CN111127080A - Big data recommendation algorithm-based customer channel drainage method - Google Patents
Big data recommendation algorithm-based customer channel drainage method Download PDFInfo
- Publication number
- CN111127080A CN111127080A CN201911230535.7A CN201911230535A CN111127080A CN 111127080 A CN111127080 A CN 111127080A CN 201911230535 A CN201911230535 A CN 201911230535A CN 111127080 A CN111127080 A CN 111127080A
- Authority
- CN
- China
- Prior art keywords
- client
- channel
- drainage
- customer
- national network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 50
- 230000001960 triggered effect Effects 0.000 claims abstract description 4
- 238000004458 analytical method Methods 0.000 claims description 21
- 230000000694 effects Effects 0.000 claims description 21
- 238000012795 verification Methods 0.000 claims description 18
- 239000011159 matrix material Substances 0.000 claims description 15
- 230000005611 electricity Effects 0.000 claims description 9
- 238000011156 evaluation Methods 0.000 claims description 9
- 238000011160 research Methods 0.000 claims description 7
- 238000010276 construction Methods 0.000 claims description 6
- 230000006870 function Effects 0.000 claims description 6
- 238000010606 normalization Methods 0.000 claims description 6
- 230000002159 abnormal effect Effects 0.000 claims description 5
- 238000002474 experimental method Methods 0.000 claims description 5
- 238000012545 processing Methods 0.000 claims description 5
- 238000012216 screening Methods 0.000 claims description 5
- 230000003542 behavioural effect Effects 0.000 claims description 4
- 238000010200 validation analysis Methods 0.000 claims description 4
- 239000002131 composite material Substances 0.000 claims description 3
- 230000008447 perception Effects 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 230000001737 promoting effect Effects 0.000 claims description 3
- 230000008439 repair process Effects 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000012360 testing method Methods 0.000 claims description 3
- 238000011161 development Methods 0.000 abstract description 3
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002860 competitive effect Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 230000002262 irrigation Effects 0.000 description 1
- 238000003973 irrigation Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 230000003014 reinforcing effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0255—Targeted advertisements based on user history
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0254—Targeted advertisements based on statistics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0269—Targeted advertisements based on user profile or attribute
- G06Q30/0271—Personalized advertisement
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Health & Medical Sciences (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Primary Health Care (AREA)
- Tourism & Hospitality (AREA)
- Probability & Statistics with Applications (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a customer channel drainage method based on a big data recommendation algorithm, and belongs to the technical field of electric power operation. The existing channel drainage method is mainly based on feeling and experience, can not accurately screen target customers, and can not mine customers with drainage potential. The method takes a client as a center to establish a client channel drainage model, starts with service and product requirements triggered by the client, and analyzes the tendency of the client to handle business by virtue of terminal equipment; meanwhile, the customer use information of each electronic channel is effectively fused, the customer groups are subdivided, the inherent characteristics and rules of customers using various electronic channels are deeply explored, the purpose of customers using the electronic channels is more comprehensively known, and guidance is provided for the development and marketing of the electronic channels; and further, the drainage of the customer channels is realized. The drainage scheme of the invention is comprehensive, rigorous and scientific, and is not easy to generate deviation.
Description
Technical Field
The invention relates to a customer channel drainage method based on a big data recommendation algorithm, and belongs to the technical field of electric power operation.
Background
As the power market is more competitive, the role of the channel is no longer limited to the sales of a single business and related services, but plays a decisive role in the market competition. The existing channel drainage method is mainly based on feeling and depending on experience, can not accurately screen target customers, can not excavate customers with drainage potential, and has incomplete, rigorous and scientific drainage scheme and easy deviation.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide a customer channel drainage method based on a big data recommendation algorithm, which is used for deeply analyzing the customer behavior characteristics of an online national network APP, constructing a channel drainage index system and a customer channel drainage recommendation model by combining the analysis result of the channel customer behavior preference, screening target customers with drainage conditions, mining the drainage scheme of customers with drainage potential, and ensuring that the drainage scheme is comprehensive, rigorous and scientific and is not easy to generate deviation.
In order to achieve the purpose, the technical scheme of the invention is as follows:
a customer channel drainage method based on big data recommendation algorithm is characterized in that a customer channel drainage model is established by taking a customer as a center, and the tendency of the customer to handle business by means of terminal equipment is analyzed starting from service and product requirements triggered by the customer; meanwhile, the customer use information of each electronic channel is effectively fused, the customer groups are subdivided, the inherent characteristics and rules of customers using various electronic channels are deeply explored, the purpose of customers using the electronic channels is more comprehensively known, and guidance is provided for the development and marketing of the electronic channels; and further, the drainage of the customer channels is realized.
