Lean fee urging method and system based on customer feature analysis

Abstract

The invention discloses a lean fee urging method and system based on customer feature analysis, relates to the technical field of decision making of user classification, and solves the problems that in the prior art, fee urging precision is insufficient, a mechanism is not perfect, and automation is insufficient due to the fact that operation decision output purely depends on human experience. According to the method, users are subjected to memory classification according to different industries and different industry standards, the fee urging channels and the success rates of the fee urging channels of each type of users are counted respectively, fee urging mode sequences of each type of users are obtained based on the fee urging success rates, and then a target function is constructed according to user satisfaction, fee urging frequency and unaffected factors of the arrearage length. And setting a corresponding constraint condition, finally solving a fee urging frequency which enables the target function to be maximum, finally dividing proportions of the fee urging modes based on the order of the fee urging success rates, and outputting a final decision. According to the invention, the fee urging service can be better supported, and the quality and efficiency of electric charge collection and customer experience are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of user classification decision-making, and in particular to a lean payment collection method and system based on customer feature analysis. Background Art

[0002] The collection of fees in the power industry is not only related to the economic benefits of the enterprise, but also an important indicator in the annual performance appraisal. In the context of increasingly fierce competition in the current power market, how to collect fees from users efficiently and accurately, while improving the user experience of being collected and reducing unnecessary complaints, has become a challenge that power companies need to solve urgently. With the continuous improvement of power grid facilities and the wide coverage of power supply networks, the user scope of collection services is constantly expanding, and users' demands for collection time periods, dates and collection methods are becoming increasingly diversified. Users expect to receive collection reminders at the right time and in the right way, which not only tests the service capabilities of power companies, but also puts higher demands on the strategy and execution of collection work. Therefore, power companies need to continuously innovate collection methods and optimize collection processes to meet the personalized needs of users, improve service quality, and ensure the smooth progress of collection work.

[0003] At present, the marketing smart brain has integrated payment collection channels such as SMS, WeChat, and voice outbound call robots, and realized multi-channel collaborative machine payment collection. It has been applied throughout the province and has replaced a considerable workload of grassroots payment collection. However, during the application process, problems such as insufficient payment collection accuracy, imperfect mechanisms, and insufficient automation caused by operational decision-making output relying solely on human experience were also discovered.

[0004] In view of this, a lean payment collection method and system based on customer characteristic analysis is needed. Summary of the invention

[0005] In view of the problems in the prior art of insufficient collection accuracy, imperfect mechanism, and insufficient automation caused by the output of operational decisions relying solely on human experience, the present invention provides a lean collection method and system based on customer feature analysis, which can form a collection strategy by considering "group + method + time" to better support collection services and improve both the quality and efficiency of electricity bill collection and customer experience. The specific technical solution is as follows: A lean collection method based on customer feature analysis, including: Classify users as ,in, L is the user level, L r Is the setting r Classification type, R Total number of user categories; Based on user classification, the collection channels and their success rates for each type of user are counted, and the collection methods for each type of user are ranked based on the collection success rate. The higher the collection success rate, the higher the ranking of the corresponding collection method. The objective function is constructed based on the factors of user satisfaction, collection frequency and arrears duration. Then, the relationship between collection frequency, user satisfaction and arrears duration is obtained through fitting. The corresponding constraints are set, and finally a collection frequency that maximizes the objective function is solved. Based on the ranking of collection success rates, the proportions of each collection method are divided and the final decision is output. The final decision is the number of collections for each collection method within the cycle.

[0006] Preferably, the calculation formula for the collection success rate is as follows: In the formula, Indicates that the payment reminder method is used for the user within time t s The number of reminders, Indicates that at time t, the j The successful payment coefficient after using the reminder method s for the first time, when hour, ,when hour, ; The time interval from when the reminder method is enabled to when the user pays the fee. The time interval threshold is set.

[0007] Preferably, the objective function is as follows: In the formula, For user satisfaction, The frequency of reminders, The duration of arrears.

[0008] Preferably, the constraints are as follows: In the formula, is the minimum acceptable collection frequency. It is the maximum acceptable collection frequency.

