CN101620598A - Method for supervising customer grouping - Google Patents
Method for supervising customer grouping Download PDFInfo
- Publication number
- CN101620598A CN101620598A CN200810039888A CN200810039888A CN101620598A CN 101620598 A CN101620598 A CN 101620598A CN 200810039888 A CN200810039888 A CN 200810039888A CN 200810039888 A CN200810039888 A CN 200810039888A CN 101620598 A CN101620598 A CN 101620598A
- Authority
- CN
- China
- Prior art keywords
- data
- information
- supervising
- group
- customer
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a method for supervising customer grouping, which comprises the following steps of: building a data warehouse or a data mart according to the service demand; adopting supervision grouping modeling to subdivide customers according to a customer unified view; and evaluating and releasing a model. Compared with the prior art, the method can accurately subdivide the customers.
Description
Technical field
The present invention relates to data mining technology, particularly relate to a kind of method for supervising customer grouping.
Background technology
Data mining technology relates to a plurality of fields such as database, artificial intelligence and mathematical statistics, pattern-recognition etc. as a chiasma type subject.It from data a large amount of, incomplete, noisy, fuzzy, at random, extract lie in wherein, people are ignorant in advance but be the information of potentially useful and the process of knowledge.Data mining more correctly called after " is excavated knowledge " from data.Also have a lot of and the akin term of this term, as Knowledge Discovery, data analysis, data fusion (data fusion) and decision support etc.The artificial intelligence field custom claims Knowledge Discovery, and the database field custom is called data mining.
Customer grouping is an important commercial problem, enterprise will provide personalized customer service and product to different customers, therefore, in the design of products ﹠ services, must a understanding be arranged to customers, so, often by client's personal information, the attribute of aspects such as behavior property and consumption information hives off to the client for we, to deepen understanding, formulate corresponding marketing strategy to the client.
In general hive off all is unsupervised learning method, and it is exactly some attribute according to object that what is called is hived off, and object is assembled in groups, makes the distance of group interior object as far as possible little, and the distance of object is tried one's best greatly between the group.
But unsupervised hiving off is defective, for example: suppose and to hive off to some, following attribute is arranged: the age, marriage, working condition is when putting these attributes together, crucial problem is the relation of just having no idea to handle between them, machine learning can not be discerned age difference ten years old and wedding difference not, and which is bigger, that is to say that we have no idea to define the standard of a distance at all.
Summary of the invention
Technical matters to be solved by this invention is exactly to provide a kind of method for supervising customer grouping for the defective that overcomes above-mentioned prior art existence.
Purpose of the present invention can be achieved through the following technical solutions: a kind of method for supervising customer grouping, it is characterized in that, and may further comprise the steps:
(1). according to business demand, set up data warehouse or Data Mart;
(2). according to unified view of customers, adopt supervision grouping modeling, the client is segmented;
(3). assessment, issue model.
Described step (1) further comprises:
(11). determine business demand;
(12). prepare data according to business demand;
(13). the data pre-service;
(14). set up each separate service view and unified view of customers.
Described data pre-service is by analyzing the processing of data, exceptional value, the conversion of the processing of null value, the extraction of data and data, process data into can be directly as the high-quality data of Data Mining Tools.
Described separate service view is included in individual essential information, cost information and the behavioural information on the specific professional dimension.
Described unified view of customers comprise the client on all data service dimensions essential information, cost information and behavioural information.
Described step (2) further comprises:
(21). determine the supervision variable;
(22): adopt the supervision variable to come standardized variable;
(23): hive off based on the supervision variable, adopt multiple clustering algorithm to make up a plurality of customer segmentation models;
Described step (3) further comprises:
(31): carry out the evaluation and test of model according to evaluate parameter and accuracy that a plurality of customer segmentation models produce, select the model of current optimum;
(32): the enforcement of model is added;
(33): the foreground of model represents, create two tables of data, be used for group information of depositing and belong to the user profile of this group respectively, deposit the quantity of group number, group, group's feature description information in group's information table, deposit group number, cell-phone number and the personal user information of ownership in the user message table.
Compared with prior art, method of the present invention has defined the index of a distance, has remedied not organizing of the existing scheme of hiving off, and has accurately accomplished client's segmentation.
Description of drawings
Fig. 1 is a process flow diagram of the present invention.
Embodiment
The invention will be further described below in conjunction with accompanying drawing.
As shown in Figure 1, a kind of method for supervising customer grouping may further comprise the steps:
(1). according to business demand, set up data warehouse or Data Mart;
(2). according to unified view of customers, adopt supervision grouping modeling, the client is segmented;
(3). assessment, issue model.
