A potential customer data mining method based on rough set and decision tree

Abstract

The present application relates to a kind of potential customer data mining methods based on rough set and decision tree, which comprises: according to existing customer data, decision tree is constructed by rough set;Obtain the customer data to be processed;Determine the corresponding existing customer data of the customer data to be processed;According to the corresponding existing customer data and decision tree, determine whether the customer to be processed is potential customer.The method provided by the present application, according to existing customer data, decision tree is constructed by rough set, according to the corresponding existing customer data and decision tree, determine whether the customer to be processed is potential customer, realizes the expansion of new customer based on existing customer data.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data mining, and particularly relates to a potential customer data mining method based on rough set and decision tree. BACKGROUND

[0002] After the big explosion of data, enterprises can obtain a large number of customer data, and if the new customers can be expanded based on the existing customer data, the total number of customers and customer stickiness can be increased, and the enterprises can obtain more economic benefits and market competition advantages.

[0003] Therefore, how to expand new customers based on existing customer data is a problem that is currently concerned. SUMMARY

[0004] (I) Technical problem to be solved

[0005] In view of the above-mentioned defects and deficiencies of the prior art, the present application provides a potential customer data mining method based on rough set and decision tree.

[0006] (II) Technical scheme

[0007] In order to achieve the above-mentioned purpose, the main technical scheme adopted by the present application includes:

[0008] A potential customer data mining method based on rough set and decision tree, the method comprising:

[0009] S101, constructing a decision tree through rough set according to existing customer data;

[0010] S102, obtaining customer data to be processed;

[0011] S103, determining the existing customer data corresponding to the customer data to be processed;

[0012] S104, determining whether the customer data to be processed is a potential customer according to the corresponding existing customer data and the decision tree.

[0013] Optionally, S101 comprises:

[0014] S101-1, determining a condition attribute set C, a decision attribute set D and a rough set threshold β;

[0015] S101-2, obtaining each condition element in the condition attribute set C in sequence,

[0016] for any condition element c i obtained, wherein i is the condition element identifier in the condition attribute set C, and the following process is executed:

[0017] determining the existing customer data that satisfies c iThe existing customer data is recorded as the first data;

[0018] Determine the membership degree of the first data to each decision element in the decision attribute set D;

[0019] Determine the number n of the first data points with a membership degree greater than β. i And the total number of the first data points, n 1i ;

[0020] Calculate any condition element c i First value and any condition element c i The second value Where N is the total number of existing customer data;

[0021] S101-3, construct a decision tree based on the first value and the second value of each condition element.

[0022] Optionally, the membership degree of the first data to each decision element in the decision attribute set D is determined, including:

[0023] In the decision attribute set D, determine the decision elements involved in each first data point;

[0024] The membership degree of each first data point to each decision element in the decision attribute set D is determined by the following formula:

[0025]

[0026] Where j is the first data identifier, n ju The number of all first data points whose decision element u has the same value as the first data point j, n 2i n is the number of different values ​​of decision element u involved in all the first data, and n4 is the theoretical total number of different values ​​of decision element u.

[0027] Optionally, S101-3 includes:

[0028] The condition element c is calculated using the following formula. i Decision values ​​de for each decision element iu :

[0029] de i =-a i1 log2a i2 ;

[0030] Use the conditional element corresponding to the maximum decision value as the classification feature;

[0031] Decision trees are constructed based on classification features.

[0032] Optionally, S103 includes:

[0033] S103-1, forming a first attribute vector of each existing customer according to existing customer data;

[0034] S103-2, clustering the existing data according to the first attribute vector to obtain at least one cluster;

[0035] S103-3, forming a second attribute vector of the to-be-processed customer according to the to-be-processed customer data;

[0036] S103-4, determining a cluster to which the second attribute vector belongs in the clusters obtained by clustering;

[0037] S103-5, taking the existing customer data corresponding to all existing customers in the cluster to which the second attribute vector belongs as the existing customer data corresponding to the to-be-processed customer data.

[0038] Optionally, S103-4 includes:

[0039] determining a center vector of each cluster in the clusters obtained by clustering;

[0040] determining a weight between each cluster and the second attribute vector;

[0041] calculating a first distance between the second attribute vector and the center vector of each cluster;

[0042] determining, as the cluster to which the second attribute vector belongs, the cluster with the minimum product of the first distance and the corresponding weight.

