Power supply marketing service method based on improved feature word weight algorithm
Technical Field
The invention relates to the field of power supply marketing service, in particular to a power supply marketing service method based on an improved feature word weight algorithm.
Background
TF-IDF (feature word weight): TF-IDF (termfrequency-inverse document frequency) is a commonly used weighting technique for information retrieval and data mining. Where TF is Term Frequency (Term Frequency) and IDF is Inverse text Frequency index (Inverse Document Frequency). TF-IDF is a statistical method to assess how important a word is for one of a set of documents or a corpus. The importance of a word increases in proportion to the number of times it appears in a document, but at the same time decreases in inverse proportion to the frequency with which it appears in the corpus. Various forms of TF-IDF weighting are often applied by search engines as a measure or rating of the degree of relevance between a document and a user query. In addition to TF-IDF, search engines on the internet use a ranking method based on link analysis to determine the order in which documents appear in search results.
Multi-tag Classification (ML-KNN): the Neighbor algorithm, or K-Nearest Neighbor (kNN) classification algorithm, is one of the simplest methods in data mining classification techniques. By K nearest neighbors is meant the K nearest neighbors, meaning that each sample can be represented by its nearest K neighbors.
The core idea of the kNN algorithm is that if most of k nearest neighbor samples of a sample in the feature space belong to a certain class, the sample also belongs to the class and has the characteristics of the sample on the class. The method only determines the category of the sample to be classified according to the category of the nearest sample or samples in the determination of classification decision. The kNN method is only relevant to a very small number of neighboring samples when making a class decision. Because the kNN method mainly determines the class by the limited neighboring samples around, rather than by the method of distinguishing the class domain, the kNN method is more suitable for the sample sets to be classified with more intersections or overlaps of the class domain than other methods.
Dynamic game: dynamic gambling (dynamic gambling) refers to the sequential order of the actions of participants, and the latter can observe the selection of the action predecessor and make corresponding selections according to the selection. Such a game is not considered to be a simultaneous decision anyway, and is called a dynamic game, also called a "multi-stage game".
Aiming at the problems of single traditional power supply marketing service mode, high service cost and poor power customer experience, power supply enterprises in China all adopt a passive mode to carry out power marketing service and evenly distribute service resources, so that the service pertinence of the individual requirements of power customers is not strong, and the power marketing service cost is huge and the power customer experience effect is not good. The traditional passive marketing service cannot meet the increasing service requirements of power customers, and the passive service mode is not beneficial to expanding the incremental distribution network market and cannot effectively consolidate the inventory of power customers.
Various solutions to the above problems have also been proposed and tried in recent years, such as: the scheme (1) is a marketing service strategy aiming at the customer satisfaction of a power supply enterprise, the strategy is based on a customer satisfaction theory, a power customer satisfaction evaluation system is constructed by using methods such as a Delphi method and the like, and the marketing service strategy is pertinently proposed according to the evaluation result; in the scheme (2), a power market marketing mode and a novel power price system under a new situation adopt menu power price and package service to guide the optimization of the marketing service resource allocation of a power supply enterprise aiming at a constantly changing power market environment; the scheme (3) is based on a power supply marketing service strategy of power market reform, and the satisfaction degree of power users on the marketing service of power supply enterprises is improved by analyzing an electric product selling strategy, a price strategy, a distribution strategy and a promotion strategy; still other solutions analyze and provide solutions from different angles for these problems, and achieve certain results, but these researches are limited to analyzing the measurement data and complaint data of customers, the researches only consider the satisfaction degree of power customers, lack the behavior, demand and value analysis of different types of customers, and do not fully account for the influence of the change of power supply marketing service strategies on the operation aspect of power supply enterprises. If different types of power customers are to be provided with differentiated services, the requirements of the power customers and the operation capacity of a power supply enterprise are comprehensively considered.
