A method for realizing multi-dimensional and fine-grained customer portrait model for power marketing informatization

Through multi-dimensional correlation and fine-grained methods, collect and analyze power marketing customer information, build an information customer portrait model, solve the problems of information redundancy and mismatch in power enterprise marketing, improve the effect of power marketing, and form a more accurate customer portrait .

CN114255070BActive Publication Date: 2024-11-15GUANGZHOU POWER SUPPLY BUREAU GUANGDONG POWER GRID CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202111479762.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2021-12-03
Publication Date
2024-11-15
Estimated Expiration
2041-12-03

AI Technical Summary

Technical Problem

There is a marketing distortion in power companies in marketing, which makes it difficult to meet user needs. The existing customer portrait methods cannot effectively analyze and meet the characteristics of electricity consumption, affecting marketing effects.

Method used

The multi-dimensional correlation fine-grained method is used to collect power marketing customer information, which is divided into user needs, electricity consumption output and electricity consumption distribution. The information customer portrait model is constructed through preprocessing and factor analysis, which enhances the correlation of information variables, reduces feature dimensions, and forms Accurate customer portrait.

Benefits of technology

It improves the fit of customer portrait construction, enhances the effect of power marketing, solves the problems of information redundancy and mismatch, and forms a more accurate customer portrait of power marketing information.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114255070B_ABST
    Figure CN114255070B_ABST
Patent Text Reader

Abstract

The present invention relates to a method for realizing a multi-dimensional associated fine-grained customer portrait model for power marketing informatization, and particularly to the technical field of power marketing, comprising the following steps: collecting power marketing customer information; collecting power demand; collecting power output; collecting power distribution; a preprocessing step; analyzing the correlation of customer information; and constructing a power marketing informatization customer portrait model. The present invention has the technical effect of ensuring the fitting degree of customer portrait construction and improving the power marketing effect.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of power marketing, and in particular to a method for realizing a multi-dimensional associated fine-grained power marketing information customer portrait model. Background Art

[0002] At present, power companies are affected by the overall environment and have already experienced marketing distortion, which has made it difficult for current power companies to meet the power needs of users, forcing them to seek new development models and vigorously develop power marketing business. As various companies pay more attention to the marketing department, relevant personnel have developed customer portraits, using the basic information of customers as the basis of portraits, using specific and abstract information tags to describe customers, in order to achieve the purpose of analyzing customer needs and provide customers with relevant services. In the era of big data, this marketing method has enabled some companies to achieve good marketing results. To this end, the customer portrait marketing model is applied to the power marketing center in order to improve the power marketing effect. Summary of the invention

[0003] The purpose of the present invention is to provide a method for realizing a multi-dimensional associated fine-grained power marketing information customer portrait model, which has the effect of ensuring the fitting degree of customer portrait construction and improving the power marketing effect.

[0004] The above-mentioned object of the present invention is achieved through the following technical solutions:

[0005] A method for realizing a multi-dimensional and fine-grained power marketing information customer portrait model comprises the following steps:

[0006] S1. Collecting power marketing customer information: Divide the power information into three parts: user demand and power output and power distribution;

[0007] S2. Collecting electricity demand: Assuming that the characteristic domain component of the power marketing customer is p, the customer's electricity intensity in the gradient direction is d, and the time for collecting customer information is t, the electricity demand Q(x,t) is:

[0008]

[0009] In the formula, δ represents the operator of power marketing customer information; u represents the characteristic component of the customer; Indicates partial derivative; x indicates the gradient direction;

[0010] S3. Collecting power output: According to formula (1), obtain the customer's power demand, track and identify the customer information, obtain the customer's basic information, and collect the customer's power output information. Assuming that the customer's feature domain information is H, the customer's total power consumption is O, and the customer information is U, the power output information D(l) is:

[0011] D(l)=O·H(k-1)+U(k) (2)

[0012] In the formula, k represents the customer information dimension;

