How should the financial big data platform be built and applied? Are there any financial cases that can be used for reference?

The construction and application of financial big data platform are two very important parts of financial big data platform. Therefore, in the next part, we will elaborate from the perspective of big data platform and what indicators banks can analyze.

I. Big Data Platform The overall architecture of a big data platform can be composed of the following parts:

As shown in the figure, there are several links from bottom to top:

1. Business application: in fact, it refers to data acquisition. How do you collect data? Collecting data online is relatively simple, and data can be collected through web pages and apps. For example, many banks now have their own apps, which can collect user behavior data at a deeper level, and can be divided into many dimensions and analyzed in detail. But for offline industries, data collection needs to be completed with the help of various business systems.

Second, data integration: actually refers to ETL, that is, users extract the required data from the data source, clean the data, and finally load the data into the data warehouse according to the predefined data warehouse model. The Kettle here is just one of ETL.

Third, data storage: refers to the construction of data warehouse, which can be simply divided into business data layer (DW), indicator layer, dimension layer and summary layer (DWA).

4. Data * * * sharing layer: provides data * * * sharing service between data warehouse and business system. Web Service and Web API represent a way to connect data, and there are some other ways to connect data according to your own situation.

Fifth, the data analysis layer: the analysis function is easy to understand, that is, various mathematical functions, such as K analysis of mean, clustering, RMF model and so on.

Column storage allows each page on disk to store only one column of values instead of a whole row of values. So the compression algorithm will be more efficient. In addition, this can reduce the I/O of the disk and improve the cache utilization, so the disk storage will be used more effectively.

Distributed computing can divide a problem that requires a lot of computing power into many small parts, then hand over these parts to many computers for simultaneous processing, and then synthesize these calculation results to get the final result.

Combining these two technologies can greatly improve the efficiency of analysis.

Yonghong MPP is the best in these two aspects at present.

Sixth, data presentation: in what form the results are presented, it is actually data visualization. Agile BI is recommended here. Different from traditional BI, it can generate reports by simple drag and drop, and the learning cost is low. In domestic agile BI, individual users recommend Tableau, and enterprise-level requirements like banks recommend Yonghong BI.

Seven, data access: this is relatively simple, depending on how you look at these data. The example in the figure is because of the B/S architecture, and the final visualization result is accessed through the browser.

Second, how to build a bank data analysis system? Building a data platform may be a project work, which will take some time to complete, but building a data analysis system has a long way to go. But if someone can share similar data application experience in the financial industry with you while making products and help you build a data analysis system, that is the real "good medicine".

Let's share a case in which Yonghong Technology helped a large bank build a data service platform.

According to the behavior path of customers handling business in banks, there can be several themes, and different themes have corresponding scenarios and indicators.

1. A customer

Customer subject: a broad table, which consists of customer attributes (customer number, customer category), indicators (total assets, products held, number of transactions, transaction amount, RFM) and contract signing (channel contract signing, business contract signing).

2. Made a deal

Transaction subject: a broad table of transaction financial attributes, business categories and payment channels.

3. Which account should I use?

Account subject: a broad table of account attributes (customer, account opening date, branch, product, interest rate and cost).

4. Through what channels?

Channel theme:

Channel attributes, dimensions and quotas form a wide table.

5. What business is involved &; product

Product theme: a broad table of product attributes, dimensions and indicators.

Third, the case in view of the space problem, you can refer to this article here:

Huaxia Bank: Big data technology serves business needs and achieves rapid sales growth.