Described step (1) further comprises:
Determine business demand; Prepare data according to business demand; The data pre-service; Set up each separate service view and unified view of customers;
Described data pre-service is by analyzing the processing of data, exceptional value, the conversion of the processing of null value, the extraction of data and data, process data into can be directly as the high-quality data of Data Mining Tools; Described separate service view is included in individual essential information, cost information and the behavioural information on the specific professional dimension; Described unified view of customers comprise the client on all data service dimensions essential information, cost information and behavioural information;
Described step (2) further comprises: determine the supervision variable; Adopt the supervision variable to come standardized variable; Hive off based on the supervision variable, adopt multiple clustering algorithm to make up a plurality of customer segmentation models;
Described step (3) further comprises: carry out the evaluation and test of model according to evaluate parameter and accuracy that a plurality of customer segmentation models produce, select the model of current optimum; The enforcement of model is added; The foreground of model represents, create two tables of data, be used for group information of depositing and belong to the user profile of this group respectively, deposit the quantity of group number, group, group's feature description information in group's information table, deposit group number, cell-phone number and the personal user information of ownership in the user message table.
Detailed process of the present invention:
Determine business demand:
To realize the understanding of data mining project theme in this step, need effectively to link up with the client, the problem and the indeterminable problem that allow the client know that data mining can solve, with the document that is organized into of the theme demand system of data mining, so that the design of the general frame of next step data mining project;
Data are prepared:
This step will propose required data according to the theme demand of data mining, in the practical project project, the needed data of data mining are not ready for the data mining personnel at the very start, because the needed data of data mining are to be distributed in the different systems sometimes, and might be the data source of isomery, to prepare data according to the demand of data mining.When this step is finished, should provide based on required data with a kind of database.
The data pre-service:
This step will be carried out pre-service to data, comprises data are analyzed, the processing of exceptional value, the processing of null value, the extraction of data and the conversion of data, thus process data into can be directly as the high-quality data of Data Mining Tools, so that the modeling of next step data mining.In this step, should null value be replaced with satisfactory value with surpassing the data deletion of certain limit, will be the desired data layout of modeling with data conversion in case of necessity.When this step finishes, should provide the high-quality that can be directly used in modeling data source.
Set up Data Mart or data warehouse:
This step will be finished and set up Data Mart or data warehouse, comprising setting up each separate service view and setting up unified view of customers.Wherein the separate service view is for comprising individual essential information, cost information and behavioural information on specific professional dimension.
Unified view of customers is client's representing on all data service dimensions, comprises client's essential information, cost information and behavioural information.
Determine the supervision variable:
This step will determine how to choose the supervision variable by deeply understanding demand, the supervision variable is in the middle of the data mining project of reality, determine that according to concrete problem in the trade type data centralization, the choosing of supervision variable chosen from the data message of transaction data often.
Allow the supervision variable come standardized variable:
This step will be come standardized variable by the supervision variable, in the trade type data centralization, the kind of sales-entity such as commodity often have thousands of in, the result that so many kind produces at minute group time is difficult to explain often, employing comes the method for standardized variable can optimize this grouping method based on the supervision variable, for example, commodity are divided into the classification of bigger one decks such as food, hive off again on this basis.
Hive off based on the supervision variable
This step will adopt clustering algorithm to come data are carried out modeling on the basis of supervision variable, and the present invention adopts multiple cluster node to carry out modeling, therefrom chooses the issue that optimization model carries out model.
The evaluation and test of model algorithm
The present invention adopts multiple clustering algorithm to carry out modeling analysis, and this step will be carried out the evaluation and test of model to evaluate parameter and accuracy that model produces, and this step will be selected the model of current optimum.
The enforcement of model is added
This step will be carried out the popularization of model on the optimization model basis of choosing, the rule set business personnel who produces for model will make an explanation, see whether rule has practical meaning, the characteristic information that then rule is changed into customer group adds the special part of describing of group of each group to.
The foreground of model represents
This step wants the foreground of implementation model to represent, and creates two tables of data, and Table A and table B are used for group information of depositing and belong to the user profile of this group respectively.The group deposits group number, group's quantity, group's feature description information in the information table.Deposit in the user message table ownership group number, cell-phone number and personal user information.
Embodiment
The present invention is that example illustrates with the cell phone television services:
The program of cell phone TV have thousands of in, directly do cluster on this basis and hive off, the result of modeling often is difficult to explain, so we adopt based on the method for supervision variable to come at first the program field to be carried out standardization.
Get the program that program watches that rank is preceding 650, carry out abstract classification on this basis, be divided into into 21 program categories based on program category.
As shown in table 1 is the concrete field information of 21 program categories:
Table 1
We adopt Data Mining Tools to carry out modeling analysis with the cluster node, and through the assessment to model, we adopt the K-means node to carry out modeling.The result of modeling and be analyzed as follows shown in:
The video display group: frequency 4.9% target mean 21.9% is characterized as movie and video programs to watch number of times the highest for being worth maximum colony, and music class and the comprehensive number of times of watching are seen sports cast hardly greater than average, form sharp contrast (repulsion physical culture) with the physical culture group;
The healthy living group: frequency 5% target mean 19.8%, sports cast is watched the highest class of number of times, and other program viewing frequency is not repelled movie and video programs all greater than mean value;
The one-tenth crowd: the favor program kind of frequency 6.5% target mean 11.6% is: the beauty, and both sexes, stars: other class program viewing time number average is less than mean value;
The group keeps up with current affairs: frequency 7.5% target mean 10.3% is liked comprehensively, legal system, and the news category program, video display beauty class program viewing number of times is less than mean value.