[0013] S4. Collecting electricity consumption distribution: Since this study divides the power marketing customer information into three parts, the customer portrait information collected in the first two parts is scattered and cannot form effective power marketing customer portrait information. Therefore, the third part of information collection is carried out. Assuming that the customer electricity consumption distribution feature is r, the output of the customer electricity consumption information in the gradient direction is o, then the customer portrait electricity consumption distribution information value C is:

[0014] C=ox+r (3)

[0015] Combining the above three formulas, we can track the collected power marketing customer portraits. However, since this study collected power marketing customer portraits three times, there may be a problem of mismatch in collected information. Therefore, we construct a related function f to detect the matching of customer portraits. The calculation formula is as follows:

[0016]

[0017] In the formula, ε represents the correlation coefficient; T represents the customer information collection period; dt represents the differential with respect to time t; Represents the mean value of information collected in the gradient direction; Indicates the correlation difference of customer information in the gradient direction;

[0018] S5, pre-processing step;

[0019] S6. Analyze the relevance of customer information;

[0020] S7. Construct a customer portrait model for power marketing informationization.

[0021] Preferably, the pre-processing step S5 comprises:

[0022] S51. When there is missing data in the data source, the data is supplemented or removed according to the amount of missing data; when there is a format error in the data in the data source, the data is removed or modified.

[0023] S52. Since the data collection sources are different during the data collection process, there may be multiple names for the same data. Therefore, the data coding is unified.

[0024] S53. Determine whether there is any logical error in the coding.

[0025] S54. When there are logical errors in the coding, the following steps need to be followed: (1) Remove duplicate data in the data source; (2) Remove fabricated (irrational) data in the data source, such as age over 100 years old, phone numbers with consecutive digits, etc.; (3) Adjust data conflicts in the data source, such as one person with different genders, etc.

[0026] S55. When there are no logical errors in the coding, there are problems such as information duplication and conflict in the data source, which reduces the correlation between the data and causes the data to deviate from the actual situation. Therefore, it is necessary to use multi-dimensional correlation and fine-grained analysis of customer information correlation.

[0027] Preferably, the analyzing the relevance of customer information S6 includes:

[0028] S61, support constraint: Assume that the main customer of power marketing collected is ψ, and its corresponding attribute value is At this time, the attribute value corresponding to the power marketing main customer It can be called an item set. Since it is impossible for only one power marketing customer to be collected, n power marketing customers can be collected in one data collection cycle T, that is, n item sets can be formed. For this reason, it is called an n-item set. At this time, with ψ and As a reference, there is Based on this, The support degree is expressed as The number of occurrences in the data source, which can also be recorded as the number of occurrences of the item set Then we have:

[0029]

[0030] In the formula, Express The support of; O(m) represents the number of occurrences of all item sets;

[0031] S62, Confidence Constraint: To avoid repeatedly adding the same attribute to the same customer profile, calculate the attribute value The probability of appearing on the power marketing main customer ψ is:

[0032]

[0033] In the formula, Indicates confidence; The power marketing client ψ contains the attributes probability; represents the power marketing subject customer ψ and attributes The probability of simultaneous occurrence; P(ψ) represents the probability of repeated occurrence of the power marketing subject customer ψ in the data source.

[0034] Preferably, the step of constructing the power marketing information customer portrait model S7 includes:

[0035] Based on the customer information collected this time, the factor analysis method will be used to enhance the correlation of information variables, reduce the number of factors through linear combination, and reduce the characteristic dimensions of customer information.

[0036] Assume that the attribute value set of the power marketing subject customer ψ is is a g-dimensional vector, the attribute value The mean vector is χ=(χ1,χ2,…,χ g ), the set of common attribute factors existing in the power marketing subject customer ψ is Among them, h represents the number of common factors, and the set of special factors is λ=(λ1,λ2,…,λ I ), where I represents the number of special factors. Then:

[0037]

[0038] In the formula, α represents the factor loading matrix. According to formula (7), the specific number of customer attribute value factors of power marketing entities can be obtained. Based on the attribute value factors, combined with the CS algorithm and the fuzzy mean algorithm, the customer portrait model of power marketing informationization is constructed.

