Six strategies for big data of commercial banks (2)

Is big data equivalent to a data warehouse?

As mentioned above, whether commercial banks have big data capability should be judged by the specific utility of data and data analysis system. To "significantly enhance the efficiency of data analysis and business decision-making", "significantly improve customer understanding and cognitive ability", "low-cost, batch to achieve a higher level of personalized customer service" three standards to measure, the current commercial bank data Warehouse construction needs to be strengthened in the following aspects.

Construction of heterogeneous Data Warehouse platform. Over the years, the commercial Bank of the data Warehouse to storage business, transaction data-based, so the procurement of high storage costs of professional data warehousing services, data in the warehouse before entering the ETL rules relatively strict, and the "Time for space" strategy for the theme split to save storage space, This results in high computational resources and reduced overall analysis efficiency when performing analytical tasks such as transaction chain recovery and transaction scenario restoration. The log data, which is closely related to user behavior data, has the characteristic of "large data volume, high frequency but low value density", which can be used to build a low-cost PC cluster, memory database and so on, and integrates with the existing Data warehouse to form a transparent heterogeneous data warehouse for data source and analysis. Improve their response speed and processing power.

Build a shared platform for business indicator extraction logic. At present, the standardization of the basic data of commercial banks has made great progress, but in practical application, there are still "business logic Information Island" phenomenon (that is, because of the lack of a shared platform, resulting in different analysts can not exchange business indicators of the extraction logic, each analyst, Each data analysis department forms an island. This phenomenon not only causes the problem of "polymorphism" of the business Index, but also induces the repeated submission of data Warehouse access request, which affects the efficiency and accuracy of data analysis, so it is necessary to set up the logical sharing platform of the business index of reasonable authority control and solve the problem of "business logic Information Island".

The establishment of Big data Innovation Project mechanism, which is dominated by the Information governance department and centered by the business unit. Big data applications require that the data analysis business chain be compressed as much as possible to further improve the degree of integration between specific business and data analysis, and in this way, we can explore the establishment of big data Innovation Project mechanism, which is dominated by information governance departments and centered by business departments. In short, it is the integration of data analysts into specific business units, by the data analysts and specific business units jointly launched the big Data application of the innovation project, after the Information governance department approval, to give the corresponding computing resources, and according to the data application project in the specific business results of the evaluation and incentive.

Does big data need only sea Dupre platform?

The Apache Software Foundation (ASF)-based Dupre (Hadoop) Open source project is undoubtedly a huge boost to big data applications, and the Hadoop HDFs system is also an important infrastructure for today's mainstream big data platforms, so is there a Hadoop platform The commercial bank has the big data processing ability?

First of all, from the completeness of the hardware and software platform, we also need to continue to invest and configure more software modules to enhance the capacity of the big data analysis platform. Hadoop is just the infrastructure of the big data analytics platform, with data analytics, Mahout, Hadoop-r, Hadoop-weka and more, in addition to the Hadoop and yarn-based hive, HBase, Pig, and Storm. Data Mining Suite for big data analysis is also essential, in addition to faster, higher performance of the spark system in the Internet enterprises have been successful applications, it is worth the attention of commercial banks and reference.

Secondly, from the point of view of data, we need to transform the front-end to obtain more dimension, higher frequency and finer granularity data. The data analysis system of Commercial Bank attaches great importance to the storage of business data for a long time, but it does not pay attention to the log of system running state and the collection of personal information of customers, which is the key of understanding the customer and troubleshooting the business problem. Therefore, commercial banks need to systematically carry out the application front-end transformation, learn from the Internet Enterprises, e-commerce enterprises, try to obtain more dimensions, more high-frequency, more granular data, and better meet the needs of big data analysis data sources.

Finally, from the implementation of the project, it is also necessary to form a "data analysis + Business application" Data analysis mode to iteratively optimize the analysis results and specific business. In traditional bi mode, the business process of data analysis can be summed up as follows: Accepting the analysis requirement from the business unit and analyzing the data and forming the report. Many of the big data analysis projects require a continuous iteration of data analysts and business people, and some projects may even be difficult to establish a definite termination point (for example, the e-commerce recommendation system is typically continuously optimized by a team), which requires commercial banks to be able to take "data analysis +" on specific big data analysis projects. Business Application "Data analysis mode to iteratively optimize analysis results and specific business.

It can be seen that the Hadoop platform is not a necessary and sufficient condition for the commercial banks to have big data capability, and the commercial banks need not only continue to invest in the hardware and software platform, but also transform them in front-end design and data analysis mode to be more adaptable to the requirements of big data analysis.

Big data is just the data analytics department?

As mentioned earlier, big data capability is based on data analysis, integration of business decision-making, customer perception, personalized service as one of the comprehensive competitiveness, therefore, big data capacity building should not only be borne by the data analysis Department.

Big data capacity-building should be integrated into development planning at the strategic level. Should do a top-level design, the big data capacity building and information bank construction, combined with the integration of online and offline construction, combined with the Internet Financial Development Strategy, collaborative business, channels, technology, data analysis and other departments, do the top-level design and planning, the formation of "full of big data" atmosphere, From data source combing, data analysis platform building, analysis model establishment, external data exchange rules and other levels to develop clear guidelines and operating standards, speed up the progress of big data capacity building.

Pay attention to the efficiency of the data analysis process. The utility size of big data analysis depends largely on the activity of the data and the speed with which the results are invested in the specific business, so that the business chain of traditional bi is compressed as much as possible. The integration of data acquisition and analysis results can be implemented in both electronic and self-service channels (e.g., product correlation recommendations based on customer personality, real-time pricing based on scenarios, personalization of self-service device interfaces, etc.), as well as the application of big data processing patterns in the traditional bi domain. Based on the intermediate data layer of high real-time, the business intelligence system with higher efficiency, stronger real-time and more customization degree is established, which realizes the real-time, mobile and customization of Business Report.

We should pay attention to the talent reserve and technology accumulation. With the rapid development of big data technology, the talent reserve and technology accumulation of data can not be accomplished overnight, and it needs a considerable amount of continuous investment. Talent Reserve, should be the spirit of "the introduction of a number of, train a batch of reserves," the principle of introducing a small batch of high-level technical personnel, through the implementation of specific projects, training large numbers of technical personnel, and through the big Data Technology competition for universities and society, financing open source community, etc., to form a broad and effective talent reserve. Technology accumulation, should be in accordance with the "open and package, for me," the idea of forming a big data pre-research team, actively carry out the screening, validation, absorption of open source projects, along the "introduction and digestion of big data open source projects-funding big data open source projects-to propose and dominate the big data open source project" path, constantly strengthen their own in the big data technology advantages, to form their own core competitiveness.

"For more information on business intelligence, business intelligence solutions and business intelligence software downloads, visit Finebi Business Intelligence official website www.finebi.com"

Six strategies for big data of commercial banks (2)