[0009] Preferably, user satisfaction and arrears duration are fitted into a function of payment reminder frequency, and the specific steps are as follows: Obtain historical data, including the frequency of payment reminders and their corresponding user satisfaction and the duration of arrears, and store them in a table or array; Check the integrity of the data, handle missing values ​​or outliers, and perform standardization or normalization; Choose polynomial regression, support vector machine regression, or neural network model to build a model for the function of customer satisfaction and collection frequency. , build a model for the function of the user's arrears time and the frequency of collection ; Use data on user satisfaction and payment reminder frequency to train model F, and use data on payment reminder frequency and user arrears duration to train model G. Adjust model parameters, perform cross-validation, and optimize model performance. Use appropriate evaluation indicators to evaluate the fit of the two models, verify the predictive ability of the models on the test set, and ensure that the models are not overfitted; Use the fitted models F and G to predict the user satisfaction and arrears duration values ​​corresponding to the new collection frequency values.

[0010] Preferably, the specific steps for solving the objective function are as follows: Step 1: Get the data set for decision making and randomly select an initial point as the starting point; Step 2: Calculate the gradient of the objective function Z at the current point; Step 3: Update the parameters along the direction of the gradient, define the learning rate, and control the step size; Step 4: Check whether the magnitude of the gradient is less than a preset threshold, or check whether the value of the objective function changes very little in several consecutive iterations. If so, the algorithm is considered to have converged. Step 5: If there is no convergence, return to step 2 and continue iterating; Step 6: When the algorithm converges, output the value of the current decision.

[0011] Preferably, users are classified based on the amount of electricity purchased, as follows: In the formula, L is the final output level, L R Is the setting R Classification type, is the threshold value set, Q It is the amount of electricity purchased by the user.

[0012] A lean collection system based on customer feature analysis, applied to the above method, includes: User classification unit, which classifies users according to different industries and different industry standards; The collection success rate statistics unit is connected to the decision-making terminal and is used to obtain the collection method and payment time point, further count the collection channels and their success rates for each type of user, and rank the collection methods for each type of user based on the collection success rate. The higher the collection success rate, the higher the ranking of the corresponding collection method; The collection strategy output unit is connected to the collection success rate statistics unit and the decision terminal, and the objective function is constructed based on the factors of user satisfaction, collection frequency and arrears duration. Then, the relationship between collection frequency, user satisfaction and arrears duration is obtained through fitting, and corresponding constraints are set. Finally, a collection frequency that maximizes the objective function is solved, and then the proportion of each collection method is divided based on the ranking of collection success rate, and the final decision is output. The final decision is the number of collection times of each collection method in the cycle; A decision terminal connected to the collection policy output unit and the user terminal, used to receive the collection policy from the collection policy output unit, and transmit collection information to the user terminal based on the collection policy; The user terminal is used to receive the reminder output by the decision terminal, and then provide the reminder success rate statistics unit with the data required for the reminder success rate calculation process after payment, including the time of receiving the reminder and the payment time.

[0013] A computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the lean collection method based on customer feature analysis as described above.

[0014] A processor is used to run a program, wherein the program, when running, executes the lean collection method based on customer feature analysis as described above.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The present invention classifies user memory according to different industries and different industry standards, and respectively counts the collection channels and success rates of each type of user, and ranks the collection methods of each type of user based on the collection success rate, and then constructs an objective function with factors that do not affect user satisfaction, collection frequency, and arrears duration, and then obtains the relationship between collection frequency and user satisfaction and arrears duration through fitting, sets corresponding constraints, and finally solves a collection frequency that maximizes the objective function, and then divides the proportion of each collection method based on the ranking of collection success rate, and outputs the final decision, which is the number of collection times of each collection method within the cycle. A collection strategy can be formed by considering "group + method + time" to better support collection services and achieve both quality and efficiency of electricity bill collection and customer experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0017] Figure 1 is a flow chart of the method of the present invention; Figure 2 is a flow chart of the data fitting process of the present invention; Figure 3 Schematic diagram of the specific steps of the gradient ascent method. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0020] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0021] It should be further understood that the term "and / or" used in the present description and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0022] In one embodiment of the present invention, a lean collection method based on customer feature analysis is provided, such as Figure 1 As shown, the following steps are included: Step 1: Classify users; Users are classified according to different industries and different industry standards, denoted as ,in, L is the user level, L rIs the setting r Classification type, R is the total number of user categories. The user categories in this step can be classified according to actual conditions. In this embodiment, an example of classifying users based on the amount of electricity purchased is given, as follows: In the formula, L is the final output level, L R Is the setting R Classification type, is the threshold value set, Q is the amount of electricity purchased by the user; it can be seen from the formula that by setting different thresholds for the amount of electricity purchased, users can be divided into different categories. Similarly, other data such as voltage level, user load, user scale, etc. can be used to classify users.