[0039] Assuming that the number of sample variables of the attribute values ​​of the power marketing subject customers is c, according to the attribute values ​​corresponding to the power marketing subject customers J cluster centers are formed. When the number of samples is c, the calculation matrix of the power marketing subject customer ψ corresponding to the attribute value is as follows:

[0040]

[0041] In the formula, represents the center vector of the cth attribute value sample variable of the Jth cluster center. According to the clustering method shown in formula (8), it is necessary to determine the fitness δ of each attribute value γ , and its calculation formula is as follows:

[0042]

[0043] In the formula, γ represents the fuzzy factor; L represents ψ i The distance to the jth cluster center; ψ i represents the power marketing customer sample; θ ij Represents ψ iThe membership degree of the jth cluster center. Combining the above formula, when the attribute value partition result obtained by the corresponding power marketing subject customer ψ satisfies the fitness of each attribute value δ γ If , the constructed customer portrait model is established. Otherwise, it is necessary to recalculate the customer information attribute value factor until the attribute value division result obtained satisfies the calculated fitness value. The loop is stopped and the final constructed customer portrait model is output.

[0044] In summary, the beneficial effects of the present invention are:

[0045] To study the customer portrait of power marketing, it is necessary to collect the information of power marketing customers first. However, the collected information is redundant. Therefore, the information is preprocessed, and the correlation of customer information is analyzed by multi-dimensional correlation and fine-grained analysis to establish a customer portrait model of power marketing informatization and form a customer portrait of power marketing informatization. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is the relevance analysis process for this customer. DETAILED DESCRIPTION

[0047] In order to facilitate the understanding of those skilled in the art, the present invention is further described below in conjunction with the embodiments and drawings, and the contents mentioned in the embodiments are not intended to limit the present invention. The present invention is described in detail below in conjunction with the drawings.

[0048] The present invention is further described in detail below in conjunction with the accompanying drawings.

[0049] See also Figure 1 , a customer portrait model for power marketing informatization based on multi-dimensional association and fine-grainedness, including the following steps:

[0050] S1. Collecting power marketing customer information: Divide the power information into three parts: user demand and power output and power distribution;

[0051] S2. Collecting electricity demand: Assuming that the characteristic domain component of the power marketing customer is p, the customer's electricity intensity in the gradient direction is d, and the time for collecting customer information is t, the electricity demand Q(x,t) is:

[0052]

[0053] In the formula, δ represents the operator of power marketing customer information; u represents the characteristic component of the customer; Indicates partial derivative; x indicates the gradient direction;

[0054] S3. Collecting power output: According to formula (1), obtain the customer's power demand, track and identify the customer information, obtain the customer's basic information, and collect the customer's power output information. Assuming that the customer's feature domain information is H, the customer's total power consumption is O, and the customer information is U, the power output information D(l) is:

[0055] D(l)=O·H(k-1)+U(k) (2)

[0056] In the formula, k represents the customer information dimension;

[0057] S4. Collecting electricity consumption distribution: Since this study divides the power marketing customer information into three parts, the customer portrait information collected in the first two parts is scattered and cannot form effective power marketing customer portrait information. Therefore, the third part of information collection is carried out. Assuming that the customer electricity consumption distribution feature is r, the output of the customer electricity consumption information in the gradient direction is o, then the customer portrait electricity consumption distribution information value C is:

[0058] C=ox+r (3)

[0059] Combining the above three formulas, we can track the collected power marketing customer portraits. However, since this study collected power marketing customer portraits three times, there may be a problem of mismatch in collected information. Therefore, we construct a related function f to detect the matching of customer portraits. The calculation formula is as follows:

[0060]

[0061] In the formula, ε represents the correlation coefficient; T represents the customer information collection period; dt represents the differential with respect to time t; Represents the mean value of information collected in the gradient direction; Indicates the correlation difference of customer information in the gradient direction;

[0062] S5, pre-processing step;

[0063] S6. Analyze the relevance of customer information;

[0064] S7. Construct a customer portrait model for power marketing informationization.

