Bank queuing duration determination method and device and bank queuing duration determination system

Abstract

The invention provides a bank queuing duration determination method, a bank queuing duration determination device and a bank queuing duration determination system. The method comprises the steps of obtaining a to-be-handled service; constructing a time generation model; and inputting the business to the time generation model to obtain a first queuing duration of the current time corresponding to the business, the first queuing duration being used for being pushed to the user to improve the user experience and / or the bank outlet adjusting the business arrangement according to the first queuing duration. According to the scheme, the problem that the queuing duration of the bank outlets cannot be accurately predicted in the prior art is solved.

Description

Technical Field

[0001] The present application relates to the technical field of queuing duration prediction. Specifically, it relates to a method for determining the bank queuing duration, an apparatus for determining the bank queuing duration, a computer program product, and a bank queuing duration determination system. Background Art

[0002] Currently, the common methods for evaluating and predicting the queuing duration at bank branches basically only consider the average time, and predict the queuing duration at bank branches based on the average time. However, this method is too simple and cannot accurately predict the queuing duration at bank branches. Summary of the Invention

[0003] The main purpose of the present application is to provide a method for determining the bank queuing duration, an apparatus for determining the bank queuing duration, a computer program product, and a bank queuing duration determination system, so as to at least solve the problem in the prior art that the queuing duration at bank branches cannot be accurately predicted.

[0004] To achieve the above object, according to one aspect of the present application, there is provided a method for determining the bank queuing duration, including: obtaining the business to be processed; constructing a time generation model, where the time generation model is trained by using multiple sets of training data through the LSTM algorithm and the BP algorithm, and each set of training data in the multiple sets of training data includes historical services obtained within a historical time period and the historical queuing duration corresponding to the historical services, where the historical queuing duration is a comprehensive representation of the distribution characteristics of time calculated based on the average value, variance, skewness, and kurtosis of multiple historical actual queuing durations, and the historical actual queuing duration is the queuing duration actually collected in the pre-collected historical time period; inputting the service into the time generation model to obtain the first queuing duration at the current time corresponding to the service, where the first queuing duration is used to be pushed to the user to improve the user experience and / or the bank branch adjusts the business arrangement according to the first queuing duration.

[0005] Optionally, before constructing the time generation model, the method further includes: calculating the historical queuing duration according to a target formula, where the target formula is:

[0006]

[0007] PTT(ρ) represents the historical queuing duration, μ represents the average value, σ 2 represents the variance, S represents the skewness, K represents the kurtosis, and p represents the percentile of the time distribution.

[0008] Optionally, after inputting the service into the time generation model to obtain the first queuing duration at the current time corresponding to the service, the method further includes: obtaining the number of people queuing at the bank branch; updating the first queuing duration according to the number of people queuing to obtain a second queuing duration, where the first queuing duration is extended when the number of people queuing is greater than a preset number threshold, and the first queuing duration is shortened when the number of people queuing is less than the preset number threshold, and the step size for adjusting the first queuing duration is positively correlated with the difference between the first queuing duration and the preset number threshold.

[0009] Optionally, obtaining the number of people queuing at the bank branch includes: obtaining number-related information, where the number-related information includes at least one or more of the type of the service, the business volume of the bank branch, the weather, and the social network score of the bank branch; using the LSTM algorithm to determine the number of people queuing according to the number-related information, where the complexity level of the type of the service is positively correlated with the number of people queuing, the business volume is positively correlated with the number of people queuing, the severity of the weather is negatively correlated with the number of people queuing, and the social network score is positively correlated with the number of people queuing.

[0010] Optionally, after inputting the service into the time generation model to obtain the first queuing duration at the current time corresponding to the service, the method further includes: obtaining user information, where the user information includes at least the bank branches visited by the user, the time period when the user handles the service, and the distances between the user and all the bank branches; using an optimization algorithm to determine the optimal bank branch according to the first queuing duration and the user information, where the optimization algorithm includes one or more of gradient optimization, adaptive estimation, whale optimization, and greedy algorithm.

[0011] Optionally, after inputting the service into the time generation model to obtain the first queuing duration at the current time corresponding to the service, the method further includes: constructing a virtual queue according to all the services handled by the users and the first queuing durations of all the services; sending the virtual queue to the user terminal and displaying the remaining number of people and the remaining queuing time on the user terminal.

[0012] Optionally, after inputting the service into the time generation model to obtain the first queuing duration of the current time corresponding to the service, the method further includes: obtaining the branch information of multiple bank branches, where the branch information includes at least one or more of the queuing time of the bank branch, the service type handled by the bank branch, the user flow of the bank branch, the counter usage of the bank branch, the location information of the bank branch, the traffic information of the bank branch, and the usage rate of self-service equipment of the bank branch; constructing a policy generation model, where the policy generation model is trained using multiple sets of training data, and each set of training data in the multiple sets of training data includes historical branch information obtained within a historical time period and the historical scheduling policy corresponding to the historical branch information, where the historical scheduling policy is information on the migration of historical services and historical resource allocation among multiple bank branches; inputting the branch information into the policy generation model to obtain the scheduling policy corresponding to the branch information, where the scheduling policy is used to provide resource scheduling suggestions to balance the queuing time.

[0013] According to another aspect of the present application, there is provided an apparatus for determining the bank queuing duration, including: a first acquisition unit, configured to acquire the service to be processed; a first construction unit, configured to construct a time generation model, where the time generation model is trained using multiple sets of training data through the LSTM algorithm and the BP algorithm, and each set of training data in the multiple sets of training data includes historical services obtained within a historical time period and the historical queuing duration corresponding to the historical services, where the historical queuing duration is a comprehensive representation of the distribution characteristics of time calculated based on the average value, variance, skewness, and kurtosis of multiple historical actual queuing durations, and the historical actual queuing duration is the queuing duration collected in advance during the actual historical time period; a first processing unit, configured to input the service into the time generation model to obtain the first queuing duration of the current time corresponding to the service, where the first queuing duration is used to be pushed to the user to improve the user experience and / or the bank branch adjusts the service arrangement according to the first queuing duration.

[0014] According to yet another aspect of the present application, there is provided a computer program product, including a computer program, where when the computer program is executed by a processor, it implements the steps of any one of the methods for determining the bank queuing duration.

[0015] According to still another aspect of the present application, there is provided a bank queuing duration determination system, including: one or more processors, a memory, and one or more programs, where the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include those for executing any one of the methods for determining the bank queuing duration.