Disclosure of Invention
The invention aims to: the power supply marketing service method based on the improved feature word weight algorithm is provided, a classification method based on the feature word weight algorithm is adopted, a multi-label classification algorithm and a dynamic game are combined, different power supply marketing services can be formulated for different power customers, and the problem that the existing passive service mode cannot meet the increasing service requirements of the power customers is solved.
The technical scheme adopted by the invention is as follows:
a power supply marketing service method based on an improved feature word weight algorithm comprises the following steps:
classifying the information data of the power customer: based on power customer data of a power supply enterprise and external power customer data published by social institutions, data classification is carried out from 'behavior-demand-value' dimensions of power customers;
constructing a power customer characteristic label index set: constructing a power customer characteristic index set by using a multi-label classification algorithm;
analyzing the influence of the power grid operation service: analyzing the cost of a power supply enterprise, and performing dynamic game analysis to obtain an optimal power supply marketing service mode;
modeling the economic contribution sketch of the TF-IDF client: aiming at the economic contribution image of the power customer, carrying out image modeling on the economic contribution of the power customer through a TF-IDF algorithm;
formulating a power supply marketing service package: the power customers are classified into certain grades according to the figures of the power customers, and different contents are served according to different grades.
In order to better implement the scheme, further, the power customer information data classification includes the behavior of the power customer in the power utilization process and factors influencing the power utilization behavior of the power customer in terms of the behavior of the power customer;
the power customer information data classification comprises the feeling, evaluation and interaction of the power customers on the power supply service in the aspect of the requirements of the power customers;
the electric power customer information data classification comprises value contribution of the electric power customer to the electric power charge and or the value-added service of the power supply enterprise and the electric power customer fund payment capacity in the value aspect of the electric power customer.
In order to better implement the scheme, further, the method for constructing the feature tag index set of the power customer comprises the following steps: acquiring K neighbor index sets of the feature data of the newly added power customer; and calculating the label probability of the feature data of the newly added power customer by using the Bayesian conditional probability, wherein the neighbor index set with high probability is the final label of the feature data of the newly added power customer, and a complete power customer feature label index set is formed.
In order to better implement the scheme, further, the method for analyzing the influence of the power grid operation service includes:
the power supply marketing service content and the business influence of a power supply enterprise are jointly used as game party parameters, and a dynamic game model is established:
step 1: establishing power supply marketing clothesThe type sequence S of the transaction is S ═ S1,S2,......,SnN is power supply marketing service content;
step 2: the sequence of the influence of the power grid operation business is Eg={Eg1,Eg2,......,Egm},
Ej={Ej1,Ej2,......,Ejm},
Ey={Ey1,Ey2,......,Eym},
Ec={Ec1,Ec2,......,Ecm},
Ex={Ex1,Ex2,......,Exm}
Wherein m is the service influence type of a power supply enterprise, Eg is the planning influence of the power supply enterprise, Ej is the construction influence, Ey is the operation and maintenance influence, Ec is the financial influence, and Ex is the traditional power supply marketing service influence;
the dynamic game objective function influenced by the power grid operation service is as follows:
wherein Δ γiFor adjustable power supply marketing service resources, Ca is the overall power supply marketing service cost;
and step 3: the total power supply marketing service cost Ca satisfies two constraints:
ca is less than or equal to Call, wherein the Call is the sum of investment plans of power supply enterprises;
ca is less than or equal to Clong, wherein Clong is the long-term investment return of power supply enterprises;
and finally, solving the minimum total power supply marketing service cost min Ca in the dynamic game objective function which meets the two constraint conditions in the step 3.
In order to better realize the scheme, further, the method for modeling the economic contribution portrait of the TF-IDF client specifically comprises the following steps:
frequency T of appearance of characteristic index of power customerfIs composed of
Wherein lwNumber of times, L, that the characteristic data z of the power consumer appears in the user profilewTotal number of words for feature data z in the user representation;
reverse file frequency I of power customer characteristic indexdfIs composed of
Wherein q issThe total number of the characteristic data texts of each power customer, and Qs is the number of the total characteristic data texts of the power customer;
reverse file frequency I to power customer characteristic indexdfAdding a fixed offset N, the TF-IDF feature extraction function F of the power customertfidfComprises the following steps:
the weight F of the feature data in the power customer feature tag index set picturewIs composed of
Wherein T isfuThe frequency of one of the power customer characteristic data texts in the index set is set;
overall portrait characteristics P of power consumerallIs composed of
Where θ is the number of characteristic indicators for the power customer.