[0065] Working principle: To study the customer portrait of power marketing, it is necessary to collect the customer information of power marketing first. However, the collected information has redundancy problems. Therefore, the information is pre-processed, and the correlation of customer information is analyzed by multi-dimensional correlation and fine-grained analysis to establish the customer portrait model of power marketing informatization and form the customer portrait of power marketing informatization.

[0066] The pre-processing step S5 comprises:

[0067] S51. When there is missing data in the data source, the data is supplemented or removed according to the amount of missing data; when there is a format error in the data in the data source, the data is removed or modified.

[0068] S52. Since the data collection sources are different during the data collection process, there may be multiple names for the same data. Therefore, the data coding is unified.

[0069] S53. Determine whether there is any logical error in the coding.

[0070] S54. When there are logical errors in the coding, the following steps need to be followed: (1) Remove duplicate data in the data source; (2) Remove fabricated (irrational) data in the data source, such as age over 100 years old, phone numbers with consecutive digits, etc.; (3) Adjust data conflicts in the data source, such as one person with different genders, etc.

[0071] S55. When there are no logical errors in the coding, there are problems such as information duplication and conflict in the data source, which reduces the correlation between the data and causes the data to deviate from the actual situation. Therefore, it is necessary to use multi-dimensional correlation and fine-grained analysis of customer information correlation.

[0072] Based on the above steps, the data preprocessing process is completed. However, in step 5, it is necessary to analyze the relevance of customer information and adopt multi-dimensional correlation granularity to improve the fit of customer portrait model construction.

[0073] The analyzing the relevance of customer information S6 includes:

[0074] S61, support constraint: Assume that the main customer of power marketing collected is ψ, and its corresponding attribute value is At this time, the attribute value corresponding to the power marketing main customer It can be called an item set. Since it is impossible for only one power marketing customer to be collected, n power marketing customers can be collected in one data collection cycle T, that is, n item sets can be formed. For this reason, it is called an n-item set. At this time, with ψ and As a reference, there is Based on this, The support degree is expressed as The number of occurrences in the data source, which can also be recorded as the number of occurrences of the item set Then we have:

[0075]

[0076] In the formula, Express The support of; O(m) represents the number of occurrences of all item sets;

[0077] S62, Confidence Constraint: To avoid repeatedly adding the same attribute to the same customer profile, calculate the attribute value The probability of appearing on the power marketing main customer ψ is:

[0078]

[0079] In the formula, Indicates confidence level; The power marketing client ψ contains the attributes probability; represents the power marketing subject customer ψ and attributes The probability of simultaneous occurrence; P(ψ) represents the probability of repeated occurrence of the power marketing subject customer ψ in the data source.

[0080] The steps of constructing the power marketing information customer portrait model S7 include:

[0081] Based on the customer information collected this time, the factor analysis method will be used to enhance the correlation of information variables, reduce the number of factors through linear combination, and reduce the characteristic dimensions of customer information.

[0082] Assume that the attribute value set of the power marketing subject customer ψ is is a g-dimensional vector, the attribute value The mean vector is χ=(χ1,χ2,…,χ g ), the set of common attribute factors existing in the power marketing subject customer ψ is Among them, h represents the number of common factors, and the set of special factors is λ=(λ1,λ2,…,λ I ), where I represents the number of special factors. Then:

[0083]

[0084] In the formula, α represents the factor loading matrix. According to formula (7), the specific number of customer attribute value factors of power marketing entities can be obtained. Based on the attribute value factors, combined with the CS algorithm and the fuzzy mean algorithm, the customer portrait model of power marketing informationization is constructed.