[0016] Applying the technical solution of the present application, starting from the phenomenon that there are significant differences in the queuing time distributions of outlets based on different business types, when considering the queuing duration of outlets, it extends from the original first moment (average value) to multiple moments, more precisely depicting the distribution characteristics of the queuing time. The LSTM-BP neural network combined model is used. This model combines the time series processing ability of the long short-term memory network (LSTM) and the optimization ability of the backpropagation neural network (BP), and can more accurately determine the queuing duration of bank outlets based on historical queuing data and business types. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings forming a part of this application are used to provide a further understanding of this application. The schematic embodiments of this application and their descriptions are used to explain this application and do not constitute an improper limitation to this application. In the drawings:

[0018] Figure 1 The hardware structure block diagram of a mobile terminal for implementing a method for determining the queuing duration of a bank according to an embodiment of the present application is shown;

[0019] Figure 2 The flowchart of a method for determining the queuing duration of a bank according to an embodiment of the present application is shown;

[0020] Figure 3 The flowchart of predicting the queuing duration is shown;

[0021] Figure 4 The flowchart of predicting the queuing duration according to the number of queuing people is shown;

[0022] Figure 5 The structure block diagram of a device for determining the queuing duration of a bank according to an embodiment of the present application is shown.

[0023] Among them, the above-mentioned accompanying drawings include the following reference numerals:

[0024] 102. Processor; 104. Memory; 106. Transmission device; 108. Input / output device. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0025] It should be noted that, without conflict, the embodiments in this application and the features in the embodiments can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0026] To enable those skilled in the art to better understand the solution of this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of this application.

[0027] It should be noted that the terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily have to be used to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so as to describe the embodiments of this application here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or are inherent to these processes, methods, products or devices.

[0028] As introduced in the background art, in the prior art, it is impossible to accurately predict the queuing duration of bank branches. To solve the above problems, the embodiments of this application provide a method for determining bank queuing duration, a device for determining bank queuing duration, a computer program product, and a system for determining bank queuing duration.

[0029] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0030] The method embodiments provided in the embodiments of this application can be executed on a mobile terminal, a computer terminal, or a similar computing device. Taking the operation on a mobile terminal as an example, Figure 1 is a hardware structure block diagram of a mobile terminal for a method of determining bank queuing duration according to an embodiment of the present invention. As Figure 1 shown, the mobile terminal may include one or more ( Figure 1 only one is shown in the figure) processors 102 (the processor 102 may include, but is not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA), and a memory 104 for storing data. Among them, the above-mentioned mobile terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those of ordinary skill in the art can understand that Figure 1 the structure shown is only schematic and does not limit the structure of the above-mentioned mobile terminal. For example, the mobile terminal may further include more or fewer components than those shown in Figure 1 , or have a different configuration from that shown in Figure 1 .

[0031] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the display method of device information in the embodiments of the present invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, implements the above-mentioned method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memories, or other non-volatile solid-state memories. In some instances, the memory 104 may further include a memory remotely disposed relative to the processor 102, and these remote memories may be connected to the mobile terminal through a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof. The transmission device 106 is used to receive or send data via a network. A specific example of the above-mentioned network may include a wireless network provided by a communication provider of the mobile terminal. In one instance, the transmission device 106 includes a network adapter (Network Interface Controller, abbreviated as NIC), which can be connected to other network devices through a base station and thus communicate with the Internet. In one instance, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0032] In this embodiment, a method for determining the bank queuing duration running on a mobile terminal, a computer terminal, or a similar computing device is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0033] Figure 2 It is a flowchart showing the method for determining the bank queuing duration according to an embodiment of the present application. As Figure 2 shown, the method includes the following steps:

[0034] Step S201, obtain the business to be handled;

[0035] Specifically, first, it is necessary to identify and obtain the specific business type that the user is about to handle at the bank branch. This is usually achieved through the user's reservation operation on the bank APP, website, or self-service terminal. The user selects or enters the business they are about to handle, such as cash deposit, loan consultation, credit card application, etc.

[0036] Step S202: Build a time generation model. The time generation model is trained using multiple sets of training data through the LSTM algorithm and the BP algorithm. Each set of the multiple sets of training data includes historical services obtained within a historical time period and the corresponding historical queuing duration of the historical services. The historical queuing duration is a comprehensive representation of the distribution characteristics of time calculated based on the average value, variance, skewness, and kurtosis of multiple historical actual queuing durations, where the historical actual queuing duration is the queuing duration actually collected in the pre - defined historical time period.

[0037] Specifically, the construction of the time generation model is based on a large amount of historical data, which includes the actual queuing durations of different service types within a historical time period. The data collection process needs to cover multiple service types and different time periods to ensure the generalization ability of the model. In the pre - processing stage, the data will be cleaned, outliers will be removed, and standardized or normalized processing will be performed to improve the efficiency and accuracy of model training.

[0038] For each set of historical data, the first moment (i.e., the average queuing duration), the second moment (variance), the third moment (skewness), and the fourth moment (kurtosis) of the queuing duration will be calculated. The calculation of these moments is based on the historical actual queuing duration, which reflects the central tendency, volatility, asymmetry, and kurtosis of the queuing duration distribution.

[0039] Use the LSTM - BP neural network combination. LSTM (Long Short - Term Memory network) is used to process sequence data and can capture long - term dependencies, which is particularly effective for time - series analysis such as queuing duration prediction. BP (Backpropagation algorithm) is used to optimize the weights of the neural network and learn through error backpropagation.

[0040] Take the historical service type and the corresponding queuing duration moment features as inputs to train the LSTM - BP neural network. The model will learn how to predict the distribution characteristics of the queuing duration of a certain service type in the future from historical data.

[0041] Step S203: Input the service into the time generation model to obtain the first queuing duration at the current time corresponding to the service. The first queuing duration is used to be pushed to the user to improve the user experience and / or the bank branch adjusts the business arrangement according to the first queuing duration.

[0042] Specifically, when the service type that the user is about to handle is received, it will use the trained time generation model to predict the queuing duration of this service type within the current time period. The predicted first queuing duration includes not only the average waiting time but also the volatility and asymmetry of the queuing time to provide more comprehensive risk information.

[0043] The first queuing duration of the prediction result will be pushed to users to help them make decisions on whether to wait or choose other times to handle business, thus improving the user experience. At the same time, bank branches can also adjust business arrangements such as the number of open windows and personnel scheduling according to the predicted queuing duration to optimize the operational efficiency of the branches.

[0044] Through this embodiment, starting from the phenomenon that there are significant differences in the queuing time distributions of branches based on different business types, when considering the queuing duration of branches, it is extended from the original first moment (average value) to multiple moments, more meticulously depicting the distribution characteristics of the queuing time. The LSTM-BP neural network combined model is used. This model combines the time series processing ability of the long short-term memory network (LSTM) and the optimization ability of the backpropagation neural network (BP), and can more accurately determine the queuing duration of bank branches based on historical queuing data and business types.

[0045] Specifically, the solution of this application can predict the queuing duration of the current business based on historical data, providing users with accurate waiting time estimates, improving the user experience, and at the same time providing data support for bank branches to adjust business arrangements according to the predicted queuing duration and improve service efficiency. Application scenarios include, but are not limited to, scenarios that require queuing such as bank counter services, self-service equipment use, and customer service centers.