In order to better implement the scheme, further, the method for formulating the power supply marketing service package specifically comprises: setting a five-level power customer, wherein the items of the first-level power customer comprise a power bill list, arrearage reminding and power utilization safety training;
the project of the secondary power customer adds fault notification service on the project of the primary power customer;
limited payment service is added to the project of the second-level power customer in the project of the third-level power customer;
the project of the four-level power customer is added with professional equipment inspection, fault equipment replacement and equipment annual inspection reports on the project of the three-level power customer;
the project of the five-level power customer is added with a power utilization analysis report and an energy efficiency monitoring analysis on the project of the four-level power customer.
In order to better realize the scheme, further, after receiving the data of the power customer, judging whether the data of the power customer is a newly-built power customer characteristic label index set, if so, performing power customer information characteristic classification, and classifying the data into the established power customer characteristic label index set; if not, adding a new power customer characteristic label index set, then performing power customer information characteristic classification, and classifying the power customer data into the power customer characteristic label index set.
In analyzing behaviors, demands and values of different types of customers, the influence of changes of power supply marketing service strategies on the operation aspect of power supply enterprises needs to be considered. In order to provide differentiated services for different types of power customers, comprehensive consideration needs to be given to the requirements of the power customers and the operation capacity of power supply enterprises. In combination with the thought, in the scheme, the requirements of the power customers and the operation capacity of the power supply enterprises are comprehensively considered, firstly, the acquired data are subjected to data classification based on the 'behavior-demand-value' dimension of the power customers, the load curve, the power factor, the arrearage or default record and the like of the users can be set for the behavior data, the customer complaints, the customer suggestions and the like can be set for the demand data, and the electricity price information, the electricity charge data or the credit assessment, the judicial information and the like of the power customers can be set for the value data.
Then electricity is constructedThe power customer characteristic label index set uses an ML-KNN algorithm, namely a multi-label classification algorithm, for newly added power customer characteristic data, a final label of the newly added power customer characteristic data is formed by using a set of K nearest-neighbor indexed power customer characteristic label index sets, and thus, a complete and newly-built power customer characteristic label index set is formed. And then analyzing the influence of the power grid operation business by using a dynamic game method to obtain an optimal power supply marketing service mode, wherein the method is specifically as described above, and is based on the type sequence S of the power supply marketing service and each influence sequence E of the power grid operation business, and combines adjustable marketing service resources delta gammaiObtaining a dynamic game objective function of the minimum value min Ca of the total power supply marketing service cost Ca, wherein different influence sequences E of the power grid operation service are different influences; the total power supply marketing service cost Ca needs to be no greater than the sum of the investment plans Call and the long-term return on investment Clong of the power supply enterprise at the same time. And then modeling the economic contribution portrait of the TF-IDF client, wherein the aim is to realize the maximization of the income of a power supply enterprise, and the adjustment principle is to incline more service resources to the power client with large economic contribution, wherein the power client can use the reverse file frequency I of the characteristic index of the power clientdfA fixed offset N is added because in formula 3, if the number of the text of the feature data of a certain power customer is 0, the inverse file frequency I occursdfExcept for 0, therefore, we need to add an offset N, which is generally a small value and has little influence on the overall result. Finally, a power supply marketing service package is formulated, users with a certain number of grades can be generally set, and according to the grading of actual service items, the power customers are correspondingly provided with the overall portrait characteristics P of the power customersallThe value of (2) is divided into different levels of the number, and different service items are given to power customers of different levels.
