decision-making has been continuously improved.Because the Internet big data risk control technology is widely used,on the one hand,it improves the efficiency of the financial institutions credit batch approval,on the other hand,there are weaknesses in data quality,model robustness and other aspects,and it is easy to be attacked by criminals through technical means and cause credit losses.This paper argues that the accumulation of telecom big data by telecom operators can effectively make up for the deficiency of Internet big data in the field of risk control,through a set of practical algorithm,and effectively extract contains in the telecommunications customer behavior characteristics of big data,with the great Internet data fusion together,build risk control model and system,has obtained the good effect of credit assessment.This paper first reviews the establishment and development of credit evaluation system at home and abroad,combined with the characteristics and needs of big data credit investigation technology,this paper analyzes the unique advantages of big data in telecom operation in credit investigation evaluation,points out its application value and application method in credit links such as before,during and after loan.In terms of specific data analysis algorithm,at present,most researches on big data risk control at home and abroad build algorithms and models based on pure mathematics and statistical principles.This paper tries to absorb the idea of interdisciplinary research,looking at big data from a signaling perspective,to explore the classic graphic signal processing theory and technology for the analysis of telecom big data.Finally,gaussian filtering,k-means clustering,histogram statistics and other planar signal processing algorithms are combined organically,extract and describe the highly invariable user behavior characteristics in telecommunication data.Taking telecom behavior characteristics as weight data items,participated in the establishment of riskcontrol score card,and obtained the expected risk identification and customer differentiation effect.Finally,this paper is based on "MPP+Hadoop" hybrid architecture is designed,the algorithm and the successful landing score card model for bank credit approval system,the Internet process and model prove that the proposed algorithm has a good prospect of utility engineering.Conclusion: Based on the relatively mature and efficient algorithm,this paper extracts the user behavior characteristics in the telecom operation data.With its participation in the construction of the big data credit risk control model,the overall reliability of the credit score model has been improved successfully.In the quantitative evaluation based on K-S value statistics,the risk control model after the introduction of telecom big data has excellent improvement effect and performance in terms of sample differentiation degree and grading accuracy.