[0046] Specifically, the disadvantages of existing inventions for predicting the queuing duration of branches include:

[0047] (1) Currently common methods for evaluating and predicting the queuing duration of branches basically only consider the average time and establish a mapping relationship between the business types of branches and the queuing time. However, issues such as changes in the number of branch windows, inconsistent business levels caused by teller transfers, and network problems are rarely taken into account, and these are all key reasons affecting changes in the handling duration.

[0048] (2) Existing queuing duration prediction models for branches that consider the volatility of the handling duration are either difficult to explain with actual mathematical tools for depicting the risk level of business handling and difficult to quantify the risk level acceptable to specific customers; or the models are extremely large and complex, resulting in slow model training due to numerous input conditions.

[0049] Therefore, a major problem of the existing solutions regarding queuing time prediction is that only the mapping relationship between the average queuing time and the service type label is established. Regarding the queuing time distribution, it is not necessarily a standard Gaussian distribution. When the queuing time distribution exhibits characteristics such as multi-modal, skewed, and long-tailed, a single average time is difficult to measure the specific queuing duration. Among the existing technologies that consider the risk index of the customer's business handling time, the existing solution is expressed as business risk entropy, but there is no actual quantifiable mathematical tool for how to obtain and calculate this business risk entropy. On the other hand, since accurate queuing time prediction is required, a model with complex input parameters can be established, which can predict more accurately, but the model training duration is another major problem.

[0050] Based on this, the solution of this application designs a method for evaluating and predicting queuing duration that quantifies the total time risk tolerance of customers handling business and does not require numerous input parameters. This method is based on the multi-order moments of queuing duration and can be used for queuing duration prediction under the background of the volatility change of queuing duration for different network services.

[0051] Specifically, the LSTM-BP neural network: LSTM stands for Long Short-Term Memory, which is a time-recurrent neural network. It is a recurrent neural network with a chain structure designed to solve the problem of long-term dependence existing in general recurrent neural networks. BP stands for Back Propagation, which is a multi-layer feedforward neural network trained according to the error backpropagation algorithm. The LSTM-BP neural network is a fusion neural network model that combines the advantages of both.

[0052] Specifically, the solution of this application designs a set of quantifiable indicators for calculating the risk degree of business handling to implement a queuing time prediction method under the condition that the time distribution presents a non-standard Gaussian distribution. Since moments can describe the distribution characteristics, using moment information such as the second moment, third moment, and third moment can further depict the distribution characteristics of queuing time.

[0053] Regarding the problem that there are extremely many influencing factors affecting queuing duration and they are mutually coupled, since these influencing conditions are finally reflected in the volatility of queuing time, compared with only analyzing the average time, more detailed description of the journey time distribution can skillfully avoid this problem and focus the problem on establishing the mapping model between service type and queuing duration distribution.

[0054] In summary, the solution of this application considers a multi - moment network point queuing duration evaluation model as the independent variable parameter of the prediction model, and combines it with the LSTM time - series prediction model to establish an LSTM - BP combined neural network queuing duration prediction method based on multi - moments. This prediction method will provide more queuing duration reference information and provide valuable queuing information for customers who care about the risk level of queuing time.

[0055] This technical solution considers the high - order moment characteristics of historical average queuing time data, combines time - series characteristics and business type characteristic parameters, and establishes a prediction model through the above independent variables to establish the influence degree of multi - moment parameters considering time distribution on queuing duration prediction.

[0056] In the specific implementation process, before constructing the time generation model, the above method further includes the following steps: calculating the above - mentioned historical queuing duration according to the target formula, where the above - mentioned target formula is:

[0057]

[0058] PTT(ρ) represents the above - mentioned historical queuing duration, μ represents the above - mentioned average value, σ 2 represents the above - mentioned variance, S represents the above - mentioned skewness, K represents the above - mentioned kurtosis, and p represents the percentile of the time distribution.

[0059] In this solution, through the above formula, the distribution characteristics of queuing duration can be captured more accurately, making the prediction of the model more accurate. Thus, in practical applications, such as bank peak - period management and resource allocation optimization, it provides a more effective decision - making basis.

[0060] Specifically, the average value, variance, skewness, and kurtosis can be calculated in any currently feasible way, and this solution will not elaborate.

[0061] is a random variable related to skewness and kurtosis. Φ represents the standard normal distribution function.

[0062] Specifically, the moment is a set of metrics for the distribution and morphological characteristics of variables. The n - th moment is defined as the integral of the product of the n - th power of a variable and its probability density function. Among them, the common first - order raw moment represents the mathematical expectation, the second - order central moment represents the variance, the third - order central moment represents the skewness, and the fourth - order central moment represents the kurtosis.

[0063] Specifically, the solution of this application applies the concept of moment in mathematical statistics. Since the moment can describe the distribution characteristics, the existing network point recommendation evaluation and prediction models only use the first - order moment, that is, the average value. In order to depict the skewed peak and long tail of the time distribution, this application's solution will introduce multi - moments. Among them, the second - order moment for time distribution represents the variance, the third - order moment represents the skewness, the fourth - order moment represents the kurtosis, etc.

[0064] Specifically, the predicted queuing duration is as Figure 3 shown and includes the following:

[0065] (1) Enter the time period, queuing duration, and information on the type of business handled by a single user at the network point, such as: cash business, commemorative coin business, etc.;

[0066] (2) Perform data preprocessing on historical data, such as data cleaning and data denoising;

[0067] (3) Take 10 minutes as a group to obtain the queuing duration time distribution of each business type within this time period;

[0068] (4) Calculate the first moment, second moment, third moment, fourth moment, etc. of this distribution based on the queuing duration time distribution;

[0069] (5) Substitute the obtained first four moments into the Cornish-Fisher expansion formula to obtain the queuing time duration index (queuing duration efficiency index);

[0070] (5) Adjust the mapping model parameters of the business type and queuing time duration index in the BP neural network;

[0071] (6) Based on the business type, establish an LSTM-BP neural network model between the time series T of the time period when the business is handled, the business type labels of {A1, A2,..., Ax} types, and the queuing time duration PTT(p) and the historical time series;

[0072] (7) Adjust the model parameters in step (5) according to the prediction effect;

[0073] (8) Establish a queuing duration prediction model for a certain type of business;

[0074] (9) Determine whether all business types have been trained;

[0075] (10) Determine the prediction model parameters used based on the business type label;

[0076] (11) Complete the queuing duration prediction for business type A.

[0077] In some embodiments, after inputting the above business into the above time generation model to obtain the first queuing duration of the current time corresponding to the above business, the above method further includes the following steps: obtaining the number of people queuing at the above bank branch; updating the above first queuing duration according to the above number of people queuing to obtain a second queuing duration, wherein when the above number of people queuing is greater than a preset number threshold, the above first queuing duration is extended, and when the above number of people queuing is less than the above preset number threshold, the above first queuing duration is shortened, wherein the step size for adjusting the above first queuing duration is positively correlated with the difference between the above first queuing duration and the above preset number threshold.

[0078] In this solution, by updating the queuing duration in real time, it can dynamically reflect the actual situation of the bank branch, help users make more reasonable waiting time estimates, and at the same time provide real-time resource scheduling requirements for the bank to ensure service quality and efficiency.