In summary, due to the adoption of the technical scheme, the invention has the beneficial effects that:
1. the power supply marketing service method based on the improved feature word weight algorithm adopts a classification method of the feature word weight algorithm, combines a multi-label classification algorithm and a dynamic game, and makes different power supply marketing services aiming at different power customers, so that differentiated services are provided for the different power customers;
2. the power supply marketing service method based on the improved feature word weight algorithm adopts a classification method of the feature word weight algorithm, combines a multi-label classification algorithm and a dynamic game, formulates different power supply marketing services aiming at different power customers, comprehensively analyzes behaviors, demands and values of the power customers, gives consideration to the influence of the change of a power supply marketing service strategy on the operation of a power supply enterprise, and ensures that the scheme has universality;
3. the power supply marketing service method based on the improved feature word weight algorithm adopts a classification method of the feature word weight algorithm, combines a multi-label classification algorithm and a dynamic game, formulates different power supply marketing services aiming at different power customers, comprehensively considers the requirements of the power customers and the operation capacity of a power supply enterprise, and ensures that the scheme has universality.
Drawings
In order to more clearly illustrate the technical solution, the drawings needed to be used in the embodiments are briefly described below, and it should be understood that, for those skilled in the art, other related drawings can be obtained according to the drawings without creative efforts, wherein:
FIG. 1 is a diagram of the power marketing service method architecture of the present invention;
fig. 2 is a block flow diagram of the present invention.
Detailed Description
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it should be understood that the described embodiments are only a part of the embodiments of the present invention, and not all embodiments, and therefore should not be considered as a limitation to the scope of protection. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.
In the description of the present invention, it is to be noted that, unless otherwise explicitly specified or limited, the terms "disposed," "connected," and "connected" are to be construed broadly, and may be, for example, fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.
The present invention will be described in detail with reference to fig. 1 to 2.
Example 1:
a power supply marketing service method based on an improved feature word weight algorithm comprises the following steps:
classifying the information data of the power customer: based on power customer data of a power supply enterprise and external power customer data published by social institutions, data classification is carried out from 'behavior-demand-value' dimensions of power customers;
constructing a power customer characteristic label index set: constructing a power customer characteristic index set by using a multi-label classification algorithm;
analyzing the influence of the power grid operation service: analyzing the cost of a power supply enterprise, and performing dynamic game analysis to obtain an optimal power supply marketing service mode;
modeling the economic contribution sketch of the TF-IDF client: aiming at the economic contribution image of the power customer, carrying out image modeling on the economic contribution of the power customer through a TF-IDF algorithm;
formulating a power supply marketing service package: the power customers are classified into certain grades according to the figures of the power customers, and different contents are served according to different grades.
Example 2
In this embodiment, on the basis of embodiment 1, the power customer information data classification includes behaviors of the power customer in a power consumption process and factors affecting the power consumption behavior of the power customer in terms of the behavior of the power customer, and the data mainly comes from an internal system of a power supply enterprise;
the power customer information data classification comprises the feeling, evaluation and interaction with a power supply enterprise of a power customer on the power supply service in the aspect of the demand of the power customer, and the data mainly comes from an internal system of the power supply enterprise;
the electric power customer information data classification comprises value contribution of the electric power customer to electric charge and or value-added service of a power supply enterprise and capital payment capacity of the electric power customer in the value aspect of the electric power customer, and the data mainly comes from external data of the customer published by an internal system and a social institution of the power supply enterprise. In the present embodiment, these electric power customer information data as in table 1 below are set.
TABLE 1
The method for constructing the electric power customer characteristic label index set comprises the following steps: acquiring K neighbor index sets of the feature data of the newly added power customer; and calculating the label probability of the feature data of the newly added power customer by using the Bayesian conditional probability, wherein the neighbor index set with high probability is the final label of the feature data of the newly added power customer, and a complete power customer feature label index set is formed.