[0023] Step 2: Based on user classification, the collection channels and their success rates for each type of user are counted, and the collection methods for each type of user are ranked based on the collection success rate. The higher the collection success rate, the higher the ranking of the corresponding collection method. Based on the user classification in step 1, statistics are collected on each user's historical collection methods and the collection success rate of each collection method.

[0024] The calculation formula for the collection success rate of collection method s is as follows: In the formula, Indicates that the payment reminder method is used for the user within time t s The number of reminders, Indicates that at time t, the j The successful payment coefficient after using the reminder method s for the first time, when hour, ,when hour, ; The time interval from when the reminder method is enabled to when the user pays the fee. The time interval threshold is set.

[0025] That is, the formula shows that after sending a reminder message to the user at time t or enabling other reminder methods to notify the user to pay, the user will If you make payment within sThe payment collection is successful. This formula shows that even if more than two payment collection methods are used, the user successfully pays within the time interval threshold, then the two or more payment collection methods are regarded as valid payment collection methods, that is, at this time, the user's payment success is regarded as the result of the combination of these payment collection methods, so each payment collection method is counted within the output range, and each payment collection method is a valid payment collection method.

[0026] For each type of user L , calculate the success rate of each collection method under this category The collection methods include SMS, WeChat, voice call robots, etc. Each collection method is ranked based on the success rate. The higher the collection success rate, the higher the ranking of the corresponding collection method. Finally, the collection method ranking of each type of user is obtained, and it can be intuitively seen which collection method is most effective for this type of user. Moreover, this ranking can be updated in real time based on real-time data.

[0027] Step 3: Construct an objective function and obtain the user's specific collection strategy by minimizing the objective function; the objective function is as follows: In the formula, For user satisfaction, The frequency of reminders, The duration of arrears.

[0028] Constraints: In the formula, is the minimum acceptable collection frequency. It is the maximum acceptable collection frequency.

[0029] Among them, user satisfaction and arrears duration are fitted into a function of the frequency of payment reminders, such as Figure 2 As shown, the specific steps are as follows: (1) Obtain historical data, including the frequency of payment reminders and their corresponding user satisfaction and the duration of arrears, and store them in a table or array; (2) Check the integrity of the data and deal with missing values ​​or outliers. In addition, since user satisfaction, payment reminder frequency, and arrears duration have different dimensions, they are standardized or normalized. (3) Select polynomial regression, support vector machine regression or neural network model to build a model for the function of user satisfaction and collection frequency. Similarly, for the function of the user's arrears time and the frequency of payment reminders, build a model ; (4) Use data on user satisfaction and payment reminder frequency to train model F, and use data on payment reminder frequency and user arrears duration to train model G. Adjust model parameters, perform cross-validation, and optimize model performance. (5) Use appropriate evaluation indicators (such as mean square error (MSE), coefficient of determination (R²), etc.) to evaluate the fitting effect of the two models, verify the predictive ability of the models on the test set, and ensure that the models are not overfitted; (6) Use the fitted models F and G to predict the user satisfaction and arrears duration values ​​corresponding to the new collection frequency values; The objective function is solved by the least squares method or the gradient ascent method. The decision output module of this embodiment adopts the gradient ascent algorithm. The gradient ascent method is an iterative method that gradually adjusts the parameters along the direction of the objective function gradient to find the maximum value of the function. Figure 3 As shown, the following are the specific steps of the gradient ascent method: Step 1: Get the data set for decision making and randomly select an initial point as the starting point; Step 2: Calculate the gradient of the objective function Z at the current point; Step 3: Update the parameters along the direction of the gradient, define the learning rate, and control the step size; Step 4: Check whether the magnitude of the gradient is less than a preset threshold, or check whether the value of the objective function changes very little in several consecutive iterations. If so, the algorithm is considered to have converged. Step 5: If there is no convergence, return to step 2 and continue iterating; Step 6: When the algorithm converges, output the value of the current decision.