[0079] Specifically, if the number of people queuing is greater than the preset number threshold, this means that the branch is busier than usual at present, so the predicted queuing duration (the first queuing duration) will be extended to reflect this situation. The step size of the adjustment (that is, how much time to increase) is proportional to the difference between the first queuing duration and the preset number threshold. In other words, the busier the branch, the more the predicted time increases.

[0080] If the number of people queuing is less than the preset number threshold, this indicates that the branch is relatively idle at present, and the predicted queuing duration will be shortened. Similarly, the step size of the adjustment is proportional to the difference between the first queuing duration and the preset number threshold. The more idle the branch, the more the predicted time decreases.

[0081] In addition, if the number of people queuing is equal to the preset number threshold, then the first queuing duration can remain unchanged.

[0082] After the above adjustment, a queuing duration closer to the actual situation, that is, the second queuing duration, is calculated, which can provide users with a more accurate waiting time prediction, help them make better decisions, and at the same time the bank can optimize resource allocation based on this information.

[0083] Suppose the prediction model of bank branch A predicts that the average queuing duration for a certain business type at 10 am is 20 minutes, the variance is 25 minutes^2, the skewness is 0.5, and the kurtosis is 4, that is, the first queuing duration is 20 minutes. The preset number threshold is set at 50 people, indicating that when the number of people in the branch is below 50, the queuing situation is relatively normal; when it exceeds 50, the queuing may intensify.

[0084] The real-time number of people queuing is 40:

[0085] Since the number of people in the queue is less than the preset number threshold, the first queuing duration will be shortened based on the preset rules. If the calculation rule for the adjustment step is to reduce the first queuing duration by 5% for every 10 people below the threshold, then when the number of people in the queue is 10 less (40 people compared to the threshold of 50 people), the first queuing duration will be reduced by 10%, that is, 2 minutes will be reduced from 20 minutes, and the second queuing duration is updated to 18 minutes.

[0086] The real-time number of people in the queue is 60:

[0087] When the real-time number of people in the queue exceeds the preset number threshold, the first queuing duration will be extended. Also following the rule that the adjustment step is proportional to the difference between the first queuing duration and the preset number threshold, assuming that for every 10 people exceeding the threshold, the first queuing duration is increased by 10%, then when the number of people in the queue exceeds 10 (60 people compared to the threshold of 50 people), the first queuing duration will be increased by 10%, that is, 2 minutes will be increased from 20 minutes, and the second queuing duration is updated to 22 minutes.

[0088] Specifically, the process of determining the queuing duration according to the number of people in the queue is as Figure 4 shown, including the following:

[0089] (1) Input the historical queuing time information of each business type;

[0090] (2) Substitute into Figure 3 the expansion formula of the fourth step in, and obtain the historical queuing duration index (historical queuing efficiency index) of each business type;

[0091] (3) Train the BP neural network model of the business type and the historical queuing duration;

[0092] (4) Establish a mapping model between the business type and the queuing duration;

[0093] (5) Train the LSTM neural network model of the historical queuing duration of business type A in the time series;

[0094] (6) Establish a time series model for the queuing duration of business type A;

[0095] (7) Use the LSTM neural network to predict the future queuing number information;

[0096] (8) Use the BP neural network, take the queuing number predicted in step (7) as the input, and predict the queuing duration data at the acceptable risk level p in the future.

[0097] In the specific implementation process, obtaining the queuing number of the above bank branch can be achieved through the following steps: obtaining information related to the number of people, where the information related to the number of people at least includes one or more of the type of the above business, the business volume of the above bank branch, the weather, and the social network score of the above bank branch; using the LSTM algorithm to determine the queuing number according to the information related to the number of people, where there is a positive correlation between the complexity of the type of the above business and the queuing number, there is a positive correlation between the business volume and the queuing number, there is a negative correlation between the severity of the weather and the queuing number, and there is a positive correlation between the social network score and the queuing number.

[0098] In this solution, multiple factors can be comprehensively considered to improve the accuracy of prediction. Especially during special periods such as weather changes and holidays, the queuing number can be predicted more accurately, providing more reliable information for banks and users.

[0099] Specifically, a number generation model can be constructed, where the number generation model is trained by the LSTM algorithm using multiple sets of training data, and each set of training data in the multiple sets of training data includes historical information related to the number of people obtained within a historical time period and the historical queuing number corresponding to the historical information related to the number of people; inputting the information related to the number of people into the number generation model to obtain the queuing number at the current time corresponding to the information related to the number of people.

[0100] Specifically, collect various information related to predicting the queuing number, including but not limited to the type of business, the business volume of the bank branch, the weather conditions, and the score of the bank branch on the social network. These information comprehensively reflect the attractiveness and busyness of the branch at a specific moment.

[0101] Business type and queuing number: There is a positive correlation between the complexity of the business and the queuing number. For example, customers involved in complex businesses such as loan consulting and account opening require longer service time, which may lead to an increase in queuing time and the number of people in the queue.

[0102] Business volume of bank branch and queuing number: The business volume of the bank branch (i.e., the number of businesses per day or within a certain time period) is also positively correlated with the queuing number. Branches with a large business volume usually attract more customers, thus increasing the queuing number.

[0103] Weather and queuing number: There is a negative correlation between the severity of the weather and the queuing number. Under bad weather conditions, such as heavy rain or high temperature, people may be more inclined to avoid going out, resulting in a decrease in the queuing number at the branch.

[0104] Social Network Rating and Queue Length: The rating of a bank branch on social networks is positively correlated with the queue length. A high rating usually indicates better service quality and customer satisfaction, which can attract more customers to visit, thus increasing the queue length.

[0105] Suppose we are predicting the queue length of Bank Branch B at 2 PM on Friday. The following is the relevant information collected:

[0106] Business Types: The main businesses are cash transactions (relatively simple) and loan consultations (relatively complex).

[0107] Business Volume: In the past week, this branch processed an average of 500 transactions per day.

[0108] Weather: It is predicted that there will be rain in the afternoon of that day, and the temperature will be moderate.

[0109] Social Network Rating: On the main social platforms, the average rating of this branch is 4.5 / 5.

[0110] Based on this information, we use the LSTM algorithm for prediction:

[0111] Impact of Business Types: Due to the high complexity of loan consultation services, an increase in their proportion may lead to an increase in the queue length. For example, if the proportion of loan consultation services increases from 20% to 30%, the queue length may increase by 5%.

[0112] Impact of Business Volume: The high business volume of Bank Branch B means more customer visits, and the queue length may increase accordingly. For example, if the business volume increases by 100 transactions per day, the queue length may increase by 3%.

[0113] Impact of Weather: Due to the predicted bad weather, people may be more inclined to handle affairs at home, and the queue length may decrease by 4% accordingly.

[0114] Impact of Social Network Rating: Branches with high ratings will attract more customers and increase the queue length. For example, if the social network rating increases by 0.5 points, the queue length may increase by 2%.