[0043] Optionally, the determining of the weight between each cluster and the second attribute vector includes:

[0044] for any cluster v, naming all the first attribute vectors in the cluster v as first vectors;

[0045] calculating a distance between each first vector and the center vector in the cluster v;

[0046] selecting n5 first vectors in descending order of the distance, and forming a vector set by using the selected first vectors and the second attribute vector;

[0047] calculating a second distance between any two elements in the vector set;

[0048] taking a standard deviation of all the second distances as the weight between any cluster x and the second attribute vector.

[0049] Optionally, before the selecting of the n5 first vectors in descending order of the distance, the method further includes:

[0050] determining a number n of the first vectors v ;

[0051] Determine the number n of first attribute vectors in each cluster other than any cluster v o wherein o is the identification of each cluster other than any cluster v.

[0052] Determine wherein n6 is the number of families.

[0053] Optionally, S104 comprises:

[0054] Record the corresponding existing customer data as second data, and determine the corresponding decision of each second data in the decision tree.

[0055] Determine the similarity between each second data according to the corresponding decision of each second data in the decision tree.

[0056] According to the highest preset number of second data and the decision tree, determine whether the to-be-processed customer is a potential customer.

[0057] Optionally, determining the similarity between each second data according to the corresponding decision of each second data in the decision tree comprises:

[0058] For any two second data x and y,

[0059] Determine the corresponding decision d x of the second data x in the decision tree.

[0060] Determine the corresponding decision d y of the second data y in the decision tree.

[0061] Calculate the distance D xy between the second data x and the second data y.

[0062] Among all the second data, determine the total amount n x of data corresponding to the decision d x .

[0063] Among all the second data, determine the total amount n y of data corresponding to the decision d y .

[0064] Calculate the similarity between the second data x and the second data y as

[0065] (III) Beneficial effects

[0066] According to the existing customer data, the decision tree is constructed through the rough set; the customer data to be processed is acquired; the existing customer data corresponding to the customer data to be processed is determined; and whether the customer to be processed is a potential customer is determined according to the corresponding existing customer data and the decision tree. The method provided by the application realizes the expansion of new customers based on the existing customer data. BRIEF DESCRIPTION OF DRAWINGS

[0067] Figure 1 A flowchart of the potential customer data mining method based on the rough set and the decision tree is provided for an embodiment of the application. DETAILED DESCRIPTION

[0068] In order to better explain the application and facilitate understanding, the application is described in detail through specific embodiments in combination with the drawings.

[0069] After the data explosion, the enterprise can acquire a large number of customer data, and if the new customers can be expanded based on the existing customer data, the total number of customers and the customer stickiness can be increased, and the enterprise can obtain more economic benefits and market competition advantages. Therefore, how to expand new customers based on the existing customer data is a problem that is currently concerned.

[0070] Based on this, the application provides a potential customer data mining method based on the rough set and the decision tree, which comprises the following steps: according to the existing customer data, the decision tree is constructed through the rough set; the customer data to be processed is acquired; the existing customer data corresponding to the customer data to be processed is determined; and whether the customer to be processed is a potential customer is determined according to the corresponding existing customer data and the decision tree. The method provided by the application realizes the expansion of new customers based on the existing customer data.

[0071] Referring to Figure 1 The implementation process of the potential customer data mining method based on the rough set and the decision tree provided by the embodiment is as follows:

[0072] S101, according to the existing customer data, the decision tree is constructed through the rough set.

[0073] The implementation process of the step is as follows:

[0074] S101-1, the condition attribute set C, the decision attribute set D and the rough set threshold β are determined.

[0075] β is in the range of (0.5, 1], for example, β is 0.85.

[0076] S101-2, sequentially acquire each condition element in the condition attribute set C,

[0077] For any acquired condition element c i , where i is the condition element identifier in the condition attribute set C, the following process is executed:

[0078] 1. Determine the existing customer data that satisfies c i , and record it as the first data.

[0079] The first data is actually existing customer data, but it is the future customer data that satisfies the special relationship (i.e., satisfies c i ).

[0080] For example, the existing customer data is shown in Table 1:

[0081] Table 1

[0082] Existing customer identification Gender Profession Is a customer Existing customer 1 Female Profession 1 Yes Existing customer 2 Female Profession 2 No Existing customer 3 Male Profession 3 No Existing customer 4 Female Profession 1 Yes Existing customer 5 Male Profession 2 No Existing customer 6 Profession 1 Yes

[0083] Where gender and occupation are condition attributes, then C = {c1, c2}, for example, c1 = gender and c2 = occupation. Whether a customer is a decision attribute, then D = {d1}, d1 = whether a customer.

