This paper analyzes the data based on a large number of data in enterprise network marketing, and processes the data according to the typical clustering analysis method in data mining technology. Taking an online marketing company as an example, this paper makes a cluster analysis of customer information, and obtains some valuable information, which will give some support to the decision-making of enterprise marketing strategy.

Paper Keywords: clustering, network marketing, strategy, customer relationship

Preface of 0

With the rapid development of modern science and technology, especially the application and development of the Internet, enterprises must promote their products through the Internet to enhance their competitiveness. Customers are very important and valuable resources. Now, how to better mine the valuable information of customers from the database, better cultivate and manage the relationship with valuable customers, abandon those customers who are unprofitable, have no development prospects and have high marketing expenses, give different policies to customers with different values, and formulate personalized marketing strategies at the same time can ensure the survival and development of enterprises. For all this, data mining is undoubtedly one of the effective and good methods. Taking an online marketing company as an example, this paper puts forward a set of operable methods to evaluate customer value, and then clusters customer information by using the common and commonly used clustering analysis algorithm in data mining technology, so as to obtain very important information and provide decision-making basis for enterprise online marketing.

1 cluster analysis

Clustering is a very important part of data mining technology, and it is also a very key part of current data mining technology. The meaning of clustering is the process of automatically classifying physical or logical data objects and finally dividing data objects into multiple classes or clusters. For clustering results, data objects should have the greatest similarity in the same classification and the smallest similarity in different classes. The practical significance of clustering is that data can be automatically classified according to a certain relationship, and the number of categories of all data objects is not known in advance, and a classification result is finally obtained through the processing of algorithms for application. For example, in the field of market research, especially for online marketing enterprises or websites, from a large number of network data analysis and clustering, it can be said that customers are divided into different categories, and personalized marketing methods are carried out according to the different purchasing power and hobbies of these categories to improve the economic benefits of enterprises. At present, most researchers are committed to improving and perfecting the clustering analysis algorithm, so as to improve the efficiency of clustering analysis. The famous algorithms are CLARANS, BRICH, DBSCAN, CURE, STING, CLIGUE and WaveCluster.

Application of cluster analysis in enterprise customer asset management.

Now, an e-commerce company analyzes. The customers of this e-commerce company are distributed all over the country and some foreign regions. Now only 10 representative big customers are listed: Jilin, Heilongjiang, Shandong, Jiangsu, Zhejiang, Anhui, Hunan, Myanmar, India, South Africa and so on. The purpose of data mining is to find some similarities from customers. Before processing these customer data, cluster analysis should be used to study whether this 10 customer has some * * * similarity, so that enterprises can give different countermeasures for different types of customers. At first, the company adopts the method of expert scoring, and also collects the opinions of local sales specialists through online questionnaires and interviews, and then synthesizes the data. Finally,

Then the specific implementation of cluster analysis can be divided into five steps:

Step 1: First, build a hierarchy for each indicator, in which the 10 major customer evaluated is the scheme level, the customer value is processed at the target level, and each indicator is the criterion level. According to this hierarchical structure, the structural diagram of each index in the customer relationship evaluation system is constructed, as shown in Figure 2- 1;

From the data, we can see two situations: one is Myanmar and South Africa. As can be seen from the data, the current value of such customers is very small, but they have great hidden value. It is bound that their growth will bring rich material benefits to the enterprise one day, so customers with development potential should take measures to stimulate their potential; Second, Anhui, India and other customers, although the current value of these customers is very small from the data point of view, but from their geographical location and economic situation, the value is great. For such customers, enterprises should take flexible measures to stimulate their purchasing power and promote the continuous development of such customers;

The second category is "maintenance" customers, who will continue to provide profits for enterprises, such as Heilongjiang and Jiangsu. According to the analysis of past transaction records, their customers are of great value at present, but they have no development potential, or in some cases they will often shrink. At present, such customers will bring huge profits to enterprises, but in the long run, they are not the main source of profits, and in some cases they will be lost. Will be lost because of the intervention of competitors from other enterprises. Therefore, on the one hand, enterprises should maintain good relations with such customers and maintain stable customer relations, on the other hand, they should take some marketing measures to stimulate the consumption of such customers and provide some personalized services and strategies;

The third kind of "eliminated" customers, like chicken ribs, are of little significance to the present and future of the enterprise. At present, the sales share is small, the marketing cost of enterprises is still high, and the annual profit rate is very low. According to the analysis, this kind of customers include Zhejiang, Hunan and Jilin, and they have no long-term development trend, so the strategy adopted by enterprises is to gradually give up them after fully excavating the current value they bring to enterprises;

The fourth category is "VIP" customers, who are the main source of economic profits of enterprises and can be said to be the guarantee of enterprise survival to some extent. He is an important customer related to the life and death of the enterprise. From the data point of view, Shandong is such an enterprise VIP customer, and his current value and potential value are great. Enterprises must take it seriously, carefully care for the relationship with such customers, the relationship with key figures of customer enterprises, and strengthen communication with such customers. For VIP customers, enterprises should carry out one-to-one marketing strategy, communicate with customers' needs, meet customers' needs as much as possible, and give some special policies to strengthen their relationship. Enhance customers' loyalty and satisfaction with the enterprise from different angles. Based on these important information, enterprises can adopt appropriate sales strategies for different customers.

3 abstract

In short, the enterprise first evaluates the customer's value from all directions and angles, then quantifies the analysis results and carries out data mining. Through cluster analysis, customers are subdivided to provide personalized services for different types of customers.

;