[0115] Based on the analysis of these factors by the LSTM algorithm, the final prediction results are as follows:

[0116] Initial Queue Length Prediction: Based on historical averages, the predicted queue length is 60 people.

[0117] Prediction after Considering the Change in Business Type Complexity: +3% (impact of the increase in the proportion of loan consultation services).

[0118] Prediction after Considering the Change in Business Volume: +2% (impact of the increase in business volume).

[0119] Prediction after considering weather changes: -4% (impact of bad weather).

[0120] Prediction after considering changes in social network ratings: +1% (impact of rating improvement).

[0121] The predicted queuing number after comprehensive adjustment is:

[0122] 60 people × (1 + 3% + 2% - 4% + 1%) = 60 people × 1.02 = 61.2 people.

[0123] The prediction system will adjust the final queuing number to about 61 people (usually an integer is taken). This value reflects the queuing number prediction considering all relevant factors, and can provide more accurate guidance for resource allocation at bank branches and the expectation of customer waiting time. Through such a prediction mechanism, the branch can pre-adjust the service staff configuration in advance, and customers can choose the best time to handle business according to the predicted queuing number, thereby improving the overall customer experience and the operation efficiency of the branch.

[0124] In some embodiments, after inputting the above business into the above time generation model to obtain the first queuing duration at the current time corresponding to the above business, the above method further includes the following steps: obtaining user information, where the above user information at least includes the above bank branches visited by the user, the time period for the user to handle business, and the distances between the user and all the above bank branches; using an optimization algorithm, according to the above first queuing duration and the above user information, determining the optimal above bank branch, where the optimization algorithm includes one or more of gradient optimization, adaptive estimation, whale optimization, and greedy algorithm.

[0125] In this solution, by intelligently recommending the optimal bank branch, not only the waiting time of users is reduced, but also the customer flow distribution at bank branches is optimized, avoiding overcrowding at some branches and improving the overall service quality and efficiency.

[0126] Specifically, obtain the historical behavior data of the user, including the bank branches the user has visited, the time period for business handling preferred by the user, and the distances between the user and all bank branches. This information helps to understand the user's preferences and geographical limitations.

[0127] The first queuing duration refers to the waiting time for a specific business at a bank branch predicted based on historical data and current conditions. This is one of the main bases for selecting the optimal branch.

[0128] An optimization algorithm is a method for finding the optimal solution in a given parameter space. In this solution, the optimization algorithm is used to comprehensively consider the queuing duration and user information to determine which bank branch a user should preferentially choose for business processing. Several optimization algorithms are mentioned here, including gradient optimization, adaptive estimation, whale optimization (a heuristic search algorithm), and greedy algorithm, which can be used alone or in combination.

[0129] In the specific implementation process, after inputting the above business into the above time generation model to obtain the first queuing duration at the current time corresponding to the above business, the above method further includes the following steps: constructing a virtual queue according to all the above businesses handled by users and the first queuing durations of all the above businesses; sending the above virtual queue to the user terminal and displaying the remaining number of people and the remaining queuing time on the above user terminal.

[0130] In this solution, the construction of the virtual queue not only enables users to clearly understand their queuing positions but also can reduce on-site waiting through remote services, appointment handling, etc., improving the convenience and satisfaction of services.

[0131] Specifically, a virtual queue model is constructed according to the types of businesses currently being handled by all users and the predicted queuing durations. This model helps banks and users understand the queuing situation at the branch in real time.

[0132] The information of the virtual queue, including the current remaining number of people in the queue and the predicted remaining queuing time, is sent to the user's terminal device, such as a mobile phone, tablet, or computer, through an application or other digital platforms.

[0133] Suppose at 14:00 on Friday afternoon, the following businesses are being handled at Bank Branch X:

[0134] Cash deposit and withdrawal business: The number of people in the queue is 10, and the average predicted queuing duration is 10 minutes.

[0135] Personal loan consultation: The number of people in the queue is 5, and the average predicted queuing duration is 25 minutes.

[0136] Account management: The number of people in the queue is 3, and the average predicted queuing duration is 15 minutes.

[0137] Combining these business types and their corresponding queuing numbers and time predictions, a virtual queue is constructed. For example, suppose Zhang is about to handle the cash deposit and withdrawal business. He can see the following virtual queue information:

[0138] The current number of people in the queue for the cash deposit and withdrawal business in the virtual queue is 10, and the remaining number of people in the queue is 10 (because Zhang is about to join).

[0139] The remaining queuing time for cash deposit and withdrawal services in the virtual queue is approximately 100 minutes (10 people × average predicted queuing time of 10 minutes).

[0140] When Zhang uses the bank's application, he will see the following information: Remaining number of people: 10 people (for cash deposit and withdrawal services). Remaining queuing time: 100 minutes (for cash deposit and withdrawal services).

[0141] In addition, Zhang can also view the queuing situation of other business types through the application, and select different business types and outlets for reservation or go directly according to personal needs. This not only provides users with a clear queuing expectation, but also helps the bank manage the customer flow of the outlets and improve service efficiency.

[0142] By constructing a virtual queue and updating the queuing situation in real time, the bank can better manage the customer flow, provide more accurate queuing time predictions, and thus enhance the customer experience. For users, this enables them to more reasonably plan the time to visit the bank outlets according to their own time arrangements and business needs, and avoid unnecessary waiting.

[0143] In some embodiments, after inputting the above business into the above time generation model to obtain the first queuing duration at the current time corresponding to the above business, the above method further includes the following steps: obtaining the outlet information of multiple above bank outlets, where the above outlet information at least includes one or more of the queuing time of the above bank outlet, the business type handled by the above bank outlet, the user flow of the above bank outlet, the counter usage situation of the above bank outlet, the location information of the above bank outlet, the traffic information of the above bank outlet, the usage rate of the self-service equipment of the above bank outlet; constructing a policy generation model, where the above policy generation model is trained using multiple sets of training data, and each set of training data in the above multiple sets of training data includes the historical outlet information obtained within a historical time period, the historical scheduling policy corresponding to the above historical outlet information, where the above historical scheduling policy is information on the migration of historical services and historical resource allocation among multiple above bank outlets; inputting the above outlet information into the above policy generation model to obtain the scheduling policy corresponding to the above outlet information, where the above scheduling policy is used to provide resource scheduling suggestions to balance the queuing time.

[0144] In this solution, not only can the queuing duration be intelligently predicted, but also a scheduling strategy can be generated to achieve dynamic optimization of bank resources, effectively balancing the queuing time at each branch, reducing customer waiting time, and improving the operational efficiency of the bank. In addition, by considering external factors, the prediction and scheduling are made closer to the actual situation, enhancing the overall practicality and flexibility of the solution. In practical applications, this approach can significantly improve customer satisfaction with bank services, reduce customer churn, and at the same time help bank managers achieve refined management and enhance the overall competitiveness of the bank.