Specifically, let the number of samples of a power customer be N, the power customer behavior be B, the index set of the power customer demand be N, and the index sets of the power customer value V be:
B={B1,B2,.......,Bn},
N={N1,N2,.....,Nn},
V={V1,V2,.........,Vn}
let BiIs the power customer behavior feature vector corresponding to the ith instance, and sets NiIs the power customer demand feature vector corresponding to the ith instance, and is set as ViIs the power customer value characteristic corresponding to the ith instanceThe eigenvector, the sample set of the power customer is:
if k is the Euclidean distance between the feature tag and the cluster center of the power customer, d is the dimensionality of the feature tag of the power customer, and β is the number of the feature tag clusters of the power customer, the vectors of the behavior, the demand and the value cluster center of the power customer are respectively as follows:
then the power customer characteristic label L is addedkIs a feature mapping function ofk:
Setting the boundary of the model as psi, and adding a new power customer feature label L
kUnique feature set of
Comprises the following steps:
according to the formula, the final power customer characteristic type label of the newly added power customer characteristic label can be obtained by calculating the neighbor index set of the newly added power customer characteristic label, and the power customer characteristic label index set is perfected.
Here, the power customer feature tags in the original power customer feature tag set are selected, and K adjacent power customer feature tags adjacent to the new power customer feature tag are selected to form a small power customer feature tag set, and the power customer feature tag set is used as the feature of the latest power customer feature tag, so that the new power customer feature tag is merged into the original power customer feature tag set.
After the power supply enterprise provides better marketing service to the electric power customer, financial settlement mode transition may appear, the load increases rapidly, the change of dimensionality such as distribution network trend multidirectional, these changes will produce great influence to the planning of power supply enterprise, the construction, the operation and maintenance, businesses such as financial affairs, increase the operation input of power supply enterprise even by a wide margin, this scheme is through the input to power supply marketing service, dynamic game analysis is done to the output, solve optimal power supply marketing service mode, in order to satisfy power supply marketing service level promotion, reduce the comprehensive cost input of power supply enterprise. The method for analyzing the influence of the power grid operation business comprises the following steps:
the power supply marketing service content and the business influence of a power supply enterprise are jointly used as game party parameters, and a dynamic game model is established:
step 1: establishing a type sequence S of power supply marketing service as S ═ S { (S)1,S2,......,SnN is power supply marketing service content;
step 2: the sequence of the influence of the power grid operation business is Eg={Eg1,Eg2,......,Egm},
Ej={Ej1,Ej2,......,Ejm},
Ey={Ey1,Ey2,......,Eym},
Ec={Ec1,Ec2,......,Ecm},
Ex={Ex1,Ex2,......,Exm}
Wherein m is the service influence type of a power supply enterprise, Eg is the planning influence of the power supply enterprise, Ej is the construction influence, Ey is the operation and maintenance influence, Ec is the financial influence, and Ex is the traditional power supply marketing service influence;
the dynamic game objective function influenced by the power grid operation service is as follows:
wherein Δ γiFor adjustable power supply marketing service resources, Ca is the overall power supply marketing service cost;
and step 3: the total power supply marketing service cost Ca satisfies two constraints:
ca is less than or equal to Call, wherein the Call is the sum of investment plans of power supply enterprises;
ca is less than or equal to Clong, wherein Clong is the long-term investment return of power supply enterprises;
and finally, solving the minimum total power supply marketing service cost min Ca in the dynamic game objective function which meets the two constraint conditions in the step 3.
The method for modeling the economic contribution portrait of the TF-IDF client specifically comprises the following steps:
frequency T of appearance of characteristic index of power customerfIs composed of
Wherein lwNumber of times, L, that the characteristic data z of the power consumer appears in the user profilewTotal number of words for feature data z in the user representation;
reverse file frequency I of power customer characteristic indexdfIs composed of
Wherein q issThe total number of the characteristic data texts of each power customer, and Qs is the number of the total characteristic data texts of the power customer;
reverse file frequency I to power customer characteristic indexdfAdding a fixed offset of 0.0001, the power customer TF-IDF feature extraction function FtfidfComprises the following steps:
the weight F of the feature data in the power customer feature tag index set picturewIs composed of
Wherein T isfuThe frequency of one of the power customer characteristic data texts in the index set is set;
overall portrait characteristics P of power consumerallIs composed of
Where θ is the number of characteristic indicators for the power customer.
