What are the audit analysis methods under the background of big data?

First, the application and methods of data mining in the era of "big data"

Data mining is to mine information and discover knowledge without clear assumptions. So the information it gets should have three characteristics: unknown, effective and practical. Therefore, data mining technology is application-oriented from the beginning. At present, data mining technology has been widely used in enterprise marketing. Including: database marketing, customer group division, background analysis, cross-selling and other market analysis behaviors, as well as customer churn analysis, customer credit score, fraud discovery and so on. In the past, the data mining of audit department focused on the analysis of a large number of data to confirm whether there are problems in the data and their performance. With the rise of performance audit, audit departments also need to make audit evaluation of various behaviors of audited units through data, which also needs the support of data.

There are many methods of data mining, mining data from different angles. Most of them can be used in audit work. 1. Data generalization. A large number of detailed data are usually stored in the database.

Through data generalization, a large number of task-related data sets can be abstracted from a lower conceptual level to a higher conceptual level. Data generalization can be applied to descriptive mining in audit data analysis,

Auditors can describe data sets from different granularity and different angles, so as to get an overview of a certain kind of data. A large number of studies have confirmed that compared with normal financial reports,

False financial reports often have some structural characteristics. Auditors can use concept description technology to mine the data stored in the audited database,

By using data generalization techniques such as attribute generalization and attribute correlation analysis, detailed financial data are expressed at a higher level, and the general attribute characteristics of financial statements are obtained.

So as to provide a basis for auditors to judge false financial reports. 2. Statistical analysis. It is a model-based method, including regression analysis, factor analysis and discriminant analysis.

This method can be used to classify and predict data. Through classified mining, the descriptions or models of all kinds of data in the audited database are mined.

Or auditors can predict and analyze a large number of financial or business historical data of the audited entity through the established statistical model, and compare the predicted value with the audit value according to the analysis, which can help auditors find audit doubts.

So it was listed as the focus of the audit. 3. Cluster analysis. Cluster analysis is to divide a group of individuals into several categories according to similarity, with the aim of making the distance between individuals in the same category as small as possible.

The distance between different types of individuals is as large as possible, and this method can provide different types of information sets for different information users. For example, auditors can use this method to identify dense and sparse areas and find the distribution pattern of audited data.

And further determine the key areas of audit. The financial statement data of an enterprise will change with the change of its business. Generally speaking,

The data changes of major items in real financial statements have certain regularity. If the change is abnormal, it means that the abnormal points in the data may hide important information.

There may be false elements in the data reflecting the audited statements. 4. Association analysis. By using association rules, it can extract frequent patterns from all the details or transactions that operate the database.

Its purpose is to mine the interrelationships hidden in data. By using association analysis, auditors can use association rules to mine and analyze the data in the audited database and find out the relationship between different data items in the audited database.

So as to find the abnormal data items, and on this basis, through further analysis, find the audit doubts.

Second, in response to the era of "big data", audit analysis should be adjusted.

From the above analysis process, we can easily see that all aspects of data storage, processing, analysis and mining in the era of "big data" have greatly changed in technology compared with traditional methods, but there is no obvious change in basic principles. There is no need for the original audit analysis mode to change accordingly because of the arrival of the era of "big data". However, the era of "big data" not only brings opportunities to audit analysis, but also brings us considerable impact. It is necessary for us to pay considerable attention to this and make corresponding adjustments in the future information construction process.

1, data storage and processing. The application demand of big data analysis affects the development of data storage infrastructure. With the continuous growth of structured data and unstructured data, as well as the diversification of analytical data sources, the previous storage system design has been unable to meet the needs of big data applications. The architecture design of block and file-based storage system needs to be adjusted to meet these new requirements. When selecting the corresponding storage system, the audit department should pay enough attention to unstructured data and make relevant preparations for collection. At the same time, with more and more units and years of data collection, the amount of data will inevitably increase on a large scale. Even mass data storage systems must have the corresponding level of scalability. The expansion of the storage system must be simple, and the capacity can be increased by adding modules or disk enclosures, even without downtime. At the same time, in order to improve the data processing ability and solve the I/O bottleneck problem, we can consider various modes of solid-state storage devices, from simple cache in the server to all-solid-state media expandable storage systems through high-performance flash memory.

2. Unstructured data processing. The diversification of unstructured data brings new challenges to data analysis, and we need a set of tools to systematically analyze and refine data. Semantic engine needs to design enough artificial intelligence to actively extract information from data.

3. Visual analysis. Users of data analysis include data analysis experts and ordinary users, but their most basic requirement for data analysis is visual analysis, because visual analysis can intuitively present the characteristics of big data, and it is also easy for readers to accept, just like looking at pictures and talking.

The construction of "one platform, two centers" is an important content of the informatization construction of the audit office at present. Through the construction of data center, the problem of data storage and processing can be solved to a considerable extent; The data-based audit analysis platform can also perform a considerable part of the functions of visual analysis to a certain extent, but the storage and processing of increasingly huge unstructured data will be the biggest challenge facing the audit department.