The method specifically comprises the following steps:
the first step is as follows: channel customer behavior preference analysis
The behavior preferences mainly include: a payment preference;
the analysis of the payment preference adopts an RFM model and an entropy method to determine the weight of each index and construct a payment channel preference model; according to the time interval information R of the last payment of the customer till now, the total payment frequency F of using a certain channel in a certain time period and the average payment amount M of the certain channel in the time period; outputting the use evaluation index of each customer to each channel; the weight is optimized through an entropy method, the payment frequency F function is highlighted, the payment amount M and the payment interval R function are weakened, and the output evaluation result can objectively and accurately reflect the channel preference of a customer;
the second step is that: building client channel drainage recommendation model
Constructing a correlation matrix of client-client characteristic indexes and a similarity matrix of non-online national network clients based on a channel drainage index system, and obtaining a drainage target client group with higher similarity to the online national network APP client characteristic indexes, namely a client group with drainage potential, by adopting a client-based collaborative filtering-Pearson correlation coefficient analysis algorithm;
the third step: customer channel drainage scene construction
The client channel drainage scene comprises offline client drainage, online client drainage of a non-online national network APP, registered but unbound online national network APP client drainage and reflowable client drainage; and then based on the characteristics of the customers, a differentiated drainage strategy is formulated by combining specific business application scenes, and the purposes of customer channel drainage and accurate marketing service are achieved.
The invention provides a method for deeply analyzing the customer behavior characteristics of an online national network APP, which is characterized in that a channel drainage index system and a customer channel drainage recommendation model are constructed by combining the behavior preference analysis result of a channel customer, and target customers with drainage conditions are screened, so that customers with drainage potential can be fully mined. The drainage scheme of the invention is comprehensive, rigorous and scientific, and is not easy to generate deviation.
As a preferable technical measure:
the first step is to construct a payment channel preference model, and the method specifically comprises the following steps:
(1) data pre-processing
Processing missing values and abnormal values in the data, screening out customers with the age of at least 6 months and the total number of times of payment of at least 3 times in the research period, and eliminating the users with the payment of less than 10 yuan per month and the total number of times of payment per month of more than 5 times because the small amount of multiple payments of the users belong to the abnormal condition;
(2) index discretization
After the data distribution of the three indexes is checked, discretizing the R, F, M and other three indexes by using an equal division method and business experience, and determining the division of each interval;
(3) determining weights
Determining R, F, M and other three indexes by using an entropy method;
(4) determining an index score
According to the scores of all the intervals, the scores of R, F, M and other three indexes are calculated respectively;
(5) determining a composite score
According to the scores and the weights of the R, F, M and other three indexes, calculating the comprehensive historical preference score of each channel of the customer;
(6) model validation
Because the RFM model is an unsupervised model, the accuracy of the RFM model cannot be directly evaluated, and the verification of the preference model of the payment channel is to compare the preference of the historical payment channel obtained by the RFM model with the payment channel of the next month.
As a preferable technical measure:
the specific implementation steps of the entropy value method (3) are as follows:
assuming that the data has n rows of records and m variables, the data can be represented by a matrix A of n × m to form n rows and m columns, namely n rows of records and m characteristic columns;
A=[x1…xm]
normalization processing of data, namely minimum and maximum value normalization;
xijthe ith row and j column elements of matrix A are represented:
calculating the proportion of the ith record under the jth index:
calculating the entropy value of the j index:
calculating the difference coefficient of the j index:
gj=1-ej
calculating the weight of the j index:
as a preferable technical measure:
the behavioral preferences further include: querying the preference;
the related business query of the national network company comprises 4 channels such as an online national network APP, WeChat, a self-service terminal and 95598;
it includes the following 6 aspects:
(1) analyzing the distribution conditions of the user age, the urban and rural categories and the like of the client under the inquiry service;
(2) analyzing the distribution conditions of the user age, the urban and rural categories and the like of the clients under each channel;
(3) analyzing the customer volume proportion condition of each channel;
(4) analyzing query services of top5 in each channel in about 6 months;
(5) analyzing the frequency of each query service top5 in the last 6 months;
(6) and analyzing the frequency of the query business under each channel within one month.
As a preferable technical measure:
the behavioral preferences further include: business handling preferences;
the related business handling of the national network company comprises 3 channels such as an online national network APP, a self-service terminal, a business hall and the like;
it includes the following 6 aspects:
(1) analyzing the distribution conditions of the user age, the urban and rural categories and the like of the clients under each business;
(2) analyzing the distribution conditions of the user age, the urban and rural categories and the like of the clients under each channel;
(3) analyzing the customer volume proportion condition of each channel;
(4) analyzing the electricity transaction business of top5 under each channel in about 6 months;
(5) analyzing the total times of the top5 of each electricity transaction service in the last 6 months;
(6) and analyzing the total times of electricity transaction business under each channel in one month.
As a preferable technical measure:
the second step, pre-screening the target client group capable of being guided
The potential drainage reflects the possibility of a client downloading and registering an online national network APP during a promotional campaign; considering that the number of low-voltage resident customers is large, and the part of customer groups are relatively easy to accept the popularization of electronic channels, the research work of channel drainage is mainly developed aiming at the low-voltage resident customers; through research, drainage objects are mainly divided into four major categories, namely off-line clients, on-line clients not using the national network APP, registered online national network APP clients without real-name authentication and account numbers, and activatable clients which are palm power APP clients and do not use the national network APP for about three months.