[0084] For example, the selected c i this time is c1, then the existing customer data that satisfies c1 (i.e., gender) is the data of existing customers 1-5, and the data of existing customers 1-5 is recorded as the first data.

[0085] 2. Determine the membership degree of the first data to each decision element in the decision attribute set D.

[0086] The specific implementation process is as follows:

[0087] 1) In the decision attribute set D, determine the decision elements involved by each first data.

[0088] 2) Determine the membership degree of each first data to each decision element in the decision attribute set D by the following formula:

[0089]

[0090] Where j is the first data identifier, n ju is the number of all first data corresponding to the decision element u with the same value as the first data j, n 2i is the number of different values of the decision element u involved by all first data, and n4 is the total number of theoretically different values of the decision element u.

[0091] Still taking the example shown in Table 1 as an example, 1) in the decision attribute set D, the decision element involved by each first data is determined to be dl (i.e. whether it is a customer). 2) The total number of first data n 1i = 5, the number of different values (i.e. "yes" or "no") of the decision element dl (i.e. whether it is a customer) involved by all first data n 2i = 2, the total number of different values (i.e. "yes" or "no" or "unknown") of the decision element dl (i.e. whether it is a customer) n4 = 3, then

[0092] For the first first data (i.e. the data of existing customer 1), the value of the corresponding decision element dl (i.e. whether it is a customer) is "yes", and in all first data, there is also the data of existing customer 4 whose value is "yes", so the number of all first data whose value of the corresponding decision element dl (i.e. whether it is a customer) is the same as that of the first first data (i.e. the data of existing customer 1) n 1u = 2 (i.e. existing customer 1 and existing customer 4),

[0093] For the second first data (i.e. the data of existing customer 2), the value of the corresponding decision element dl (i.e. whether it is a customer) is "no", and in all first data, there is also the data of existing customer 3 and the data of existing customer 5 whose value is "no", so the number of all first data whose value of the corresponding decision element dl (i.e. whether it is a customer) is the same as that of the second first data (i.e. the data of existing customer 2) n 2u = 3 (i.e. existing customer 2, existing customer 3 and existing customer 5),

[0094] For the third first data (i.e. the data of existing customer 3), the value of the corresponding decision element dl (i.e. whether it is a customer) is "no", and in all first data, there is also the data of existing customer 2 and the data of existing customer 5 whose value is "no", so the number of all first data whose value of the corresponding decision element dl (i.e. whether it is a customer) is the same as that of the third first data (i.e. the data of existing customer 3) n 3u = 3 (i.e. existing customer 2, existing customer 3 and existing customer 5),

[0095] For the fourth first data (i.e. the data of existing customer 4), the value of the corresponding decision element dl (i.e. whether it is a customer) is "yes", and in all first data, there is also the data of existing customer 1 whose value is "yes", so the number of all first data whose value of the corresponding decision element dl (i.e. whether it is a customer) is the same as that of the fourth first data (i.e. the data of existing customer 4) n 4u = 2 (i.e. existing customer 1 and existing customer 4),

[0096] For the fifth first data (i.e. the data of the existing customer 5), the value of the decision element d1 (i.e. whether it is a customer) is "no", and in all first data, the values of "no" are also the data of the existing customer 2 and the data of the existing customer 3, therefore, the number n of all first data corresponding to the value of the decision element d1 (i.e. whether it is a customer) of the fifth first data (i.e. the data of the existing customer 5) which is the same as the value 5u = 3 (i.e. the existing customer 2, the existing customer 3 and the existing customer 5),

[0097] 3, determine the number n of the first data whose membership degree is greater than β i , and the total number of first data n 1i .

[0098] In A 1u , A 2u , A 3u , A 4u , A 5u , determine the number greater than β, which is n i . The total number of first data n 1i = 5.

[0099] 4, calculate the first value of any condition element c i and the second value of any condition element c i

[0100] Wherein, N is the total number of existing customer data.

[0101] For example, the total number of existing customer data is 6, i.e. N = 6.

[0102] S101-3, according to the first value of each condition element and the second value of each condition element, construct the decision tree node corresponding to any condition element c i .

[0103] The specific implementation process is as follows:

[0104] 1) Calculate the decision value de iu of the condition element c i for each decision element by the following formula:

[0105] de i = -a i1 log2a i2

[0106] 2) Take the condition element corresponding to the maximum decision value as the classification feature.