Claims (7)
1. a method for supervising customer grouping is characterized in that, may further comprise the steps:
(1). according to business demand, set up data warehouse or Data Mart;
(2). according to unified view of customers, adopt supervision grouping modeling, the client is segmented;
(3). assessment, issue model.
2. a kind of method for supervising customer grouping according to claim 1 is characterized in that, described step (1) further comprises:
(11). determine business demand;
(12). prepare data according to business demand;
(13). the data pre-service;
(14). set up each separate service view and unified view of customers.
3. a kind of method for supervising customer grouping according to claim 2, it is characterized in that, described data pre-service is by analyzing the processing of data, exceptional value, the conversion of the processing of null value, the extraction of data and data, process data into can be directly as the high-quality data of Data Mining Tools.
4. a kind of method for supervising customer grouping according to claim 2 is characterized in that, described separate service view is included in individual essential information, cost information and the behavioural information on the specific professional dimension.
5. a kind of method for supervising customer grouping according to claim 2 is characterized in that, described unified view of customers comprise the client on all data service dimensions essential information, cost information and behavioural information.
6. a kind of method for supervising customer grouping according to claim 1 is characterized in that, described step (2) further comprises:
(21). determine the supervision variable;
(22): adopt the supervision variable to come standardized variable;
(23): hive off based on the supervision variable, adopt multiple clustering algorithm to make up a plurality of customer segmentation models;
7. a kind of method for supervising customer grouping according to claim 1 is characterized in that, described step (3) further comprises:
(31): carry out the evaluation and test of model according to evaluate parameter and accuracy that a plurality of customer segmentation models produce, select the model of current optimum;
(32): the enforcement of model is added;
(33): the foreground of model represents, create two tables of data, be used for group information of depositing and belong to the user profile of this group respectively, deposit the quantity of group number, group, group's feature description information in group's information table, deposit group number, cell-phone number and the personal user information of ownership in the user message table.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200810039888A CN101620598A (en) | 2008-06-30 | 2008-06-30 | Method for supervising customer grouping |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200810039888A CN101620598A (en) | 2008-06-30 | 2008-06-30 | Method for supervising customer grouping |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101620598A true CN101620598A (en) | 2010-01-06 |
Family
ID=41513840
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN200810039888A Pending CN101620598A (en) | 2008-06-30 | 2008-06-30 | Method for supervising customer grouping |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101620598A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106204931A (en) * | 2016-06-28 | 2016-12-07 | 国网山东省电力公司济南市历城区供电公司 | A kind of keyholed back plate that takes the most in real time manages system and method |
| CN107274066A (en) * | 2017-05-19 | 2017-10-20 | 浙江大学 | A kind of shared traffic Customer Value Analysis method based on LRFMD models |
-
2008
- 2008-06-30 CN CN200810039888A patent/CN101620598A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106204931A (en) * | 2016-06-28 | 2016-12-07 | 国网山东省电力公司济南市历城区供电公司 | A kind of keyholed back plate that takes the most in real time manages system and method |
| CN107274066A (en) * | 2017-05-19 | 2017-10-20 | 浙江大学 | A kind of shared traffic Customer Value Analysis method based on LRFMD models |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111444236B (en) | Mobile terminal user portrait construction method and system based on big data | |
| CN102591942B (en) | Method and device for automatic application recommendation | |
| CN106570718B (en) | Information delivery method and delivery system | |
| CN103686382A (en) | Program recommendation method | |
| CN107894998B (en) | Video recommendation method and device | |
| CN103034508A (en) | Software recommending method and software recommending system | |
| CN103106285A (en) | Recommendation algorithm based on information security professional social network platform | |
| CN105139021B (en) | A kind of method and system for realizing TV user Fast Classification based on rough set theory | |
| CN103488788A (en) | Method and device for recommending applications automatically | |
| CN105095219A (en) | Weibo recommendation method and terminal | |
| CN105744370A (en) | Radio and television system based on group viewing behaviors and personalized program recommendation method thereof | |
| CN111489201A (en) | Method, device and storage medium for analyzing customer value | |
| CN106960354A (en) | Method and device is recommended in a kind of precision based on customer life cycle | |
| CN108765052A (en) | Electric business recommendation/method for pushing and device, storage medium and computing device | |
| CN101452478A (en) | Information processing device and method, program, and recording medium | |
| CN108447064A (en) | A kind of image processing method and device | |
| CN104484435A (en) | Method for cross-over analysis of user behavior | |
| CN117436679A (en) | Meta-universe resource matching method and system | |
| CN105468740B (en) | A kind of storage, analysis method and the device of game player's data | |
| Vaca Ruiz et al. | Modeling dynamics of attention in social media with user efficiency | |
| CN101620598A (en) | Method for supervising customer grouping | |
| CN101661507A (en) | Method for merging data and system thereof | |
| KR101664358B1 (en) | Apparatus and method for multi-dimensional customer's clustering using topic analysis | |
| CN114385899A (en) | User group accurate identification system and method based on big data analysis | |
| CN114092123A (en) | Satisfaction intelligent analysis system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20100106 |