As a preferable technical measure:
validating a model
(1) Definition of successful drainage
Definition of drainage success: downloading and registering the online national network APP by the client participating in the online national network APP popularization activity within 1 week after the popularization activity period and the activity end period, namely, judging that the drainage is successful;
(2) also includes a successful drainage verification method;
simulating to adopt a contrast test mode to verify the model effectiveness, and specifically verifying the model effectiveness by establishing an experimental group and a contrast group for contrast verification; randomly extracting 1000 clients from each city client, and setting the clients as a control group; extracting the same number of target customer groups from the same city according to the label characteristics, and setting the target customer groups as experiment groups; and simultaneously pushing the online national network APP to two groups of clients respectively, calculating the client proportion of successful drainage of the two groups of client groups to the online national network APP in the same month, comparing, repeatedly repeating the above processes for multiple times, and judging that the model is effective if the successful drainage rate formed by multiple verification experiments is higher than that of random sampling.
As a preferable technical measure:
draining off-line customers:
aiming at the services of paying, inquiring, complaints, fault repair and the like which are frequently handled in an online channel, but a few online channels are used as customers, diversified activities such as free experience in business halls, discount on order placement on an online platform, field activities of a third party and the like are developed, the experience perception of the customers is improved, and then the drainage of the customers is realized.
As a preferable technical measure:
online customer drainage of non-online national network APP:
the client habits are used in other electronic channels of non-online national network APP, and based on the preference behavior characteristics of the target client, such as sensitive electric bill, love for financing, attention to power failure and the like, personalized service is provided in a mode of a target client group, and the accuracy, quality and efficiency of marketing service are guaranteed.
As a preferable technical measure:
the registered but unbound online national network APP client drainage:
aiming at a client which is registered with the online national network APP and does not perform real-name authentication and binding because of curiosity of a new product or a new pulling activity promoted by the online national network APP, the activity degree of the online national network APP is low, and personalized promotion is performed based on the preference behavior characteristics of the client, so that the viscosity of the client is continuously enhanced, and the activity degree of the client is improved;
the reflowable client stream:
the preference behavior is specific to the old customers of the palm electric power APP but not using the Internet and national network APP or the customers registered and bound with the Internet and national network APP but not used in the last 3 months.
Compared with the prior art, the invention has the following beneficial effects:
the invention provides a method for deeply analyzing the customer behavior characteristics of an online national network APP, which is characterized in that a channel drainage index system and a customer channel drainage recommendation model are constructed by combining the behavior preference analysis result of a channel customer, and target customers with drainage conditions are screened, so that customers with drainage potential can be fully mined. The drainage scheme of the invention is comprehensive, rigorous and scientific, and is not easy to generate deviation.
Drawings
FIG. 1 is a diagram of a client channel drainage recommendation model construction according to the present invention;
FIG. 2 is a flow chart of a billing channel preference model of the present invention;
FIG. 3 is a flow chart of a drainage recommendation model construction of the present invention;
FIG. 4 is a flow chart of model verification according to the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
On the contrary, the invention is intended to cover alternatives, modifications, equivalents and alternatives which may be included within the spirit and scope of the invention as defined by the appended claims. Furthermore, in the following detailed description of the present invention, certain specific details are set forth in order to provide a better understanding of the present invention. It will be apparent to one skilled in the art that the present invention may be practiced without these specific details.
As shown in fig. 1-4, a client channel drainage method based on big data recommendation algorithm, which is to establish a client channel drainage model with a client as a center, and analyze the tendency of the client to handle business by means of a terminal device starting from service and product requirements triggered by the client; meanwhile, the customer use information of each electronic channel is effectively fused, the customer groups are subdivided, the inherent characteristics and rules of customers using various electronic channels are deeply explored, the purpose of customers using the electronic channels is more comprehensively known, and guidance is provided for the development and marketing of the electronic channels; and further, the drainage of the customer channels is realized.
The method specifically comprises the following steps:
the first step is as follows: channel customer behavior preference analysis
The behavior preferences mainly include: payment preference, query preference analysis and service handling preference analysis.
The second step is that: building client channel drainage recommendation model
Constructing a correlation matrix of client-client characteristic indexes and a similarity matrix of non-online national network clients based on a channel drainage index system, and obtaining a drainage target client group with higher similarity to the online national network APP client characteristic indexes, namely a client group with drainage potential, by adopting a client-based collaborative filtering-Pearson correlation coefficient analysis algorithm;
the third step: customer channel drainage scene construction
The client channel drainage scene comprises offline client drainage, online client drainage of a non-online national network APP, registered but unbound online national network APP client drainage and reflowable client drainage; and then based on the characteristics of the customers, a differentiated drainage strategy is formulated by combining specific business application scenes, and the purposes of customer channel drainage and accurate marketing service are achieved.