[0145] Specifically, real-time and historical data of multiple bank branches are collected, including queuing time, types of services handled, customer flow, counter usage, location information, traffic information, and the usage rate of self-service equipment, etc. This information is crucial for understanding the operational status of each branch and customer behavior.

[0146] Multiple sets of historical training data are used, with each set containing branch information and corresponding historical scheduling strategies (such as historical business migration and resource allocation information) within a historical time period. The model is trained through deep learning or machine learning algorithms to enable it to predict the optimal resource scheduling strategy based on the current branch information to balance the queuing time.

[0147] Suppose at 15:00 on Friday afternoon, the real-time information of three branches within the bank area is collected:

[0148] Branch A: Queuing time is 15 minutes, mainly handling cash deposit and withdrawal services, medium customer flow, counter usage rate is 80%, located in the commercial center, convenient transportation, self-service equipment usage rate is 60%.

[0149] Branch B: Queuing time is 30 minutes, mainly handling loan consultation services, high customer flow, counter usage rate is 95%, located near the residential area, average transportation, self-service equipment usage rate is 70%.

[0150] Branch C: Queuing time is 5 minutes, mainly handling account management services, low customer flow, counter usage rate is 50%, located in the office area, convenient transportation, self-service equipment usage rate is 80%.

[0151] The strategy generation model is trained based on historical data, which includes the above information of each branch within a specific historical time period (such as the past year) and the actual scheduling strategies adopted (such as adding counter staff, adjusting the business processing order, guiding customers to use self-service equipment, etc.). Suppose the historical data shows that during high customer flow periods, adding counter staff can effectively shorten the queuing time, and guiding customers to use self-service equipment can improve service efficiency.

[0152] When the system inputs the currently collected branch information into the strategy generation model, the model predicts the following scheduling strategy suggestions:

[0153] Branch A: It is recommended to add 1 cash teller and recommend customers to come for business during off-peak hours through the application.

[0154] Branch B: It is recommended to add 2 loan consulting tellers and guide some customers to make appointments in advance through the online reservation system to reduce on-site queuing.

[0155] Branch C: It is recommended to maintain the status quo. However, due to the low utilization rate of the counter, some employees can be considered to be assigned to Branch B to support its high-traffic needs.

[0156] In this embodiment, specific resource scheduling and customer guidance strategies are recommended. For example, the queuing time at Branch A is expected to be reduced from 15 minutes to 12 minutes, the queuing time at Branch B is reduced from 30 minutes to 25 minutes, and the queuing time at Branch C remains at 5 minutes. The implementation of these strategies helps to balance the queuing time among branches, improve overall customer satisfaction and bank operation efficiency.

[0157] By constructing a strategy generation model and adjusting resource allocation in real time, the bank can better respond to the dynamic needs of each branch and ensure efficient and fast customer service during high-traffic periods. At the same time, through customer guidance and the recommended use of self-service equipment, the bank can relieve the pressure on the counter and further optimize the queuing time.

[0158] Specifically, the key point of the solution of this application is to propose using the multi-order moments of the time distribution to predict the queuing time risk of business processing in view of the non-normal characteristics of the business processing time distribution.

[0159] Specifically, compared with the traditional branch recommendation method that only considers time, the solution of this application adds the consideration of the queuing time distribution in the non-normal case and introduces the characteristic parameters of multi-order moments, optimizes the queuing duration prediction method for customers under different risk acceptance levels, and establishes a queuing duration prediction model based on multi-order moments, where the number of moments in the multi-order moments can be freely adjusted. The solution of this application only deduces to the fourth-order moment, but it does not mean that it is only applicable to the first four-order moments and can continue to be deduced further.

[0160] Specifically, according to the characteristic that the business processing time distributions of different businesses are different, the solution of this application establishes a total percentile duration effect prediction model for business processing considering the business type.

[0161] Specifically, customers can input their own urgency of business processing as a parameter to obtain the queuing duration under the acceptable time risk level, which is rarely explored in existing research.

[0162] Specifically, the solution of the present application starts from the phenomenon that the queuing time distributions of outlets of different business types have significant differences. When considering the queuing duration of outlets, it extends from the original first moment (average value) to multiple moments, more precisely depicting the distribution characteristics of the queuing time, and giving a specific description of the queuing time risk of customers. Based on this model, customers can more accurately estimate their own queuing duration according to their acceptable risks.

[0163] Specifically, the solution of the present application does not introduce a large number of input parameters. However, due to the accurate analysis of the queuing duration distribution, it can achieve the same effect as a large-parameter model, and because of the small number of input parameters, the total percentile duration prediction model for business processing considering business types established accordingly has the characteristic of fast model training.

[0164] The embodiment of the present application also provides a device for determining the bank queuing duration. It should be noted that the device for determining the bank queuing duration in the embodiment of the present application can be used to execute the method for determining the bank queuing duration provided in the embodiment of the present application. This device is used to implement the above-mentioned embodiments and preferred implementation manners, and those that have been described will not be repeated. As used hereinafter, the term "module" can be a combination of software and / or hardware that can achieve a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware is also possible and contemplated.

[0165] The following introduces the device for determining the bank queuing duration provided in the embodiment of the present application.

[0166] Figure 5 is a structural block diagram of a device for determining the bank queuing duration according to the embodiment of the present application. As Figure 5 shown, the device includes:

[0167] A first acquisition unit 10, configured to acquire the business to be processed;

[0168] A first construction unit 20, configured to construct a time generation model, where the time generation model is trained by using a multi-group of training data through an LSTM algorithm and a BP algorithm. Each group of the multi-group of training data includes historical services obtained within a historical time period and the historical queuing duration corresponding to the historical services. The historical queuing duration is a comprehensive representation of the distribution characteristics of time calculated based on the average value, variance, skewness, and kurtosis of multiple historical actual queuing durations, and the historical actual queuing duration is the queuing duration actually collected in the pre-collected historical time period;

[0169] The first processing unit 30 is configured to input the above-mentioned service into the above-mentioned time generation model to obtain a first queuing duration at the current time corresponding to the above-mentioned service, where the above-mentioned first queuing duration is used to be pushed to the user to improve the user experience and / or the bank branch adjusts the service arrangement according to the above-mentioned first queuing duration.

[0170] Through this embodiment, based on the phenomenon that there are significant differences in the queuing time distributions of different service types at bank branches, when considering the queuing duration at bank branches, it is extended from the original first moment (average value) to multiple moments, more precisely depicting the distribution characteristics of the queuing time. Using the LSTM-BP neural network combined model, this model combines the time series processing ability of the long short-term memory network (LSTM) and the optimization ability of the backpropagation neural network (BP), and can more accurately determine the queuing duration of bank branches based on historical queuing data and service types.

[0171] In the specific implementation process, the above-mentioned device further includes a calculation unit, and the calculation unit is configured to calculate the above-mentioned historical queuing duration according to a target formula before constructing the time generation model, where the above-mentioned target formula is:

[0172]

[0173] PTT(ρ) represents the above-mentioned historical queuing duration, μ represents the above-mentioned average value, σ 2 represents the above-mentioned variance, S represents the above-mentioned skewness, K represents the above-mentioned kurtosis, and p represents the percentile of the time distribution.