[0085] Assuming that the number of sample variables of the attribute values ​​of the power marketing subject customers is c, according to the attribute values ​​corresponding to the power marketing subject customers J cluster centers are formed. When the number of samples is c, the calculation matrix of the power marketing subject customer ψ corresponding to the attribute value is as follows:

[0086]

[0087] In the formula, represents the center vector of the cth attribute value sample variable of the Jth cluster center. According to the clustering method shown in formula (8), it is necessary to determine the fitness δ of each attribute value γ , and its calculation formula is as follows:

[0088]

[0089] In the formula, γ represents the fuzzy factor; L represents ψ i The distance to the jth cluster center; ψ i represents the power marketing customer sample; θ ij Represents ψ i The membership degree of the jth cluster center. Combining the above formula, when the attribute value partition result obtained by the corresponding power marketing subject customer ψ satisfies the fitness of each attribute value δ γ If , the constructed customer portrait model is established. Otherwise, it is necessary to recalculate the customer information attribute value factor until the attribute value division result obtained satisfies the calculated fitness value. The loop is stopped and the final constructed customer portrait model is output.

[0090] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Although the present invention is disclosed as a preferred embodiment as above, it is not used to limit the present invention. Any technician familiar with the profession can make some changes or modifications to equivalent embodiments of equivalent changes by using the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any simple modification, equivalent changes and modifications made to the above embodiments according to the technology of the present invention are within the scope of the technical solution of the present invention without departing from the content of the technical solution of the present invention.

Claims

1. A method for realizing a multi-dimensional and fine-grained power marketing information customer portrait model, characterized in that: The following steps are involved: S1. Collecting power marketing customer information: Divide the power information into three parts: user demand and power output, and power distribution; S2. Collecting electricity demand: Assuming that the characteristic domain component of the power marketing customer is p, the customer's electricity intensity in the gradient direction is d, and the time for collecting customer information is t, the electricity demand Q(x,t) is: In the formula, δ represents the operator of power marketing customer information; u represents the characteristic component of the customer; Indicates partial derivative; x indicates the gradient direction; S3. Collecting power output: According to formula (1), the customer's power demand is obtained, the customer information is tracked and identified, the customer's basic information is obtained, and the customer's power output information is collected; assuming that the customer's feature domain information is H, the customer's total power consumption is O, and the customer information is U, the power output information D(l) is: D(l)=O×H(k-1)+U(k) (2) In the formula, k represents the customer information dimension; S4. Collecting electricity consumption distribution: Since this study divides the power marketing customer information into three parts, and in the first two parts, the collected customer portrait information is scattered and cannot form effective power marketing customer portrait information; therefore, the third part of information collection is carried out. Assuming that the customer electricity consumption distribution feature is r, the output of the customer electricity consumption information in the gradient direction is o, then the customer portrait electricity consumption distribution information value C is: C=ox+r(3) Combining the above three formulas, we can track the collected power marketing customer portraits; however, since this study collected power marketing customer portraits three times, there may be a problem of mismatch in collected information. Therefore, we construct a related function f to detect the matching of customer portraits. The calculation formula is as follows: In the formula, ε represents the correlation coefficient; T represents the customer information collection period; dt represents the differential with respect to time t; Represents the mean value of information collected in the gradient direction; Indicates the correlation difference of customer information in the gradient direction; S5, pre-processing step; S6. Analyze the relevance of customer information; S7. Construct a customer portrait model for power marketing informatization. Based on the customer information collected this time, factor analysis methods will be used to enhance the relevance of information variables, reduce the number of factors in a linear combination manner, and reduce the characteristic dimensions of customer information; Assume that the attribute value set of the power marketing subject customer ψ is is a g-dimensional vector, the attribute value The mean vector is χ=(χ1,χ2,…,χ g ), the set of common attribute factors existing in the power marketing subject customer ψ is Among them, h represents the number of common factors, and the set of special factors is λ=(λ1,λ2,…,λ I ), where I represents the number of special factors; then: In the formula, α represents the factor loading matrix. According to formula (7), the specific number of customer attribute value factors of power marketing entities can be obtained. Based on the attribute value factors, combined with the CS algorithm and the fuzzy mean algorithm, the customer portrait model of power marketing informationization is constructed. Assuming that the number of sample variables of the attribute values ​​of the power marketing subject customers is c, according to the attribute values ​​corresponding to the power marketing subject customers J cluster centers are formed. When the number of samples is c, the calculation matrix of the power marketing subject customer ψ corresponding to the attribute value is as follows: In the formula, represents the center vector of the cth attribute value sample variable of the Jth cluster center; according to the clustering method shown in formula (8), it is necessary to determine the fitness δ of each attribute value γ , and its calculation formula is as follows: In the formula, γ represents the fuzzy factor; L represents ψ i The distance to the jth cluster center; ψ i represents the sample of power marketing customers; θ ij represents the membership of ψi to the jth cluster center; based on the above formula, when the attribute value partition result obtained by the corresponding power marketing subject customer ψ satisfies the fitness of each attribute value δ γ If , the constructed customer portrait model is established, otherwise, it is necessary to recalculate the customer information attribute value factor until the attribute value division result obtained satisfies the calculated fitness value, stop the loop, and output the final constructed customer portrait model.