[0107] ​​3) Constructing decision tree according to classification features.

[0108] The process of constructing decision tree according to classification features adopts existing process. For example, each node in the constructed decision tree contains four elements: tree.feature is the node split feature; tree.location is the split interval when the split feature is continuous attribute; location is the discrete value of the attribute when the split feature is discrete attribute; tree.child is a structure body storing child nodes, and tree.value is the attribute sample value. For any node, the maximum decision value is the tree.feature split feature of the node, and the existing split method is adopted according to the attribute value. The internal nodes of the classification decision tree are constructed by the existing recursive method until all sample decision values are completely the same or only contain a single conditional attribute, and the recursion is stopped.

[0109] S102, obtaining customer data to be processed.

[0110] S103, determining existing customer data corresponding to the customer data to be processed.

[0111] The implementation process of this step is as follows:

[0112] S103-1, forming a first attribute vector of each existing customer according to the existing customer data.

[0113] S103-2, clustering the existing data according to the first attribute vector to obtain at least one cluster.

[0114] The clustering scheme of this step can adopt existing schemes, such as clustering by k-means method.

[0115] S103-3, forming a second attribute vector of the customer to be processed according to the customer data to be processed.

[0116] S103-4, determining the cluster to which the second attribute vector belongs in the cluster obtained by clustering.

[0117] The specific implementation process is as follows:

[0118] 1. Determining the center vector of each cluster in the cluster obtained by clustering.

[0119] In the clustering process (such as k-means clustering process), it is determined that each first attribute vector belongs to which cluster based on the distance between the first attribute vector and the center vector of each cluster, and the center vector of each cluster is directly obtained in this step.

[0120] 2. Determining the weight between each cluster and the second attribute vector.

[0121] The specific determination scheme is:

[0122] For any cluster v,

[0123] 1) name all the first attribute vectors in any cluster v as first vectors.

[0124] The first vector is the first attribute vector, which is renamed for subsequent convenient description of all the first attribute vectors in any cluster v.

[0125] 2) calculate the distance between each first vector and the center vector in any cluster v.

[0126] The distance calculation method in this step uses the existing method, such as Euclidean distance.

[0127] 3) select n5 first vectors in order from small to large distance, and form a vector set with the selected first vectors and the second attribute vector.

[0128] Wherein, n5 is determined before forming the vector set, and the determination scheme is:

[0129] (1) determine the number of first vectors n v .

[0130] (2) determine the number of first attribute vectors in each cluster other than any cluster v n o ,

[0131] Wherein, o is the identification of each cluster other than any cluster v.

[0132] (3) determine

[0133] Wherein, n6 is the number of families.

[0134] 4) calculate the second distance between any two elements in the vector set.

[0135] The distance calculation method in this step uses the existing method, such as Euclidean distance.

[0136] 5) take the standard deviation of all second distances as the weight between any cluster x and the second attribute vector.

[0137] The standard deviation calculation scheme in this step uses the existing scheme.

[0138] S103-5, take all the existing customer data corresponding to the existing customers in the cluster to which the second attribute vector belongs as the existing customer data corresponding to the to-be-processed customer data.

[0139] Through step 103, the most similar data to the to-be-processed customer data will be found in all the existing customers, and based on the most similar data, it can be known whether the to-be-processed customer is a potential customer.

[0140] S104, determining whether the to-be-processed customer is a potential customer according to the corresponding existing customer data and the decision tree.

[0141] The implementation process of this step is as follows:

[0142] 1. Record the corresponding existing customer data as second data, and determine the corresponding decision of each second data in the decision tree.

[0143] The second data is existing customer data that meets a specific condition (i.e., the corresponding existing customer data obtained in step S103-5).

[0144] 2. Determine the similarity between each second data according to the corresponding decision of each second data in the decision tree.

[0145] The implementation process of this step is as follows:

[0146] For any two second data x and y,

[0147] 1) Determine the corresponding decision d of the second data x in the decision tree x .

[0148] 2) Determine the corresponding decision d of the second data y in the decision tree y .

[0149] 3) Calculate the distance D between the second data x and the second data y xy .

[0150] The distance calculation method of this step uses existing methods, such as Euclidean distance.

[0151] 4) In all second data, determine the total amount of data n corresponding to the decision d x . x

[0152] 5) In all second data, determine the total amount of data n corresponding to the decision d y . y

[0153] 6) Calculate the similarity between the second data x and the second data y as

[0154] 3. Determine whether the to-be-processed customer is a potential customer according to the highest similarity of the preset number of second data and the decision tree.