As shown in FIG. 2, one embodiment of the billing preference analysis of the present invention:
at present, the payment of related services of a national network company comprises 7 channels such as an online national network APP, a payment treasure, WeChat, an E treasure, a self-service terminal, a business hall, a bank and the like, and analysis is performed according to three aspects such as the distribution condition of the family age, the urban and rural categories and the like of customers under each channel, the payment behavior frequency under each channel in nearly 6 months, the customer volume proportion condition of each channel and the like.
And (3) constructing a payment channel preference model by adopting an RFM (recursive modeling) model and an entropy method (determining the weight of each index). And outputting the use evaluation index of each customer to each channel according to the time interval information (R) of the last payment of the customer to the present, the total payment frequency (F) of using a certain channel within 6 months and the average payment amount (M) of using the certain channel within 6 months. The weight is optimized through an entropy method, the payment frequency (F) function is highlighted, the payment amount (M) and the payment interval (R) function are weakened, and the output evaluation result can objectively and accurately reflect the channel preference of the customer.
The invention constructs a specific embodiment of a payment channel preference model, which comprises the following steps:
the method specifically comprises the following steps:
(1) data pre-processing
And processing the missing value and the abnormal value in the data, and screening out customers with the age of at least 6 months and the total number of times of payment of at least 3 times in the research period, wherein the customers with small amount and more payment of the customers belong to abnormal conditions, so that the customers with the payment of less than 10 yuan per month and the total number of times of payment per month of more than 5 times per month are eliminated.
(2) Index discretization
After the data distribution of the three indexes is checked, the three indexes of R, F, M and the like are discretized by utilizing an equal division method and business experience, and the division of each interval is determined (see the following table 5.1)
(3) Determining weights
The weights of R, F, M and other three indexes are determined by entropy method (see Table 1 below)
The entropy method is realized by the following steps:
assuming that the data has n rows of records and m variables, the data can be represented by a matrix A of n x m (n rows and m columns, i.e. n rows of records, m characteristic columns)
A=[x1...xm]
Normalization of data (minimum maximum normalization)
xij denotes the ith row and j column elements of matrix a.
Calculating the proportion of the ith record under the jth index
Calculating entropy of j index
Calculating the difference coefficient of the j index
gj=1-ej
Calculating the weight of the j index
(4) Determining an index score
According to the scores of all the intervals, the scores of R, F, M and other three indexes are respectively calculated (see the following table 1)
(5) Determining a composite score
And (8) calculating the comprehensive historical preference score of each channel of the customer according to the scores and the weights of the three indexes, such as R, F, M.
TABLE 1 RFM model index grouping
(6) Model validation
Since the RFM model is an unsupervised model and the accuracy of the RFM model cannot be directly evaluated, the historical payment channel preference model verification is to compare the historical payment channel preference obtained by the model with the next month payment channel. The verification evaluation result of the whole channel is 92.53%, and the verification evaluation result of each channel is shown in the following table 2; a "target channel preference customer base" output by the customer channel preference model.
Objectivity, the grouping rule of each index is determined by using an equal division bit method (see table 1), and the result of model verification (see table 2).
TABLE 2 Payment channel preference model verification results
Serial number | Channel for irrigation | Verification of evaluation results |
1 | All channel | 92.53% |
2 | Internet and national network APP | 68.67% |
3 | Bank | 96.08% |
4 | Payment device | 93.00% |
5 | 8111% | |
6 | Business hall | 78.64% |
7 | Electric E treasure | 06% |
8 | Self-service terminal | 64.56% |
9 | Others | 39.16% |
One specific embodiment of the query preference analysis of the present invention:
the related business query of the national network company comprises 4 channels such as an online national network APP, WeChat, a self-service terminal and 95598. The analysis was developed from 6 aspects:
(1) analyzing the distribution conditions of the user age, the urban and rural categories and the like of the client under the inquiry service;
(2) analyzing the distribution conditions of the user age, the urban and rural categories and the like of the clients under each channel;
(3) analyzing the customer volume proportion condition of each channel;
(4) analyzing query services of top5 in each channel in about 6 months;
(5) analyzing the frequency of each query service top5 in the last 6 months;
(6) and analyzing the frequency of the query business under each channel within one month.
The invention relates to a specific embodiment of service handling preference analysis, which comprises the following steps:
the related business transaction of the national network company comprises 3 channels such as an online national network APP, a self-service terminal and a business hall. The analysis was developed from 6 aspects:
(1) analyzing the distribution conditions of the user age, the urban and rural categories and the like of the clients under each business;
(2) analyzing the distribution conditions of the user age, the urban and rural categories and the like of the clients under each channel;
(3) analyzing the customer volume proportion condition of each channel;
(4) analyzing the electricity transaction business of top5 under each channel in about 6 months;
(5) analyzing the total times of the top5 of each electricity transaction service in the last 6 months;
(6) analyzing the total times of electricity transaction services in each channel within one month;
as shown in FIG. 3, one embodiment of the present invention for constructing a client channel drainage recommendation model is as follows:
2.1 definition of drainage
Generally, the drainage is to guide other channel customers to a target channel by using a customer-expanding mode and a channel adding mode, the operation is very simple, namely, the great field of the customers is used for making value contribution so as to enhance the trust of the customers, and the aim of adding new customers in the shortest time is to do so.