[0174] In this solution, through the above formula, the distribution characteristics of the queuing duration can be captured more accurately, making the prediction of the model more accurate, so as to provide a more effective decision-making basis in practical applications, such as bank peak period management, resource allocation optimization, etc.

[0175] In some embodiments, the above-mentioned device further includes a second acquisition unit and an update unit. The second acquisition unit is configured to acquire the number of people queuing at the above-mentioned bank branch after inputting the above-mentioned service into the above-mentioned time generation model to obtain a first queuing duration at the current time corresponding to the above-mentioned service; the update unit is configured to update the above-mentioned first queuing duration according to the above-mentioned number of people queuing to obtain a second queuing duration, where the above-mentioned first queuing duration is extended when the above-mentioned number of people queuing is greater than a preset number threshold, and the above-mentioned first queuing duration is shortened when the above-mentioned number of people queuing is less than the above-mentioned preset number threshold, where the step size for adjusting the above-mentioned first queuing duration is positively correlated with the difference between the above-mentioned first queuing duration and the above-mentioned preset number threshold.

[0176] In this solution, by updating the queuing duration in real time, it can dynamically reflect the actual situation of bank branches, help users make more reasonable waiting time estimates, and at the same time provide real-time resource scheduling requirements for banks to ensure service quality and efficiency.

[0177] In the specific implementation process, the second acquisition unit includes an acquisition module and a determination module. The acquisition module is used to acquire information related to the number of people, where the information related to the number of people at least includes one or more of the type of the above-mentioned business, the business volume of the above-mentioned bank branch, the weather, and the social network score of the above-mentioned bank branch; the determination module is used to use the LSTM algorithm to determine the above-mentioned queuing number according to the above-mentioned information related to the number of people, where there is a positive correlation between the complexity of the type of the above-mentioned business and the above-mentioned queuing number, there is a positive correlation between the above-mentioned business volume and the above-mentioned queuing number, there is a negative correlation between the severity of the above-mentioned weather and the above-mentioned queuing number, and there is a positive correlation between the above-mentioned social network score and the above-mentioned queuing number.

[0178] In this solution, multiple factors can be comprehensively considered to improve the accuracy of prediction. Especially during special periods such as weather changes and holidays, it can more accurately predict the queuing number and provide more reliable information for banks and users.

[0179] In some embodiments, the above-mentioned device further includes a third acquisition unit and a determination unit. The third acquisition unit is used to acquire user information after inputting the above-mentioned business into the above-mentioned time generation model to obtain the first queuing duration at the current time corresponding to the above-mentioned business, where the above-mentioned user information at least includes the above-mentioned bank branches visited by the user, the time period when the user handles the business, and the distances between the user and all the above-mentioned bank branches; the determination unit is used to use an optimization algorithm to determine the optimal above-mentioned bank branch according to the above-mentioned first queuing duration and the above-mentioned user information, where the optimization algorithm includes one or more of gradient optimization, adaptive estimation, whale optimization, and greedy algorithm.

[0180] In this solution, by intelligently recommending the optimal bank branch, it not only reduces the waiting time of users, but also optimizes the customer flow distribution of bank branches, avoids overcrowding in some branches, and improves the overall service quality and efficiency.

[0181] In the specific implementation process, the above-mentioned device further includes a second construction unit and a display unit. The second construction unit is used to construct a virtual queue according to all the above-mentioned businesses handled by users and the first queuing durations of all the above-mentioned businesses after inputting the above-mentioned business into the above-mentioned time generation model to obtain the first queuing duration at the current time corresponding to the above-mentioned business; the display unit is used to send the above-mentioned virtual queue to the user terminal and display the remaining number of people and the remaining queuing time on the above-mentioned user terminal.

[0182] In this solution, the construction of the virtual queue not only enables users to clearly understand their queuing positions, but also reduces on-site waiting through remote services, appointment handling, etc., improving the convenience and satisfaction of services.

[0183] In some embodiments, the above device further includes a fourth acquisition unit, a third construction unit, and a second processing unit. The fourth acquisition unit is configured to, after inputting the above service into the above time generation model to obtain the first queuing duration of the current time corresponding to the above service, acquire the network point information of multiple above bank branches, where the above network point information includes at least one or more of the queuing time of the above bank branch, the service types handled by the above bank branch, the user flow of the above bank branch, the counter usage of the above bank branch, the location information of the above bank branch, the traffic information of the above bank branch, and the usage rate of the self-service equipment of the above bank branch; the third construction unit is configured to construct a strategy generation model, where the above strategy generation model is obtained by training using multiple groups of training data, and each group of the above multiple groups of training data includes historical network point information obtained within a historical time period and the historical scheduling strategy corresponding to the above historical network point information, where the above historical scheduling strategy is information on the migration of historical services and historical resource allocation among multiple above bank branches; the second processing unit is configured to input the above network point information into the above strategy generation model to obtain the scheduling strategy corresponding to the above network point information, where the above scheduling strategy is used to provide resource scheduling suggestions to balance the queuing time.

[0184] In this solution, not only can the queuing duration be intelligently predicted, but also a scheduling strategy can be generated to achieve dynamic optimization and allocation of bank resources, effectively balancing the queuing times of each network point, reducing user waiting, and improving the operating efficiency of the bank. In addition, by considering external factors, the prediction and scheduling are closer to the actual situation, enhancing the overall practicality and flexibility of the solution. In practical applications, this method can significantly improve the customer satisfaction of bank services, reduce user churn, and at the same time help bank managers achieve refined management and enhance the overall competitiveness of the bank.

[0185] The above device for determining the queuing duration of a bank includes a processor and a memory. The first acquisition unit, the first construction unit, and the first processing unit, etc. are all stored in the memory as program units, and the corresponding functions are implemented by the processor executing the above program units stored in the memory. The above modules are all located in the same processor; or, the above each module is located in different processors in any combination form.

[0186] The processor contains a kernel, and the kernel retrieves the corresponding program units from the memory. One or more kernels can be set, and by adjusting the kernel parameters, the problem that the queuing duration of bank branches cannot be accurately predicted in the prior art can be solved.

[0187] The memory may include non-permanent memory in a computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0188] Embodiments of the present invention provide a computer-readable storage medium, where the computer-readable storage medium stores a program, and when the program runs, it controls the device where the computer-readable storage medium is located to execute the method for determining the bank queuing duration.

[0189] Embodiments of the present invention provide a processor, where the processor is used to run a program, and when the program runs, it executes the method for determining the bank queuing duration.

[0190] Embodiments of the present invention provide a device, which includes a processor, a memory, and a program stored on the memory and executable on the processor. When the processor executes the program, it implements at least the steps of the method for determining the bank queuing duration. The device herein may be a server, a PC, a PAD, a mobile phone, etc.

