Mining the Gold Mine of Enterprise Data

As early as the 1980s, Baoshan Iron & Steel Co., Ltd. (hereinafter referred to as Baosteel) noticed that analyzing the production and operation data of an enterprise can create greater economic benefits. In the early 1990s, Baosteel introduced the technical solutions of the internationally mature SAS company and started the first data mining project: improving the quality of steel plates for shipbuilding. The project has created millions of dollars in benefits. At present, Baosteel has established data warehouses in terms of production, technology, quality, energy, sales, and financial costs, and conducted data mining.

Companies like Baosteel that have taken data mining as a powerful weapon to improve their core competitiveness are still rare in China. But in foreign countries, the application of data mining has been very common.

Since the end of the last century, most enterprises in North America and Europe are preparing to build data warehouses and start enterprise-level data mining. In the "Fortune" Global 500 companies, 98% have applied business intelligence solutions. They are either on data warehouse projects or data mining, and they are closely integrated with business to support enterprises to make correct business decisions.

Discovering laws and predicting laws from massive data

In the past ten years, with the extensive application of computers and networks, enterprises have accumulated more and more data, from suppliers, distribution Information on all aspects of business, customers, and markets is exploding. According to IBM's analysis, today's companies use less than 1% of their total data holdings. Just imagine, what if the other 99% of data could be used?

For data to truly be an asset to a company, it must serve business decisions and strategic development. Otherwise, a large amount of data may become a burden, or even become garbage. The data mining technology of Shali Gold Rush came into being and showed more and more vitality.

The so-called data mining, also known as knowledge discovery in the database (Knowledge Discover Database, KDD), is based on mathematical statistics, artificial intelligence, machine learning and other technologies to extract reliable and novel data from a large amount of data , high-level processing of patterns that are valid and understandable. According to the specific business goals of the enterprise, it mines and analyzes a large amount of data, discovers the inherent laws in time, and makes predictions that conform to the laws of enterprise operation and development.

According to Zheng Xiaojun, senior consultant of IBM China Global Business Intelligence Solutions, data mining is divided into two aspects in practical application.

The first is to discover the laws, through mathematical statistics and artificial intelligence methods, to discover the laws hidden in the data, including correlation laws, frequency and periodic laws, and similarity laws.

A typical example is targeted marketing. Data mining tools can identify customers who are most likely to respond to future emails based on a wealth of data from past emails. Another example is the analysis of retail data to identify products that appear to be unrelated on the surface, but in many cases are actually sold together.

The second is the prediction rule, including the prediction of pure value (value), and the prediction of the discriminant category, that is, yes or no. Prediction rules need to be deployed (deploy) rules, that is, deploy rules to all previous data, or to new data.

For example, the bank finds out which customers have good credit, what kind of behavior pattern does the good credit show, and what kind of behavior pattern does the bad credit show? In the next stage, we will take this rule to see a new customer customers, analyze his behavior patterns, evaluate him, evaluate him, and predict his potential value to the bank.

All in all, data mining can be used in all production and business operations of enterprises that involve discovering and predicting laws. It is not difficult to understand why the vast majority of the world's leading companies have adopted data mining technology.

Data mining applies to any industry, any sector

Most of the earliest companies that adopted data mining tools were in information-intensive industries, such as large banks, insurance companies, telecommunications companies, and sales. Shouldn't it be so urgent for other industries to adopt data mining?

Zheng Xiaojun believes that data mining tools are not limited to information-intensive enterprises. If the business problems encountered by the enterprise are not simple report statistics and data query to solve the problem, but need to discover and predict the law, data mining can be used. Data mining can and should be used as long as a company has a large database and has a strong desire to improve company management through software technology and to outperform its competitors.

Yang Tao, director of business solutions for SAS in China, said that the successful cases of SAS abroad are completely cross-industry. In addition to banking, telecommunications, insurance, and sales, there are also successful applications in manufacturing, power, transportation, and even government. For example, the Logistics Department of the US Department of Defense, which has won the World Data Mining Award, is a huge department, and its operation was very time-consuming and laborious. Using data mining technology, they effectively used funds, manpower and material resources to support optimized operations.

There is still a misconception that data mining is only used for customer-facing purposes. Indeed, as industries have become more and more customer-centric in recent years, data mining has been heavily used in this regard. But data mining is not only limited to customer relationship management (CRM), but also includes supplier-oriented applications and internal-oriented applications. IBM has a laboratory in New York that studies the optimization of supply chain (SCM), which is achieved through data mining.

Within an organization, data mining can be applied to any business unit of the enterprise. For example, the financial intelligent management of enterprises, how to analyze the return on investment ratio, how to control costs, and how to make budgets. In addition to financial management, data mining is also widely used in human resource management, risk management, quality management, and even in the comprehensive performance appraisal of senior leaders, and in the high-level strategic management of enterprises.

Leading enterprises use data mining to improve their competitiveness

In China, the application of data mining technology has just started, and the real use of data mining tools to support enterprises in making correct production and management decisions

Success stories are few and far between.

First of all, this is because most enterprises still have a vague concept of data mining technology, and they do not have enough understanding of how to improve the competitiveness of enterprises in the fierce market economy environment and the economic benefits it can bring.

The second is because of cultural differences. Western enterprises have formed a scientific management ideology and system, and it is very common to discover laws based on quantitative scientific analysis and guide the production and operation decisions of enterprises. However, most enterprises in China have not yet formed professional management, and are not highly dependent on data. They are not used to finding patterns in data, but prefer to do things by habit.

A typical example is: in order to formulate corresponding market strategies, when it is necessary to classify customers, it is not like Western companies to scientifically analyze the differences in customer behavior patterns, but to subjectively consider the social identity of customers. what a difference. This general idea needs to be improved urgently, otherwise it is not conducive to the development of data mining.

In fact, the technology of data mining itself is not advanced. Zheng Xiaojun said that the techniques used, such as mathematical statistical methods, were proposed in the 1940s and 1960s. However, after all, data mining is a relatively focused technical field, and it is of course difficult to use on the premise that most Chinese enterprises do not pay too much attention to experts. Western companies will let corresponding experts undertake specific operations involving specialized technical fields, which is of course easier to use.