2. According to claim 1, a method for realizing a multi-dimensional, associated, and fine-grained power marketing information customer portrait model is characterized by: The pre-processing step S5 comprises: S51. When there is missing data in the data source, the data is supplemented or removed according to the amount of missing data; when there is a format error in the data in the data source, the data is removed or modified; S52. Since the data collection sources are different during the data collection process, there are situations where the same data has multiple names. Therefore, the data coding is unified; S53, determining whether there is a logical error in the coding; S54. When there are logical errors in the coding, the following steps need to be followed: (1) remove duplicate data in the data source; (2) remove fabricated data in the data source; (3) adjust the data conflicts in the data source; S55. When there are no logical errors in the coding, there are information duplication and conflict problems in the data source, which reduces the correlation between the data and causes the data to deviate from the actual situation. Therefore, it is necessary to use multi-dimensional correlation and fine-grained analysis of customer information correlation.

3. According to claim 2, a method for realizing a multi-dimensional, associated, and fine-grained power marketing information customer portrait model is characterized in that: The analyzing the relevance of customer information S6 includes: S61, support constraint: Assume that the main customer of power marketing collected is ψ, and its corresponding attribute value is At this time, the attribute value corresponding to the power marketing main customer It can be called an item set; since it is impossible for only one main customer of power marketing to be collected, n main customers of power marketing can be collected in one data collection cycle T, that is, n item sets can be formed, so it is called an n-item set; at this time, with ψ and As a reference, there is Based on this, The support degree is expressed as The number of occurrences in the data source, which can also be recorded as the number of occurrences of the item set Then we have: In the formula, Express The support of; O(m) represents the number of occurrences of all item sets; S62, Confidence Constraint: To avoid repeatedly adding the same attribute to the same customer profile, calculate the attribute value The probability of appearing on the power marketing main customer ψ is: In the formula, Indicates confidence level; The power marketing client ψ contains the attributes The probability of represents the power marketing subject customer ψ and attributes The probability of simultaneous occurrence; P(ψ) represents the probability of repeated occurrence of the power marketing subject customer ψ in the data source.

Citation Information

Patent Citations

  • A method and a system for constructing a portrait of a power grid user

    CN109359868A

  • Recommendation method for recommending customer service staff to power consumers

    CN111724039A

  • Electric power business hall service method based on artificial intelligence

    CN113240394A

  • Comprehensive energy service package configuration method

    CN113706208A

  • Training method and system of neural network model and prediction method and system

    US20210264272A1