[0155] The preset number is a pre-set value, which can be an empirical value or a value determined by big data, etc. The determination scheme of the preset number is not limited in this embodiment.

[0156] ​​The most similar data to the to-be-processed customer data can be found through step 103, and the more similar data can be found more accurately in the most similar data based on the similarity in this step, and whether the to-be-processed customer is a potential customer is determined based on the more similar decision value.

[0157] For example, the preset number of second data with the highest similarity are input into the decision tree to obtain a final decision, and whether the to-be-processed customer is a potential customer is determined based on the decision.

[0158] The embodiment provides a potential customer data mining method based on a rough set and a decision tree, which comprises the following steps: constructing a decision tree through a rough set according to existing customer data; obtaining to-be-processed customer data; determining corresponding existing customer data of the to-be-processed customer data; and determining whether the to-be-processed customer is a potential customer according to the corresponding existing customer data and the decision tree. The method provided in the embodiment can construct a decision tree through a rough set according to existing customer data, and determine whether the to-be-processed customer is a potential customer according to the corresponding existing customer data and the decision tree, thereby realizing expansion of new customers based on the existing customer data.

[0159] For a better understanding of the above technical solutions, exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to enable a clearer, more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art.

[0160] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system or a computer program product. Therefore, the present application can be in the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can be in the form of a computer program product implemented on one or more computer usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer usable program code.

[0161] The present application is described with reference to flowcharts and / or block diagrams of methods, devices (systems) and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions.

[0162] It should be noted that the description used in the claims is intended to cover any implementation of the application in which the described structure, material or acts are included. Although the application has been described and illustrated with a certain degree of particularity, it is the patentee's intention that only a limited number of the disclosure be claimed, but that the patent shall be measured in terms of the full breadth and scope of the claims. The term comprising, when used in the specification, including the claims, does not exclude the presence of other elements or steps than those listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The application can be implemented by means of both hardware and software, and any combination thereof. When in the claims a certain prior art is identified that the application is not based on, it is the intention that the claims are not limited to exclude the identified prior art. The word "comprise", "comprising", "comprises" or "comprising" when used in the specification, including the claims, means that other elements can also be present in addition to those listed in the claim. The word "first", "second", "third", etc. are used to identify different components in the description and claims, and are not intended to denote an ordinal number. The components can be identified as part of the component name.

[0163] Furthermore, it is noted that the specification and drawings are, of course, subjected to various modifications and that such modifications are intended to be included within the scope of the application. It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting. It is also noted that, as used in the specification and the appended claims, the singular forms "a", "an" and "the" include plural referents unless the context clearly dictates otherwise. The term "comprising" (and grammatical variations thereof) as used herein is used in the conventional sense in this specification to indicate the presence of true, stated features, integers, steps or components, but not to preclude the presence or addition of one or more other features, integers, steps, components or groups thereof. The term "consisting of" is used herein to refer explicitly to the components of the compositions or methods listed in the claim, and excludes the presence of other, non-recited ingredients. The term "consisting essentially of" is used herein to refer to compositions or methods that do not materially differ in any meaningful way from the compositions or methods recited in the claim. Thus, variations of the described compositions and methods that do not materially affect the basic and novel characteristics of the compositions or methods can be considered equivalents thereof as defined in connection with the claims.

[0164] Although preferred embodiments of the application have been described, those skilled in the art will recognize that modifications and variations of the preferred embodiments can be made without departing from the spirit or scope of the application. Accordingly, it is intended that the application encompass all such modifications and variations as fall within the scope of the claims. It is also to be understood that the terminology used herein is for the purpose of describing the particular embodiments only and is not intended to be limiting. It is further noted that, as used in the specification and the appended claims, the singular forms "a", "an" and "the" include plural referents unless the context clearly dictates otherwise. The term "comprising" (and grammatical variations thereof) as used herein is used in the conventional sense in this specification to indicate the presence of true, stated features, integers, steps or components, but not to preclude the presence or addition of one or more other features, integers, steps, components or groups thereof. The term "consisting of" is used herein to refer explicitly to the components of the compositions or methods listed in the claim, and excludes the presence of other, non-recited ingredients. The term "consisting essentially of" is used herein to refer to compositions or methods that do not materially differ in any meaningful way from the compositions or methods recited in the claim. Thus, variations of the described compositions and methods that do not materially affect the basic and novel characteristics of the compositions or methods may be considered equivalents thereof as defined in connection with the claims.