[0030] Step 4: Based on step 3, the optimal collection frequency for each user type is obtained, and then the proportion of each collection method is divided based on the ranking of collection success rate. For example, the ratio of SMS, WeChat, and voice outbound call robot is 3:2:1. The lower the collection success rate, the lower the proportion. For example, when the optimal frequency is 6 times per week, among the six collection times in a week, SMS collection accounts for three times, WeChat collection accounts for two times, and voice outbound call robot accounts for one time. The order of each collection method is random.

[0031] In one embodiment of the present invention, a lean payment collection system based on customer feature analysis is provided, comprising: User classification unit, which classifies users according to different industries and different industry standards, denoted as ,in, L is the user level, L r Is the setting r Classification type, RThe total number of user categories.

[0032] The collection success rate statistics unit is connected to the decision-making terminal and is used to obtain the collection method and payment time point, and further count the collection channels and their success rates for each type of user. The collection method ranking for each type of user is obtained based on the collection success rate. The higher the collection success rate, the higher the corresponding collection method ranking.

[0033] The collection strategy output unit is connected to the collection success rate statistics unit and the decision terminal, and constructs the objective function with user satisfaction, collection frequency and arrears duration as uninfluencing factors. Then, the relationship between collection frequency, user satisfaction and arrears duration is obtained through fitting, and corresponding constraints are set. Finally, a collection frequency that maximizes the objective function is solved. Then, based on the ranking of collection success rates, the proportions of various collection methods are divided, and the final decision is output. The final decision is the number of collection times for each collection method within the cycle.

[0034] The decision terminal is connected to the collection policy output unit and the user terminal, and is used to receive the collection policy from the collection policy output unit, and transmit the collection information to the user terminal based on the collection policy.

[0035] The user terminal is used to receive the reminder output by the decision terminal, and then provide the reminder success rate statistics unit with the data required for the reminder success rate calculation process after payment, such as the time of receiving the reminder and the payment time.

[0036] In summary, the unit of the present invention classifies users according to different industries and different industry standards, and counts the collection channels and success rates of each type of user respectively. The collection methods of each type of user are ranked based on the collection success rate, and then the objective function is constructed with factors that do not affect user satisfaction, collection frequency, and arrears duration. Then, the relationship between collection frequency, user satisfaction, and arrears duration is obtained through fitting, and corresponding constraints are set. Finally, a collection frequency that maximizes the objective function is solved, and then the proportions of each collection method are divided based on the ranking of collection success rates, and the final decision is output. The final decision is the number of collection times for each collection method within the cycle. A collection strategy can be formed by considering "group + method + time" to better support collection services and achieve both quality and efficiency of electricity bill collection and customer experience.

[0037] Those of ordinary skill in the art will appreciate that the units of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition of each example has been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0038] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical function division, and there may be other division methods in actual implementation, for example, multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored, etc.

[0039] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0040] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-0nlyMemory), random access memory (RAM, RandomAccessMemory), mobile hard disk, magnetic disk or optical disk, etc., which can store program code.

[0041] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A lean payment collection method based on customer feature analysis, characterized in that: include: Classify users as ,in, L is the user level, L r Is the setting r Classification type, R Total number of user categories; Based on user classification, the collection channels and their success rates for each type of user are counted, and the collection methods for each type of user are ranked based on the collection success rate. The higher the collection success rate, the higher the ranking of the corresponding collection method. The objective function is constructed based on the factors of user satisfaction, collection frequency and arrears duration. Then, the relationship between collection frequency, user satisfaction and arrears duration is obtained through fitting. The corresponding constraints are set, and finally a collection frequency that maximizes the objective function is solved. Based on the ranking of collection success rates, the proportions of each collection method are divided and the final decision is output. The final decision is the number of collections for each collection method within the cycle.