2.2 drainage object analysis
The potential drainage reflects the possibility of a customer downloading and registering an online national network APP during a promotional campaign. Considering that the number of low-voltage resident customers is large, and the part of customer groups are relatively easy to accept the popularization of electronic channels, the research work of channel drainage is mainly developed aiming at the low-voltage resident customers. Through research, drainage objects are mainly divided into four major categories, namely off-line clients, on-line clients not using the national network APP, registered online national network APP clients without real-name authentication and account numbers, and activatable clients which are palm power APP clients and do not use the national network APP for about three months.
2.3 big data analysis model construction
The client channel drainage recommendation model constructs a correlation matrix (shown in table 4) of client-client characteristic indexes and a similarity matrix (shown in table 5) of non-online national network clients based on a channel drainage index system (shown in table 3), and obtains a drainage target client group with higher similarity to the online national network APP client characteristic indexes by adopting a collaborative filtering-Pearson correlation coefficient analysis algorithm based on clients, namely the client group with drainage potential. And then based on the characteristics of the customers, a differentiated drainage strategy is formulated by combining specific business application scenes, and the purposes of customer channel drainage and accurate marketing service are achieved.
TABLE 3 channel drainage index System
TABLE 4 collaborative filtering recommendation algorithm based on customers
TABLE 5 customer similarity matrix
Customer A | Customer B | Customer C | Client D | Client E | |
Customer A | X | X | X | X | X |
Customer B | 0.87 | X | X | X | X |
Customer C | 0.24 | 0.46 | X | X | X |
Client D | 0.65 | 0.11 | 0.25 | X | X |
Client E | 0.98 | 0.76 | 0.46 | 0.77 | X |
2.4 model validation
(1) Definition of successful drainage
Definition of drainage success: and downloading and registering the online national network APP by the client participating in the online national network APP popularization activity within 1 week after the popularization activity period and the activity end, namely, the drainage is considered to be successful.
(2) Successful drainage verification method
And simulating to adopt a comparison test mode to verify the model effectiveness, and specifically verifying the model effectiveness by establishing an experimental group and a comparison verification mode of a comparison group.
As shown in fig. 4, it is assumed that 1000 customers are randomly selected from each city customer and set as a control group; and extracting the same number of target customer groups from the same city according to the label characteristics, and setting the target customer groups as experimental groups. And simultaneously pushing the online national network APP to two groups of clients respectively, calculating the client proportion of successful drainage of the two groups of client groups to the online national network APP in the same month, comparing, repeatedly repeating the above processes for multiple times, and judging that the model is effective if the successful drainage rate formed by multiple verification experiments is higher than that of random sampling.
The specific embodiment of the customer channel drainage scene design of the invention comprises the following steps:
(1) offline customer drainage
Aiming at the services of paying, inquiring, complaints, fault repair and the like which are frequently handled in an online channel, but a few online channels are used as customers, diversified activities such as free experience in business halls, discount on order placement on an online platform, field activities of a third party and the like are developed, the experience perception of the customers is improved, and then the drainage of the customers is realized.
(2) Non-online national network APP online customer drainage
The client habits are used in other electronic channels of non-online national network APP, and based on the preference behavior characteristics of the target client, such as sensitive electric bill, love for financing, attention to power failure and the like, personalized service is provided in a mode of a target client group, and the accuracy, quality and efficiency of marketing service are guaranteed.
(3) Registered but unbound online national network APP customer drainage
Aiming at the customers who probably register the online national network APP due to curiosity of new products or update activities promoted by the online national network APP but do not carry out real name authentication and binding, the activity degree of the online national network APP is lower, and the personalized promotion is carried out based on the preference behavior characteristics, so that the viscosity of the customers is continuously enhanced, and the activity degree of the customers is improved.