[0191] A computer program product includes a non-volatile computer-readable storage medium, where the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the steps of the method for determining the bank queuing duration in various embodiments of the present application.

[0192] The present application also provides a system for determining the bank queuing duration, which includes one or more processors, a memory, and one or more programs. Among them, the one or more programs are stored in the memory and are configured to be executed by the one or more processors. The one or more programs include those for executing any one of the methods for determining the bank queuing duration.

[0193] Obviously, those skilled in the art should understand that the above-mentioned modules or steps of the present invention can be implemented by a general-purpose computing device. They can be concentrated on a single computing device or distributed on a network composed of multiple computing devices. They can be implemented by program codes executable by the computing device. Thus, they can be stored in a storage device and executed by the computing device. And in some cases, the steps shown or described herein can be executed in a different order, or they can be separately fabricated into individual integrated circuit modules, or multiple modules or steps among them can be fabricated into a single integrated circuit module for implementation. In this way, the present invention is not limited to any specific combination of hardware and software.

[0194] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) that contain computer-usable program code.

[0195] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in Figure 1 one or more of the processes Figure 1 or multiple processes and / or blocks.

[0196] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that realizes the functions specified in Figure 1 one or more of the processes Figure 1 or multiple processes and / or blocks.

[0197] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for realizing the functions specified in Figure 1 one or more of the processes Figure 1 or multiple processes and / or blocks.

[0198] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and a memory.

[0199] The memory may include non-permanent memory in the computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash memory (flash RAM). The memory is an example of a computer-readable medium.

[0200] A computer-readable medium includes both permanent and non-permanent, removable and non-removable media and can implement information storage by any method or technology. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory, or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD), or other optical storage, magnetic cassette tapes, disk storage, or other magnetic storage devices, or any other non-transitory medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer-readable medium does not include transitory computer-readable media such as modulated data signals and carrier waves.

[0201] It should also be noted that the term "comprising", "including" or any other variation thereof is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that comprises the element.

[0202] The foregoing is only a preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

Claims

1. A method for determining the queuing time in a bank, characterized in that, Including: Obtain the business to be processed; Construct a time generation model, where the time generation model is trained using multiple sets of training data through the LSTM algorithm and the BP algorithm. Each set of training data in the multiple sets of training data includes historical services obtained within a historical time period and the corresponding historical queuing duration of the historical services. The historical queuing duration is a comprehensive representation of the distribution characteristics of time calculated based on the average value, variance, skewness, and kurtosis of multiple historical actual queuing durations, and the historical actual queuing duration is the queuing duration of the actual historical time period collected in advance; Input the service into the time generation model to obtain the first queuing duration at the current time corresponding to the service, where the first queuing duration is used to be pushed to the user to improve the user experience and / or the bank branch adjusts the service arrangement according to the first queuing duration.

2. The method according to claim 1, characterized in that, Before constructing the time generation model, the method further includes: Calculate the historical queuing duration according to the target formula, where the target formula is: PTT(ρ) represents the historical queuing duration, μ represents the average value, and σ 2 represents the variance, S represents the skewness, K represents the kurtosis, and p represents the percentile of the time distribution.

3. The method according to claim 1, characterized in that After inputting the service into the time generation model to obtain the first queuing duration at the current time corresponding to the service, the method further includes: Obtain the number of people queuing at the bank branch; Update the first queuing duration according to the number of people queuing to obtain the second queuing duration. When the number of people queuing is greater than the preset number threshold, the first queuing duration is extended; when the number of people queuing is less than the preset number threshold, the first queuing duration is shortened. The step size for adjusting the first queuing duration is positively correlated with the difference between the first queuing duration and the preset number threshold.

4. The method according to claim 3, wherein Obtaining the number of people queuing at the bank branch includes: Obtain the number-related information, where the number-related information includes at least one or more of the type of the service, the business volume of the bank branch, the weather, and the social network score of the bank branch; Adopt the LSTM algorithm to determine the number of people queuing according to the number-related information. The complexity of the type of the service is positively correlated with the number of people queuing, the business volume is positively correlated with the number of people queuing, the severity of the weather is negatively correlated with the number of people queuing, and the social network score is positively correlated with the number of people queuing.

5. The method according to claim 1, characterized in that After inputting the service into the time generation model to obtain the first queuing duration at the current time corresponding to the service, the method further includes: Obtain user information, where the user information includes at least the bank branches visited by the user, the time period when the user processes the service, and the distances between the user and all the bank branches; Adopt an optimization algorithm to determine the optimal bank branch according to the first queuing duration and the user information, where the optimization algorithm includes one or more of gradient optimization, adaptive estimation, whale optimization, and greedy algorithm.

6. The method according to claim 1, wherein After inputting the service into the time generation model to obtain the first queuing duration at the current time corresponding to the service, the method further includes: Construct a virtual queue according to the services handled by all users and the first queuing duration of all the services. Send the virtual queue to the user terminal and display the remaining number of people and the remaining queuing time on the user terminal.

7. The method according to claim 1, characterized in that, After inputting the service into the time generation model to obtain the first queuing duration at the current time corresponding to the service, the method further includes: Obtain the branch information of multiple bank branches, where the branch information includes at least one or more of the queuing time of the bank branch, the service types handled by the bank branch, the user flow of the bank branch, the counter usage of the bank branch, the location information of the bank branch, the traffic information of the bank branch, and the usage rate of the self-service equipment of the bank branch. Construct a policy generation model, where the policy generation model is trained using multiple sets of training data, and each set of training data in the multiple sets of training data includes historical branch information obtained within a historical time period and the historical scheduling policy corresponding to the historical branch information, where the historical scheduling policy is information on the migration of historical services and historical resource allocation among multiple bank branches. Input the branch information into the policy generation model to obtain the scheduling policy corresponding to the branch information, where the scheduling policy is used to provide resource scheduling suggestions to balance the queuing time.

8. A device for determining the queuing duration in a bank, characterized in that, Include: A first acquisition unit for acquiring the service to be handled. A first construction unit for constructing a time generation model, where the time generation model is trained using multiple sets of training data through the LSTM algorithm and the BP algorithm, and each set of training data in the multiple sets of training data includes historical services obtained within a historical time period and the historical queuing duration corresponding to the historical services, where the historical queuing duration is a comprehensive representation of the distribution characteristics of time calculated based on the average value, variance, skewness, and kurtosis of multiple historical actual queuing durations, and the historical actual queuing duration is the queuing duration actually collected in the pre-collected historical time period. A first processing unit for inputting the service into the time generation model to obtain the first queuing duration at the current time corresponding to the service, where the first queuing duration is used to be pushed to the user to improve the user experience and / or the bank branch adjusts the service arrangement according to the first queuing duration.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method for determining the bank queuing duration according to any one of claims 1 to 7.

10. A bank queuing duration determination system, characterized in that, Include: One or more processors, a memory, and one or more programs, where the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include the method for determining the bank queuing duration according to any one of claims 1 to 7.

Images