2. A lean payment collection method based on customer feature analysis according to claim 1, characterized in that: The calculation formula for the collection success rate is as follows: In the formula, Indicates that the payment reminder method is used for the user within time t s The number of reminders, Indicates that at time t, the j The successful payment coefficient after using the reminder method s for the first time, when hour, ,when hour, ; The time interval from when the reminder method is enabled to when the user pays the fee. The time interval threshold is set.

3. The lean payment collection method based on customer feature analysis according to claim 1 is characterized in that: The objective function is as follows: In the formula, For user satisfaction, The frequency of reminders, The duration of arrears.

4. A lean payment collection method based on customer feature analysis according to claim 3, characterized in that: The constraints are as follows: In the formula, is the minimum acceptable collection frequency. It is the maximum acceptable collection frequency.

5. A lean payment collection method based on customer feature analysis according to claim 4, characterized in that: User satisfaction and overdue payment duration are fitted into a function of payment reminder frequency. The specific steps are as follows: Obtain historical data, including the frequency of payment reminders and their corresponding user satisfaction and the duration of arrears, and store them in a table or array; Check the integrity of the data, handle missing values ​​or outliers, and perform standardization or normalization; Choose polynomial regression, support vector machine regression, or neural network model to build a model for the function of customer satisfaction and collection frequency. , build a model for the function of the user's arrears time and the frequency of collection ; Use data on user satisfaction and payment reminder frequency to train model F, and use data on payment reminder frequency and user arrears duration to train model G. Adjust model parameters, perform cross-validation, and optimize model performance. Use appropriate evaluation indicators to evaluate the fit of the two models, verify the predictive ability of the models on the test set, and ensure that the models are not overfitted; Use the fitted models F and G to predict the user satisfaction and arrears duration values ​​corresponding to the new collection frequency values.

6. A lean payment collection method based on customer feature analysis according to claim 5, characterized in that: The specific steps to solve the objective function are as follows: Step 1: Get the data set for decision making and randomly select an initial point as the starting point; Step 2: Calculate the gradient of the objective function Z at the current point; Step 3: Update the parameters along the direction of the gradient, define the learning rate, and control the step size; Step 4: Check whether the magnitude of the gradient is less than a preset threshold, or check whether the value of the objective function changes very little in several consecutive iterations. If so, the algorithm is considered to have converged. Step 5: If there is no convergence, return to step 2 and continue iterating; Step 6: When the algorithm converges, output the value of the current decision.

7. The lean payment collection method based on customer feature analysis according to claim 1 is characterized in that: Users are classified based on the amount of electricity purchased, as follows: In the formula, L is the final output level, L R Is the setting R Classification type, is the threshold value set, Q It is the amount of electricity purchased by the user.

8. A lean collection system based on customer feature analysis, characterized in that: The method applied to any one of claims 1 to 7, comprising: User classification unit, which classifies users according to different industries and different industry standards; The collection success rate statistics unit is connected to the decision-making terminal and is used to obtain the collection method and payment time point, further count the collection channels and their success rates for each type of user, and rank the collection methods for each type of user based on the collection success rate. The higher the collection success rate, the higher the ranking of the corresponding collection method; The collection strategy output unit is connected to the collection success rate statistics unit and the decision terminal, and the objective function is constructed based on the factors of user satisfaction, collection frequency and arrears duration. Then, the relationship between collection frequency, user satisfaction and arrears duration is obtained through fitting, and corresponding constraints are set. Finally, a collection frequency that maximizes the objective function is solved, and then the proportion of each collection method is divided based on the ranking of collection success rate, and the final decision is output. The final decision is the number of collection times of each collection method in the cycle; A decision terminal connected to the collection policy output unit and the user terminal, used to receive the collection policy from the collection policy output unit, and transmit collection information to the user terminal based on the collection policy; The user terminal is used to receive the reminder output by the decision terminal, and then provide the reminder success rate statistics unit with the data required for the reminder success rate calculation process after payment, including the time of receiving the reminder and the payment time.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the lean collection method based on customer feature analysis as described in any one of claims 1 to 7.

10. A processor, characterized in that: The processor is used to run a program, wherein the program, when running, executes the lean collection method based on customer feature analysis as described in any one of claims 1 to 7.

Images