The method for formulating the power supply marketing service package specifically comprises the following steps: setting a five-level power customer, wherein the items of the first-level power customer comprise a power bill list, arrearage reminding and power utilization safety training;
the project of the secondary power customer adds fault notification service on the project of the primary power customer;
limited payment service is added to the project of the second-level power customer in the project of the third-level power customer;
the project of the four-level power customer is added with professional equipment inspection, fault equipment replacement and equipment annual inspection reports on the project of the three-level power customer;
the project of the five-level power customer is added with a power utilization analysis report and an energy efficiency monitoring analysis on the project of the four-level power customer. As shown in table 2, in addition, the power customers can increase the service contents by purchasing the member level, so as to meet the service expectations of different power customer groups.
TABLE 2
After receiving the data of the power customer, judging whether the data of the power customer is a newly-built power customer characteristic label index set, if so, performing power customer information characteristic classification, and classifying the data into the established power customer characteristic label index set; if not, adding a new power customer characteristic label index set, then performing power customer information characteristic classification, and classifying the power customer data into the power customer characteristic label index set.
The working principle is as follows: in analyzing behaviors, demands and values of different types of customers, the influence of changes of power supply marketing service strategies on the operation aspect of power supply enterprises needs to be considered. In order to provide differentiated services for different types of power customers, comprehensive consideration needs to be given to the requirements of the power customers and the operation capacity of power supply enterprises. In combination with the thought, in the scheme, the requirements of the power customers and the operation capacity of the power supply enterprises are comprehensively considered, firstly, the acquired data are subjected to data classification based on the 'behavior-demand-value' dimension of the power customers, the load curve, the power factor, the arrearage or default record and the like of the users can be set for the behavior data, the customer complaints, the customer suggestions and the like can be set for the demand data, and the electricity price information, the electricity charge data or the credit assessment, the judicial information and the like of the power customers can be set for the value data.
And then constructing a power customer characteristic label index set, wherein an ML-KNN algorithm, namely a multi-label classification algorithm, is used, and for newly-added power customer characteristic data, a final label of the newly-added power customer characteristic data is formed by using a set of K nearest-neighbor indexed power customer characteristic label index sets, so that a complete and newly-built power customer characteristic label index set is formed. And then analyzing the influence of the power grid operation business by using a dynamic game method to obtain an optimal power supply marketing service mode, wherein the method is specifically as described above, and is based on the type sequence S of the power supply marketing service and each influence sequence E of the power grid operation business, and combines adjustable marketing service resources delta gammaiTo obtain a totalThe dynamic game objective function of the minimum value min Ca of the power supply marketing service cost Ca is obtained, wherein different influence sequences E of the power grid operation business are different influences; the total power supply marketing service cost Ca needs to be no greater than the sum of the investment plans Call and the long-term return on investment Clong of the power supply enterprise at the same time. And then modeling the economic contribution portrait of the TF-IDF client, wherein the aim is to realize the maximization of the income of a power supply enterprise, and the adjustment principle is to incline more service resources to the power client with large economic contribution, wherein the power client can use the reverse file frequency I of the characteristic index of the power clientdfA fixed offset N is added because in formula 3, if the number of the text of the feature data of a certain power customer is 0, the inverse file frequency I occursdfExcept for 0, therefore, we need to add an offset N, which is generally a small value and has little influence on the overall result. Finally, a power supply marketing service package is formulated, users with a certain number of grades can be generally set, and according to the grading of actual service items, the power customers are correspondingly provided with the overall portrait characteristics P of the power customersallThe value of (2) is divided into different levels of the number, and different service items are given to power customers of different levels.
Other parts of this embodiment are the same as those of embodiment 1, and thus are not described again.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention in any way, and all simple modifications and equivalent variations of the above embodiments according to the technical spirit of the present invention are included in the scope of the present invention.