(4) Reflowable client drainage
The old customer to palm electric power APP but not use the customer of online national network APP yet or register and bind the customer that online national network APP but not used yet in nearly 3 months, based on its preference behavior characteristic, carry out accurate marketing to it and promote, promote customer loyalty, reinforcing user's sense of dependence.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (10)
1. A customer channel drainage method based on big data recommendation algorithm is characterized in that,
establishing a client channel drainage model by taking a client as a center, starting from service and product requirements triggered by the client, and analyzing the tendency of the client to handle business by virtue of terminal equipment; meanwhile, the customer use information of each electronic channel is effectively fused, the customer groups are subdivided, and the inherent characteristics and rules of customers using various electronic channels are deeply explored; further, the drainage of the customer channels is realized;
the method specifically comprises the following steps:
the first step is as follows: channel customer behavior preference analysis
The behavior preferences mainly include: a payment preference;
the analysis of the payment preference adopts an RFM model and an entropy method to determine the weight of each index and construct a payment channel preference model; according to the time interval information R of the last payment of the customer till now, the total payment frequency F of using a certain channel in a certain time period and the average payment amount M of the certain channel in the time period; outputting the use evaluation index of each customer to each channel; the weight is optimized through an entropy method, the payment frequency F function is highlighted, the payment amount M and the payment interval R function are weakened, and the output evaluation result can objectively and accurately reflect the channel preference of a customer;
the second step is that: building client channel drainage recommendation model
Constructing a correlation matrix of client-client characteristic indexes and a similarity matrix of a client of a non-online national network client-online national network based on a channel drainage index system, and obtaining a drainage target client group with higher similarity to the client characteristic indexes of the online national network APP (application), namely a client group with drainage potential, by adopting a client-based collaborative filtering-Pearson correlation coefficient analysis algorithm;
the third step: customer channel drainage scene construction
The client channel drainage scene comprises offline client drainage, online client drainage of a non-online national network APP, registered but unbound online national network APP client drainage and reflowable client drainage; and then based on the characteristics of the customers, a differentiated drainage strategy is formulated by combining specific business application scenes, and the purposes of customer channel drainage and accurate marketing service are achieved.
2. The big data recommendation algorithm-based client channel drainage method of claim 1,
the first step is to construct a payment channel preference model, and the method specifically comprises the following steps:
(1) data pre-processing
Processing missing values and abnormal values in the data;
(2) index discretization
After the data distribution of the three indexes is checked, discretizing the R, F, M and other three indexes by using an equal division method and business experience, and determining the division of each interval;
(3) determining weights
Determining R, F, M and other three indexes by using an entropy method;
(4) determining an index score
According to the scores of all the intervals, the scores of R, F, M and other three indexes are calculated respectively;
(5) determining a composite score
According to the scores and the weights of the R, F, M and other three indexes, calculating the comprehensive historical preference score of each channel of the customer;
(6) model validation
Because the RFM model is an unsupervised model, the accuracy of the RFM model cannot be directly evaluated, and the verification of the preference model of the payment channel is to compare the preference of the historical payment channel obtained by the RFM model with the payment channel of the next month.
3. The big data recommendation algorithm-based client channel drainage method of claim 1,
the specific implementation steps of the entropy value method (3) are as follows:
assuming that the data has n rows of records and m variables, the data can be represented by a matrix A of n × m to form n rows and m columns, namely n rows of records and m characteristic columns;
A=[x1,...xm]
normalization processing of data, namely minimum and maximum value normalization;
xijthe ith row and j column elements of matrix A are represented:
calculating the proportion of the ith record under the jth index:
calculating the entropy value of the j index:
calculating the difference coefficient of the j index:
gj=1-ej
calculating the weight of the j index:
4. the big data recommendation algorithm-based client channel drainage method of claim 1,
the behavioral preferences further include: querying the preference;
the related business query of the national network company comprises 4 channels such as an online national network APP, WeChat, a self-service terminal and 95598;
it includes the following 6 aspects:
(1) analyzing the distribution conditions of the user age, the urban and rural categories and the like of the client under the inquiry service;
(2) analyzing the distribution conditions of the user age, the urban and rural categories and the like of the clients under each channel;
(3) analyzing the customer volume proportion condition of each channel;
(4) analyzing query services of top5 in each channel in about 6 months;
(5) analyzing the frequency of each query service top5 in the last 6 months;
(6) and analyzing the frequency of the query business under each channel within one month.
5. The big data recommendation algorithm-based client channel drainage method of claim 4,
the behavioral preferences further include: business handling preferences;
the related business handling of the national network company comprises 3 channels such as an online national network APP, a self-service terminal, a business hall and the like;
it includes the following 6 aspects:
(1) analyzing the distribution conditions of the user age, the urban and rural categories and the like of the clients under each business;
(2) analyzing the distribution conditions of the user age, the urban and rural categories and the like of the clients under each channel;
(3) analyzing the customer volume proportion condition of each channel;
(4) analyzing the electricity transaction business of top5 under each channel in about 6 months;
(5) analyzing the total times of the top5 of each electricity transaction service in the last 6 months;
(6) and analyzing the total times of electricity transaction business under each channel in one month.
6. The big data recommendation algorithm-based client channel drainage method of claim 1,
the second step, pre-screening the target client group capable of being guided
The potential drainage reflects the possibility of a client downloading and registering an online national network APP during a promotional campaign; considering that the number of low-voltage resident customers is large, and the part of customer groups are relatively easy to accept the popularization of electronic channels, the research work of channel drainage is mainly developed aiming at the low-voltage resident customers; the drainage objects are mainly divided into four major categories, namely off-line clients, on-line clients not using the national network APP, registered online national network APP clients without real name authentication and account numbers bound, and activatable clients which are palm power APP clients and do not use the national network APP for nearly three months.
