SAS Yang Yue: Large data variable small data data analysis layers progressive

November 24, sponsored by ZDNet to the top of the network, Intel co-organizer, VMware and SAS supported the theme of "large data?" The fifth session of the Forum on the Scientific development of government decision-making was held in Kunshan. A number of government information experts, large data practice elite and IT industry opinion leaders attended the meeting, and in the Conference on the large data in the actual application of information technology and other issues of in-depth exchanges. Among them, the SAS government industry senior manager Yang Yue analyzes the characteristics and key points of the data value from various aspects of technology and application, and sums up the elements that the enterprise should do well in the data analysis when dealing with large data.

Data analysis is divided into two stages

The big Data age has both opportunities and challenges, and Yang Yue that the biggest challenge at the moment is that the ability to analyze and execute data is much faster than the data changes. This is precisely the result of the three characteristics of large data-large data, rapid data growth, complex data types, and the gradual reduction of the cost of data storage is one of the reasons.

However, this challenge is the root cause of the emergence of data mining technology. In fact, the process that data produces value is the process of data mining analysis and utilization. Data mining provides an opportunity for enterprises to "discover problems, predict the future and optimize their business", so that the enterprise data can be refined management and the decision basis is exported.

In the process of data analysis, we should pay attention to the transformation of two stages, the first stage is the transformation from commercial problem to data analysis, and the second stage is the transformation from data analysis result to business solution. The final analysis results are mainly affected by the definition factors of the problem, the choice of variables, the model and the adjustment factors of the parameters. In the process of analysis, the data will be changed into information, knowledge, and eventually become the wisdom, become an important part of the intangible assets of enterprises.

As a special data analysis method, data mining technology is also oriented to business application, and it is the ultimate aim to meet business requirements. Data mining, which is done without a clear assumption, often leads to more unpredictable or counterintuitive information or knowledge in "small data." In addition, Yang Yue also stressed that data mining will be more affected by the analyst's business knowledge, thinking mode, and the strength of the data analysis service provider.

Eight levels of analytical ability

From the point of view of analyzing the mining system, the analysis layer above the data layer in the application layer is often easily overlooked by people. The analytical methods and capabilities that are often needed in the analysis layer are considered to be divided into eight levels from low to High: Yang Yue

• Regular reports: Periodic builds that reflect the situation in a given area in the short term and cannot be used to make long-term decisions

• Ad hoc enquiries: constantly ask questions and find answers

Multidimensional Analysis: Through multi-level drilling analysis to find the problem

• Alert: Record the time when the problem occurs and notify when the problem occurs again

• Statistical analysis: Statistics and summary rules in historical data, including some more complex analysis

• Forecast: Accurate forecast market demand, wide application

• Predictive modeling: In a large customer base, customers are divided by predicting different responses from different demand customers

• Optimization: Find the best way to achieve a goal based on existing resources and requirements

For these eight analytical capabilities, Yang Yue that most customers still have only the top four analysis capabilities, which is also the main function of traditional business intelligence. They analyze what happened in the past by summarizing historical data, but lack the ability to anticipate the future. If the enterprise will face complex business problems, or want to be able to predict future trends, you need to have four of in-depth data analysis and mining capabilities to create new data value. And the real best solution is to use all the analytical skills to get the highest level of business intelligence.

High performance data analysis platform to enhance customer experience

In the practical application of data mining, SAS has established a complete set of data mining project methodology model including defining business problem, System environment assessment, data preparation and so on. Yang Yue said that the implementation of methodological models, the coordination of analysts, organizations and institutions is the focus. The demand usually comes from the business unit, but in fact the system problem is often easy to become the information barrier, so the innovation of the Organization and the perfection of the system become the key.

With regard to data analysis and mining solutions, SAS has also brought a case for the financial industry. A credit card issuer faces many problems with the data, making it difficult to make a marketing strategy. Limited by the amount of data, they can only through the transaction point of statistical data, the old system daily Analysis report update time is too long, while the static report information coverage is insufficient, resulting in customer service department can not capture user information, product departments can not be based on user behavior data customization product solutions.

However, after the introduction of High-performance data analysis platform, through the single user's card trading, income changes, negative feedback, such as a series of information tracking and mining, the bank can be able to different users for accurate real-time positioning, and targeted to the user needs to make immediate response, Thus the bank's customer service level has been upgraded to a level that fully embodies the advantages of in-depth data mining and business intelligence optimization.

From the importance of understanding to "large data" to "small data", from the comprehensive construction analysis layer eight ability to the various analytical technique method grasps, the SAS presents a set of ideas is in the big Data age, the enterprise must do is not all-inclusive, but with small achievement big, with the detail achievement whole, transforms the big data to the real business value great wisdom.

(Responsible editor: The good of the Legacy)