[0165] It will be apparent to those skilled in the art that various modifications and variations can be made to the present

Claims

1. A potential customer data mining method based on rough set and decision tree, characterized by, The method comprises: S101, constructing a decision tree according to existing customer data through a rough set; S102, obtaining to-be-processed customer data; S103, determining existing customer data corresponding to the to-be-processed customer data; S104, determining whether the to-be-processed customer is a potential customer according to the corresponding existing customer data and the decision tree; The S101 comprises: S101-1, determining a condition attribute set C, a decision attribute set D and a rough set threshold β; S101-2, sequentially obtaining each condition element in the condition attribute set C, For any condition element c acquired i where i is the condition element identification in the condition attribute set C, the following procedure is executed: determining the existing customer data satisfying c i in the existing customer data, and recording the existing customer data as first data; determining the membership degree of the first data belonging to each decision element in the decision attribute set D; determining the number n of first data whose membership degree is greater than β i and the total number n of first data 1i ; Calculate any condition element c i First value and any condition element c i The second value Where N is the total number of existing customer data; S101-3, constructing a decision tree according to the first value of each condition element and the second value of each condition element; The S104 comprises: record the corresponding existing customer data as second data, and determine the corresponding decision of each second data in the decision tree; determining the similarity between each second data according to the corresponding decision of each second data in the decision tree; determining whether the to-be-processed customer is a potential customer according to the preset number of second data with the highest similarity and the decision tree.

2. The method of claim 1, wherein, The determining of the membership degree of the first data belonging to each decision element in the decision attribute set D comprises: determining the decision element involved by each first data in the decision attribute set D; determining the membership degree of each first data belonging to each decision element in the decision attribute set D through the following formula: Wherein, j is the first data identification, n ju The number of all first data corresponding to the same value of decision element u of the first data j, n 2i The number of different values of decision element n involved in all first data, n4 is the total number of different values of decision element u.

3. The method of claim 2, wherein, The S101-3 comprises: The conditional element c is calculated by the following equation i The decision value de for each decision element iu : de i = -a i1 log2a i2 ; taking the condition element corresponding to the maximum decision value as a classification feature; constructing a decision tree according to the classification feature.

4. The method of claim 1, wherein, The S103 comprises: S103-1, forming a first attribute vector of each existing customer according to the existing customer data; S103-2, clustering the existing data according to the first attribute vector to obtain at least one cluster; S103-3, forming a second attribute vector of the to-be-processed customer according to the to-be-processed customer data; S103-4, determining the cluster to which the second attribute vector belongs in the cluster obtained by clustering; S103-5, taking the existing customer data corresponding to all existing customers in the cluster to which the second attribute vector belongs as the existing customer data corresponding to the to-be-processed customer data.

5. The method of claim 4, wherein, The S103-4 comprises: determining a center vector of each cluster in the cluster obtained by clustering; determining the weight between each cluster and the second attribute vector; calculating the first distance between the second attribute vector and the center vector of each cluster; determining the cluster with the minimum product of the first distance and the corresponding weight as the cluster to which the second attribute vector belongs.

6. The method of claim 5, wherein, The determining of the weight between each cluster and the second attribute vector comprises: for any cluster v, naming all first attribute vectors in the any cluster v as first vectors; calculating the distance between each first vector and the center vector in the any cluster v; selecting n5 first vectors in order of small to large distance, forming a vector set with the selected first vectors and the second attribute vector; calculating the second distance between any two elements in the vector set; taking the standard deviation of all second distances as the weight between any cluster x and the second attribute vector.

7. The method of claim 6, wherein, Before the selecting n5 first vectors in order of small to large distance, the method further comprises: determining the number n of first vectors v ; determining the number n of first attribute vectors in each cluster other than said any cluster v o wherein o is an identification of each cluster other than said any cluster v determining wherein n6 is a group number.

8. The method of claim 1, wherein, The similarity between each second data is determined according to the corresponding decision in the decision tree of each second data, comprising: For any two second data x and y, determining a corresponding decision d in the decision tree for the second data x x ; determining a corresponding decision d in the decision tree for the second data y y ; calculating a distance D between the second data x and the second data y xy ; In all second data, determine the corresponding decision d x of the total amount of data n x ; In all second data, determine the corresponding decision d y of the total amount of data n y ; The similarity between the second data x and y is calculated as

Images