7. The big data recommendation algorithm-based client channel drainage method of claim 6,
also includes a successful drainage verification method;
simulating to adopt a contrast test mode to verify the model effectiveness, and specifically verifying the model effectiveness by establishing an experimental group and a contrast group for contrast verification; randomly extracting 1000 clients from each city client, and setting the clients as a control group; extracting the same number of target customer groups from the same city according to the label characteristics, and setting the target customer groups as experiment groups; and simultaneously pushing the online national network APP to two groups of clients respectively, calculating the client proportion of successful drainage of the two groups of client groups to the online national network APP in the same month, comparing, repeatedly repeating the above processes for multiple times, and judging that the model is effective if the successful drainage rate formed by multiple verification experiments is higher than that of random sampling.
8. The big data recommendation algorithm-based client channel drainage method according to any one of claims 1-7,
draining off-line customers:
aiming at the services of paying, inquiring, complaints, fault repair and the like which are frequently handled in an online channel, but a few online channels are used as customers, diversified activities such as free experience in business halls, discount on order placement on an online platform, field activities of a third party and the like are developed, the experience perception of the customers is improved, and then the drainage of the customers is realized.
9. The big data recommendation algorithm-based client channel drainage method of claim 8,
online customer drainage of non-online national network APP:
the client habits are used in other electronic channels of non-online national network APP, and based on the preference behavior characteristics of the target client, such as sensitive electric bill, love for financing, attention to power failure and the like, personalized service is provided in a mode of a target client group, and the accuracy, quality and efficiency of marketing service are guaranteed.
10. The big data recommendation algorithm-based client channel drainage method of claim 9,
the registered but unbound online national network APP client drainage:
aiming at a client which is registered with the online national network APP and does not perform real-name authentication and binding because of curiosity of a new product or a new pulling activity promoted by the online national network APP, the activity degree of the online national network APP is low, and personalized promotion is performed based on the preference behavior characteristics of the client, so that the viscosity of the client is continuously enhanced, and the activity degree of the client is improved;
the reflowable client stream:
the preference behavior is specific to the old customers of the palm electric power APP but not using the Internet and national network APP or the customers registered and bound with the Internet and national network APP but not used in the last 3 months.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911230535.7A CN111127080A (en) | 2019-12-04 | 2019-12-04 | Big data recommendation algorithm-based customer channel drainage method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911230535.7A CN111127080A (en) | 2019-12-04 | 2019-12-04 | Big data recommendation algorithm-based customer channel drainage method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111127080A true CN111127080A (en) | 2020-05-08 |
Family
ID=70497494
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911230535.7A Pending CN111127080A (en) | 2019-12-04 | 2019-12-04 | Big data recommendation algorithm-based customer channel drainage method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111127080A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111931039A (en) * | 2020-06-28 | 2020-11-13 | 重庆锐云科技有限公司 | Potential customer recommendation method and device, computer equipment and storage medium |
CN113065905A (en) * | 2021-05-07 | 2021-07-02 | 国家电网有限公司大数据中心 | Construction method, device, equipment and storage medium for online electricity purchasing drainage label |
CN113315768A (en) * | 2021-05-27 | 2021-08-27 | 焦作大学 | Database access credible system based on Internet of things |
CN113449184A (en) * | 2021-06-23 | 2021-09-28 | 平安科技(深圳)有限公司 | Recommendation method and device of reach channel, computer equipment and storage medium |
CN115983739A (en) * | 2023-03-21 | 2023-04-18 | 四川云瞻信息技术有限公司 | Decentralized-based business promotion channel matching method |
CN117522419A (en) * | 2024-01-08 | 2024-02-06 | 广东深玎科技有限公司 | Resource allocation method applied to customer relationship management system |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106096825A (en) * | 2016-06-03 | 2016-11-09 | 广东电网有限责任公司 | A kind of electrical power services channel analysis method and system |
CN106991173A (en) * | 2017-04-05 | 2017-07-28 | 合肥工业大学 | Collaborative filtering recommending method based on user preference |
CN107423859A (en) * | 2017-08-07 | 2017-12-01 | 国家电网公司客户服务中心 | A kind of built-up pattern modeling method and system |
CN107465739A (en) * | 2017-08-01 | 2017-12-12 | 中国联合网络通信集团有限公司 | The method and device of entity channel user drainage |
CN107506924A (en) * | 2017-08-25 | 2017-12-22 | 国网河南省电力公司电力科学研究院 | Power customer payment channel preference analyzing and predicting method based on RFM models |
CN107992883A (en) * | 2017-11-22 | 2018-05-04 | 福建省计量科学研究院 | A kind of metering industry customer's divided method based on CRFM models |
CN109461025A (en) * | 2018-10-23 | 2019-03-12 | 国网湖南省电力公司节能管理分公司 | A kind of electric energy substitution potential customers' prediction technique based on machine learning |
-
2019
- 2019-12-04 CN CN201911230535.7A patent/CN111127080A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106096825A (en) * | 2016-06-03 | 2016-11-09 | 广东电网有限责任公司 | A kind of electrical power services channel analysis method and system |
CN106991173A (en) * | 2017-04-05 | 2017-07-28 | 合肥工业大学 | Collaborative filtering recommending method based on user preference |
CN107465739A (en) * | 2017-08-01 | 2017-12-12 | 中国联合网络通信集团有限公司 | The method and device of entity channel user drainage |
CN107423859A (en) * | 2017-08-07 | 2017-12-01 | 国家电网公司客户服务中心 | A kind of built-up pattern modeling method and system |
CN107506924A (en) * | 2017-08-25 | 2017-12-22 | 国网河南省电力公司电力科学研究院 | Power customer payment channel preference analyzing and predicting method based on RFM models |
CN107992883A (en) * | 2017-11-22 | 2018-05-04 | 福建省计量科学研究院 | A kind of metering industry customer's divided method based on CRFM models |
CN109461025A (en) * | 2018-10-23 | 2019-03-12 | 国网湖南省电力公司节能管理分公司 | A kind of electric energy substitution potential customers' prediction technique based on machine learning |
Non-Patent Citations (4)
Title |
---|
张宏达;: "基于决策树算法的电子渠道精准营销效能分析" * |
梁林森: "用户电子服务渠道偏好模型的构建与应用" * |
涂莹等: "基于协同过滤算法的电能替代潜力用户挖掘模型研究" * |
蔡丽华;黄婷;蔡丽平;: "用户电子服务渠道偏好研究与应用" * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111931039A (en) * | 2020-06-28 | 2020-11-13 | 重庆锐云科技有限公司 | Potential customer recommendation method and device, computer equipment and storage medium |
CN113065905A (en) * | 2021-05-07 | 2021-07-02 | 国家电网有限公司大数据中心 | Construction method, device, equipment and storage medium for online electricity purchasing drainage label |
CN113315768A (en) * | 2021-05-27 | 2021-08-27 | 焦作大学 | Database access credible system based on Internet of things |
CN113315768B (en) * | 2021-05-27 | 2022-07-26 | 焦作大学 | Database access trusted system based on Internet of things |
CN113449184A (en) * | 2021-06-23 | 2021-09-28 | 平安科技(深圳)有限公司 | Recommendation method and device of reach channel, computer equipment and storage medium |
CN115983739A (en) * | 2023-03-21 | 2023-04-18 | 四川云瞻信息技术有限公司 | Decentralized-based business promotion channel matching method |
CN115983739B (en) * | 2023-03-21 | 2023-06-20 | 四川云瞻信息技术有限公司 | Service popularization channel matching method based on decentralization |
CN117522419A (en) * | 2024-01-08 | 2024-02-06 | 广东深玎科技有限公司 | Resource allocation method applied to customer relationship management system |
CN117522419B (en) * | 2024-01-08 | 2024-03-22 | 广东深玎科技有限公司 | Resource allocation method applied to customer relationship management system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111127080A (en) | Big data recommendation algorithm-based customer channel drainage method | |
US7523051B2 (en) | Information processing apparatus, information processing method, and program | |
CN103853948A (en) | User identity recognizing and information filtering and searching method and server | |
CN106447434A (en) | Personal credit ecological platform | |
CN107689008A (en) | A kind of user insures the method and device of behavior prediction | |
CN103582886B (en) | Valid keyword selection system and valid keyword selection method thereof | |
US9183301B2 (en) | Search engine optimization performance valuation | |
CN104813315A (en) | Customer-data analysis/evaluation system | |
CN103150696A (en) | Method and device for selecting potential customer of target value-added service | |
CN102722833A (en) | Persistent metadata for a user-controlled policy of personal data disclosure and usage for online advertising | |
CN110335596A (en) | Products Show method, apparatus, equipment and storage medium based on speech recognition | |
CN106408325A (en) | User consumption behavior prediction analysis method based on user payment information and system | |
CN105741161A (en) | Method and system for recognizing click farming users in taxi businesses on basis of driver credit | |
CN105978729A (en) | System and method for pushing mobile phone information based on user surfing log and position | |
CN109190669A (en) | A kind of intelligent recommendation method, electronic equipment and computer readable storage medium | |
CN117745355A (en) | Advertisement release system and release method based on mobile terminal | |
CN113450158A (en) | Bank activity information pushing method and device | |
Sun et al. | Predictive modeling of potential customers based on the customers clickstream data: A field study | |
CN108022150A (en) | recommendation method and system based on O2O data | |
KR20110034175A (en) | Method and server apparatus for calculating user conversion rate | |
CN108268506A (en) | A kind of processing method of promotion message, device, terminal and computer readable storage medium | |
KR100563451B1 (en) | On-line Advertising Method And System Using Community Site And Keyword Relative To The Community Site | |
KR20120022026A (en) | Survey server for panel survey, survey analysis method using the same, panel mobile terminal for survey and survey method using the same | |
CN113313506B (en) | Method and system for realizing user management of operator | |
KR100799628B1 (en) | The commerce information analysis service provision method which uses the real-time integration of information |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |