A bank branch teller deployment method and device
By performing feature engineering and prediction model processing on bank branch data, accurately predicting the daily workload, the problem of redundancy or insufficient personnel in teller scheduling is solved, and the resource utilization efficiency of bank branch is improved.
Patent Information
- Application Number
- CN202111091118.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-09-17
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2041-09-17
AI Technical Summary
It is difficult for bank branch tellers to accurately estimate customer service needs, resulting in redundant or insufficient personnel and inefficient efficiency.
By obtaining bank branch data for feature engineering, using prediction models to predict the daily workload and determine the number of tellers required.
It improves the accuracy of teller allocation, rationally utilizes human resources, and improves scheduling efficiency.
Smart Images

Figure CN113807947B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of big data technology, and in particular to a method and device for allocating tellers at bank outlets. Background Art
[0002] At present, the counter channels of bank branches are important channels for banks to initiate offline business. The number of counters in bank branches is usually fixed, and bank branches arrange tellers to handle offline business through shifts.
[0003] In the prior art, the teller scheduling of bank branches is mainly done by the person in charge of the bank branch based on experience. It is difficult to accurately estimate the customer service demand of each working day of the bank branch through scheduling based on human experience, resulting in frequent staff redundancy or shortage when scheduling tellers, which makes the human resources of the branch not reasonably utilized and the scheduling efficiency low. Summary of the invention
[0004] In view of the problems in the prior art, an embodiment of the present invention provides a method and device for allocating tellers at bank outlets, which can at least partially solve the problems in the prior art.
[0005] On the one hand, the present invention provides a method for deploying bank branch tellers, comprising:
[0006] Obtain bank branch data;
[0007] Performing feature engineering processing on the bank branch data to obtain branch feature data;
[0008] Based on the branch feature data and the bank branch workload prediction model, a single-day predicted workload is obtained; wherein the bank branch workload prediction model is obtained by training based on bank branch training data;
[0009] The number of bank branch tellers required is determined based on the daily predicted workload.
[0010] On the other hand, the present invention provides a bank branch teller deployment device, comprising:
[0011] The first acquisition module is used to acquire bank branch data;
[0012] A first acquisition module is used to perform feature engineering processing on the bank branch data to obtain branch feature data;
[0013] A prediction module, used to obtain a daily predicted workload based on the branch feature data and a bank branch workload prediction model; wherein the bank branch workload prediction model is obtained by training based on bank branch training data;
[0014] A determination module is used to determine the number of bank branch tellers required based on the daily predicted workload.
[0015] On the other hand, the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the bank branch teller deployment method described in any of the above embodiments are implemented.
[0016] On the other hand, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the bank branch teller deployment method described in any of the above embodiments.
[0017] The bank branch teller deployment method and device provided in the embodiment of the present invention can obtain bank branch data, perform feature engineering processing on the bank branch data, obtain branch feature data, obtain a single-day predicted workload based on the branch feature data and a bank branch workload prediction model, and determine the number of bank branch tellers required based on the single-day predicted workload. The bank branch workload prediction model can more accurately predict the single-day predicted workload, thereby enabling more reasonable deployment of bank branch tellers, thereby improving the accuracy of bank branch teller deployment. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:
[0019] Figure 1 It is a flow chart of a method for allocating bank branch tellers provided in the first embodiment of the present invention.
[0020] Figure 2 It is a flow chart of a method for allocating bank branch tellers provided in a second embodiment of the present invention.
[0021] Figure 3 It is a flow chart of a method for allocating bank branch tellers provided in a third embodiment of the present invention.
[0022] Figure 4 It is a flow chart of a method for allocating bank branch tellers provided in a fourth embodiment of the present invention.
[0023] Figure 5 It is a structural diagram of a bank branch teller's deployment device provided in the fifth embodiment of the present invention.
[0024] Figure 6 It is a structural diagram of a bank branch teller's deployment device provided in a sixth embodiment of the present invention.
[0025] Figure 7 It is a structural diagram of a bank branch teller's deployment device provided by the seventh embodiment of the present invention.
[0026] Figure 8 It is a structural diagram of a bank branch teller's deployment device provided in the eighth embodiment of the present invention.
[0027] Fig. 9 It is a schematic diagram of the physical structure of an electronic device provided by the ninth embodiment of the present invention. DETAILED DESCRIPTION
[0028] In order to make the purpose, technical scheme and advantages of the embodiments of the present invention more clear, the embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention. It should be noted that, in the absence of conflict, the embodiments in this application and the features in the embodiments can be combined with each other arbitrarily.
[0029] Figure 1 FIG. 1 is a flow chart of a method for allocating bank branch tellers provided by a first embodiment of the present invention. Figure 1 As shown, the method for deploying bank branch tellers provided by the embodiment of the present invention includes:
[0030] S101. Obtain bank branch data;
[0031] Specifically, the server can obtain bank branch data, which may include basic branch data, branch feature data, branch other data, and branch historical data within a preset time period, which is set according to actual needs and is not limited in the embodiment of the present invention. The preset time period is set according to actual needs and is not limited in the embodiment of the present invention.
[0032] Among them, the basic data of the outlets may include the daily counter workload, the daily smart device business volume, and the number of tellers at the bank outlets every day in the past two years. The smart devices include but are not limited to ATMs, self-service payment machines and other devices. The characteristic data of the outlets may include the distance between the bank outlets and the commercial area, the distance between the outlets and the factories, the distance between the bank outlets and the schools, the number of commercial areas, the number of factories, the number of schools, etc. The historical data of the outlets may include the number of customers in a specific time period, the time customers wait to handle business in a specific time period, the time customers handle business in a specific time period, etc. The specific time period includes but is not limited to different time periods such as years, quarters, months, weeks, and days. Other data of the outlets may include the type of outlets, the type of the day, etc. The type of outlets includes but is not limited to business departments, branches, etc. The type of the day includes but is not limited to holidays, paydays, social security days, credit card repayment days, etc.
[0033] S102, performing feature engineering processing on the bank branch data to obtain branch feature data;
[0034] Specifically, after obtaining the bank branch data, the server performs feature engineering processing on the bank branch data to obtain branch feature data. By performing feature engineering processing on the bank branch data, features can be extracted from the bank branch data to perform subsequent single-day forecast workload prediction. The specific process of feature engineering processing is set according to actual needs and is not limited in the embodiment of the present invention.
[0035] For example, the numerical values of basic data and characteristic data of outlets can be directly retained; the type data included in other data of outlets can be one-hot encoded; and different weight coefficients can be set for different indicators for historical data of outlets for summary calculation.
[0036] S103, based on the branch feature data and the bank branch workload prediction model, obtain a single-day predicted workload; wherein the bank branch workload prediction model is obtained by training based on bank branch training data;
[0037] Specifically, the server inputs the branch feature data into the bank branch workload prediction model, and after being processed by the bank branch workload prediction model, the daily predicted workload of the bank branch can be output, and the daily predicted workload indicates the workload that the bank branch needs to complete in a single day. The bank branch workload prediction model is obtained by training based on the bank branch training data. The specific training process of the bank branch workload prediction model is described below and will not be repeated here.
[0038] S104: Determine the number of bank branch tellers required based on the daily predicted workload.
[0039] Specifically, after obtaining the predicted daily workload, the server can determine the number of bank branch tellers required to be configured based on the predicted daily workload, and can divide the predicted daily workload by the standard workload of a single bank branch teller to obtain the number of bank branch tellers required. The standard workload of a single bank branch teller is set based on actual experience and is not limited in the embodiments of the present invention.
[0040] For example, the daily workload of each teller in the bank branch within a period of time can be counted, and the average daily workload of each teller within a period of time can be calculated as the standard workload. The period of time is set according to actual needs and is not limited in the embodiment of the present invention.
[0041] The method for deploying bank branch tellers provided in the embodiment of the present invention can obtain bank branch data, perform feature engineering processing on the bank branch data, obtain branch feature data, obtain a single-day predicted workload based on the branch feature data and a bank branch workload prediction model, and determine the number of bank branch tellers required based on the single-day predicted workload. The bank branch workload prediction model can more accurately predict the single-day predicted workload, thereby enabling more reasonable deployment of bank branch tellers, thereby improving the accuracy of bank branch teller deployment.
[0042] Figure 2 FIG. 1 is a flow chart of a method for allocating bank branch tellers provided by a second embodiment of the present invention. Figure 2 As shown, on the basis of the above embodiments, further, the step of obtaining the bank branch workload prediction model based on the bank branch training data training includes:
[0043] S201, obtaining the bank branch training data;
[0044] Specifically, bank branch-related data may be collected, and bank branch training data may be screened out therefrom. The server may obtain the bank branch training data. The bank branch training data may include basic branch data, branch feature data, branch other data, and branch historical data within a set time period, which may be set according to actual needs and are not limited in the embodiment of the present invention. The set time period may be set according to actual needs and are not limited in the embodiment of the present invention.
[0045] S202, performing feature engineering processing on the bank branch training data to obtain branch training feature data; wherein the branch training feature data includes multiple groups of training data, and each group of training data is labeled with a corresponding single-day historical workload;
[0046] Specifically, after obtaining the bank branch training data, the server performs feature engineering processing on the bank branch training data to obtain branch training feature data. The branch training feature data can be divided into multiple groups of training data by day, and each group of training data is labeled with the corresponding single-day historical workload. The single-day historical workload is the single-day counter workload of the bank branch, which can be obtained from the basic data of the branch. Among them, the number of groups of training data included in the bank branch training data is set according to actual needs, and the embodiment of the present invention does not limit it.
[0047] S203. Based on each group of training data, the daily historical workload corresponding to each group of training data, and the hybrid model, the bank branch workload prediction model is trained; wherein the hybrid model includes multiple original models; and the loss function used to determine whether the model training is terminated is constructed based on the average waiting time of customers.
[0048] Specifically, each group of training data and the daily historical workload corresponding to each group of training data are input into the hybrid model, the hybrid model is trained, the parameters of each original model included in the hybrid model are continuously updated, and the average waiting time of customers is calculated through the output of each original model and the loss function during the training process until the average waiting time of customers obtained by calculation converges, and the training of the model is stopped. After the trained hybrid model passes the verification, the workload prediction model of the bank branch is obtained. Wherein, the hybrid model includes multiple original models. During the model training process, each group of training data and the daily historical workload corresponding to each group of training data are respectively input into each original model to train each original model. Each original model corresponds to a model training algorithm. The model training algorithm includes but is not limited to a generalized linear regression algorithm, a tree-based regression algorithm, a time series algorithm, etc., which is selected according to actual needs and is not limited in the embodiment of the present invention.
[0049] The hybrid model includes various original models, which can be set based on experience; and can also be obtained by screening from multiple initial models.
[0050] Figure 3 FIG. 1 is a flow chart of a method for allocating bank branch tellers provided in a third embodiment of the present invention. Figure 3 As shown, on the basis of the above embodiments, further, each original model included in the hybrid model is obtained by screening multiple initial models, and the specific screening steps include:
[0051] S301, based on each group of training data, the single-day historical workload corresponding to each group of training data, and each initial model, train each model to be screened;
[0052] Specifically, the server inputs each set of training data and the single-day historical workload corresponding to each set of training data into each initial model, trains each initial model, and uses each trained and verified initial model as a model to be screened. Each initial model corresponds to a model training algorithm.
[0053] S302, based on the network point historical feature data and each model to be screened, obtaining a prediction value corresponding to each model to be screened; wherein the network point historical feature data is obtained from the network point training feature data;
[0054] Specifically, the server inputs the branch historical feature data into each model to be screened, and after being processed by each model to be screened, the server can output the prediction value corresponding to each model to be screened. The prediction value corresponding to the model to be screened is the daily workload of the bank branch predicted by the model to be screened. The branch historical feature data is obtained from the branch training feature data, and part of the branch training feature data can be selected from the branch training feature data as the branch historical feature data. It is understandable that the time interval of the time period corresponding to the branch historical feature data is the same as the time interval of the set time period corresponding to the branch training feature data.
[0055] S303, based on the predicted value corresponding to each model to be screened, the single-day historical workload corresponding to each model to be screened, and the residual sum of squares formula, calculate and obtain the residual sum of squares corresponding to each model to be screened;
[0056] Specifically, the server inputs the predicted value corresponding to each model to be screened and the single-day historical workload corresponding to each model to be screened into the residual sum of squares formula, and can calculate the residual sum of squares corresponding to each model to be screened. The residual sum of squares formula is preset.
[0057] For example, the residual sum of squares formula is The residual sum of squares corresponding to the RSS model to be screened, y i represents the i-th predicted value corresponding to the model to be screened, It represents the i-th single-day historical workload corresponding to the model to be screened. The i-th predicted value and the i-th single-day historical workload correspond to the same day. n represents the number of predicted values. i is a positive integer and i is less than or equal to n.
[0058] S304, based on the residual sum of squares corresponding to each model to be screened and the Akaike information criterion formula, calculate and obtain the AIC value corresponding to each model to be screened;
[0059] Specifically, the server inputs the residual sum of squares corresponding to each model to be screened into the Akaike information criterion (AIC) formula, and can calculate the AIC value corresponding to each model to be screened. The AIC formula is preset.
[0060] For example, the AIC formula is Among them, AIC represents the AIC value, k represents the number of parameters of the model to be screened, RSS represents the residual sum of squares corresponding to the model to be screened, and n represents the number of predicted values corresponding to the model to be screened used to calculate the residual sum of squares corresponding to the model to be screened.
[0061] S305. Based on the AIC value corresponding to each model to be screened, screen out the original model included in the mixed model; wherein the original model uniquely corresponds to one initial model.
[0062] Specifically, the server may select multiple models to be screened from multiple models to be screened as original models included in the hybrid model based on the AIC value corresponding to each model to be screened, wherein each original model uniquely corresponds to one initial model.
[0063] Figure 4 FIG. 4 is a flow chart of a method for allocating bank branch tellers provided in a fourth embodiment of the present invention. Figure 4 As shown, on the basis of the above embodiments, further, the original model included in the mixed model is screened out based on the AIC value corresponding to each model to be screened, including:
[0064] S401, comparing the AIC values corresponding to the models to be screened, and obtaining the minimum AIC value;
[0065] Specifically, the server compares the AIC values corresponding to the models to be screened and can obtain the minimum AIC value.
[0066] For example, the server arranges the AIC values corresponding to the models to be screened in order from large to small, and the AIC value corresponding to the model to be screened that is last sorted is the minimum AIC value.
[0067] S402, calculating the difference between each other AIC value and the minimum AIC value, and obtaining the difference corresponding to each other AIC value; wherein each other AIC value refers to each AIC value except the minimum AIC value among the AIC values corresponding to each model to be screened;
[0068] Specifically, the server calculates the difference between each other AIC value and the minimum AIC value as the difference corresponding to each other AIC value, wherein the other AIC value refers to an AIC value other than the minimum AIC value among the AIC values corresponding to each model to be screened.
[0069] For example, if there are m models to be screened, there will be m AIC values corresponding to the models to be screened, there is one minimum AIC value, and there are m-1 other AIC values. The server can calculate the differences corresponding to the m-1 other AIC values.
[0070] S403, calculating the model weight of the to-be-screened model corresponding to each other AIC value according to the difference corresponding to each other AIC value and the model weight formula; wherein the model weight formula is preset;
[0071] Specifically, after obtaining the sum of the differences corresponding to each other AIC value, the server brings the differences corresponding to each other AIC value into the model weight formula, and can calculate the model weight of the to-be-screened model corresponding to each other AIC value, respectively. The model weight formula is preset.
[0072] For example, the model weight formula is:
[0073]
[0074] Among them, w i Indicates the model weight of the model to be screened corresponding to the i-th other AIC value, ΔAIC i represents the difference corresponding to the i-th other AIC values, e represents a natural constant, R represents the total number of models to be screened corresponding to other AIC values, i is a positive integer, and i is less than or equal to R.
[0075] S404. Obtain the original model included in the hybrid model according to the model weights of the to-be-screened models corresponding to the other AIC values.
[0076] Specifically, the larger the model weight, the more effective the model to be screened corresponding to the model weight. The server can screen out multiple models to be screened as original models included in the hybrid model according to the model weights of the models to be screened corresponding to each other AIC value.
[0077] For example, the model weights of the models to be screened corresponding to other AIC values may be ranked from large to small, and the models to be screened ranked third are taken as the original models included in the hybrid model.
[0078] On the basis of the above embodiments, further, the loss function is:
[0079]
[0080] Among them, w q represents the average waiting time of customers, 1 / λ represents the average time interval of customer arrival, μ represents the maximum number of customers served by a bank branch per unit time, and L s Represents the average demand for the service desk.
[0081] Specifically, 1 / λ represents the average time interval between customer arrivals, which can be obtained based on the historical data of the interval between customer number collection at bank branches. μ represents the maximum number of customers served by a bank branch within a unit time, which can be set according to the actual situation of the bank branch. s represents the average demand for service desks. Since each service desk will be equipped with a teller, the demand for service desks is equal to the number of tellers required at bank branches. During the model training process, each original model will output the corresponding prediction value. The quotient of each prediction value divided by the standard workload of a single bank branch teller is calculated as the service desk demand corresponding to each original model. Then, the average value of the service desk demand corresponding to each original model is calculated as the average service desk demand L. s The server is configured by the formula Can calculate the average customer waiting time w q During the model training process, the server will calculate the average waiting time of customers w q Compared with the set threshold, if w q If it is less than the set threshold, it means that the average waiting time of the customers has converged and the model training can be stopped.
[0082] On the basis of the above embodiments, further, the obtaining of bank branch data includes:
[0083] The basic data of the bank branch, the characteristic data of the bank branch, other data of the bank branch and the historical data of the bank branch within a preset time period are obtained as the bank branch data.
[0084] Specifically, the server may obtain basic data, characteristic data, other data and historical data of the bank branch within a preset time period as bank branch data. The preset time period is set according to actual needs and is not limited in the embodiment of the present invention.
[0085] For example, the preset time period is set to 1 month, and the daily forecast workload after the current time is predicted based on the basic data of the outlets, the characteristic data of the outlets, the other data of the outlets and the historical data of the outlets one month before the current time.
[0086] It is understandable that the bank branch teller deployment method provided in the embodiment of the present invention can be applied to various bank branches, and the daily predicted workload of each bank branch can be presented through a visual chart. Since the customer financial service needs of bank branches will change over time, using the same bank branch workload prediction model all the time may cause deviations in the prediction results. The prediction accuracy of the bank branch workload prediction model can be regularly calculated. If the prediction accuracy of the bank branch workload prediction model is lower than a threshold, the bank branch workload prediction model can be retrained to obtain the bank branch workload prediction model.
[0087] Figure 5 FIG. 5 is a schematic diagram of the structure of a bank branch teller's deployment device provided by a fifth embodiment of the present invention. Figure 5 As shown, the bank branch teller allocation device provided by the embodiment of the present invention includes a first acquisition module 501, a first obtaining module 502, a prediction module 503 and a determination module 504, wherein:
[0088] The first acquisition module 501 is used to acquire bank branch data; the first acquisition module 502 is used to perform feature engineering processing on the bank branch data to obtain branch feature data; the prediction module 503 is used to obtain a single-day predicted workload based on the branch feature data and a bank branch workload prediction model; wherein the bank branch workload prediction model is obtained by training based on bank branch training data; the determination module 504 is used to determine the number of bank branch tellers required based on the single-day predicted workload.
[0089] Specifically, the first acquisition module 501 can acquire bank branch data, which may include basic branch data, branch feature data, branch other data, and branch historical data within a preset time period, which are set according to actual needs and are not limited in the embodiment of the present invention. The preset time period is set according to actual needs and is not limited in the embodiment of the present invention.
[0090] After obtaining the bank branch data, the first acquisition module 502 performs feature engineering processing on the bank branch data to obtain branch feature data. By performing feature engineering processing on the bank branch data, features can be extracted from the bank branch data to perform subsequent single-day forecast workload prediction. The specific process of feature engineering processing is set according to actual needs and is not limited in the embodiment of the present invention.
[0091] The prediction module 503 inputs the branch feature data into the bank branch workload prediction model, and after being processed by the bank branch workload prediction model, it can output the bank branch's daily predicted workload, which indicates the workload that the bank branch needs to complete in a single day. The bank branch workload prediction model is obtained by training based on the bank branch training data.
[0092] After obtaining the predicted daily workload, the determination module 504 can determine the number of bank branch tellers required to be configured based on the predicted daily workload, and can divide the predicted daily workload by the standard workload of a single bank branch teller to obtain the number of bank branch tellers required. The standard workload of a single bank branch teller is set according to actual experience and is not limited in the embodiment of the present invention.
[0093] The bank branch teller allocation device provided in the embodiment of the present invention can obtain bank branch data, perform feature engineering processing on the bank branch data to obtain branch feature data, obtain a single-day predicted workload based on the branch feature data and a bank branch workload prediction model, and determine the number of bank branch tellers required based on the single-day predicted workload. The bank branch workload prediction model can more accurately predict the single-day predicted workload, thereby enabling more reasonable allocation of bank branch tellers, thereby improving the accuracy of bank branch teller allocation.
[0094] Figure 6 : is a schematic diagram of the structure of the bank branch teller's deployment device provided by the sixth embodiment of the present invention, such as Figure 6 As shown, on the basis of the above embodiments, further, the bank branch teller allocation device provided by the embodiment of the present invention further includes a second acquisition module 505, a second acquisition module 506 and a first training module 507, wherein:
[0095] The second acquisition module 505 is used to acquire the bank branch training data; the second acquisition module 506 is used to perform feature engineering processing on the bank branch training data to obtain branch training feature data; wherein, the branch training feature data includes multiple groups of training data, and each group of training data is labeled with the corresponding single-day historical workload; the first training module 507 is used to train and obtain the bank branch workload prediction model based on each group of training data, the single-day historical workload corresponding to each group of training data, and a hybrid model; wherein, the hybrid model includes multiple original models; the loss function used to determine whether the model training is terminated is constructed based on the average waiting time of customers.
[0096] Figure 7 FIG. 7 is a schematic diagram of the structure of a bank branch teller's deployment device provided by the seventh embodiment of the present invention. Figure 7As shown, on the basis of the above embodiments, further, the bank branch teller allocation device provided by the embodiment of the present invention further includes a second training module 508, a third obtaining module 509, a first calculation module 510, a second calculation module 511 and a screening module 512, wherein:
[0097] The second training module 508 is used to train and obtain each model to be screened based on the network point training feature data, the single-day historical workload corresponding to the network point training feature data, and each initial model; the third acquisition module 509 is used to obtain the prediction value corresponding to each model to be screened based on the network point training feature data and each model to be screened; the first calculation module 510 is used to calculate the residual sum of squares corresponding to each model to be screened based on the prediction value corresponding to each model to be screened, the single-day historical workload corresponding to each model to be screened, and the residual sum of squares formula; the second calculation module 511 is used to calculate the AIC value corresponding to each model to be screened based on the residual sum of squares corresponding to each model to be screened and the Akaike information criterion formula; the screening module 512 is used to screen out the original model included in the mixed model based on the AIC value corresponding to each model to be screened; wherein, the original model uniquely corresponds to one of the initial models.
[0098] Figure 8 FIG. 8 is a schematic diagram of the structure of a bank branch teller's deployment device provided by an eighth embodiment of the present invention. Figure 8 As shown, on the basis of the above embodiments, further, the screening module 512 includes a comparison unit 5121, a first calculation unit 5122, a second calculation unit 5123 and an acquisition unit 5124, wherein:
[0099] The comparison unit 5121 is used to compare the AIC values corresponding to each model to be screened and obtain the minimum AIC value; the first calculation unit 5122 is used to calculate the difference between each other AIC value and the minimum AIC value to obtain the difference corresponding to each other AIC value; wherein each other AIC value refers to each AIC value except the minimum AIC value among the AIC values corresponding to each model to be screened; the second calculation unit 5123 calculates the model weight of the model to be screened corresponding to each other AIC value according to the difference corresponding to each other AIC value and the model weight formula; wherein the model weight formula is preset; the acquisition unit 5124 is used to obtain the original model included in the hybrid model according to the model weight of the model to be screened corresponding to each other AIC value.
[0100] The embodiments of the apparatus provided in the embodiments of the present invention can be specifically used to execute the processing flow of the above-mentioned method embodiments, and its functions are not described in detail here, and reference can be made to the detailed description of the above-mentioned method embodiments.
[0101] Fig. 9is a schematic diagram of the physical structure of an electronic device provided by an embodiment of the present invention, such as Fig. 9 As shown, the electronic device may include: a processor 901, a communication interface 902, a memory 903 and a communication bus 904, wherein the processor 901, the communication interface 902 and the memory 903 communicate with each other through the communication bus 904. The processor 901 may call the logic instructions in the memory 903 to execute the following method: obtaining bank branch data; performing feature engineering processing on the bank branch data to obtain branch feature data; obtaining a single-day predicted workload based on the branch feature data and the bank branch workload prediction model; wherein the bank branch workload prediction model is obtained by training based on the bank branch training data; and determining the number of bank branch tellers required based on the single-day predicted workload.
[0102] In addition, the logic instructions in the above-mentioned memory 903 can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a 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: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.
[0103] The present embodiment discloses a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the methods provided by the above-mentioned method embodiments, for example, including: obtaining bank branch data; performing feature engineering processing on the bank branch data to obtain branch feature data; obtaining a single-day predicted workload based on the branch feature data and a bank branch workload prediction model; wherein the bank branch workload prediction model is obtained by training based on bank branch training data; and determining the number of bank branch tellers required based on the single-day predicted workload.
[0104] The present embodiment provides a computer-readable storage medium, which stores a computer program, and the computer program enables the computer to execute the methods provided by the above-mentioned method embodiments, for example, including: obtaining bank branch data; performing feature engineering processing on the bank branch data to obtain branch feature data; obtaining a single-day predicted workload based on the branch feature data and a bank branch workload prediction model; wherein the bank branch workload prediction model is obtained by training based on bank branch training data; and determining the number of bank branch tellers required based on the single-day predicted workload.
[0105] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0106] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0107] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0108] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1The steps for the functions specified in one or more boxes.
[0109] In the description of this specification, the description with reference to the terms "one embodiment", "a specific embodiment", "some embodiments", "for example", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.
[0110] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.
Claims
1. A method for deploying bank branch tellers, characterized in that: include: Obtain bank branch data; Performing feature engineering processing on the bank branch data to obtain branch feature data; Based on the branch feature data and the bank branch workload prediction model, a single-day predicted workload is obtained; wherein the bank branch workload prediction model is obtained by training based on bank branch training data; Determine the number of bank branch tellers required based on the daily predicted workload; The step of obtaining the bank branch workload prediction model based on the bank branch training data training includes: Obtaining the bank branch training data; Performing feature engineering processing on the bank branch training data to obtain branch training feature data; wherein the branch training feature data includes multiple groups of training data, each group of training data is labeled with the corresponding single-day historical workload; Based on each group of training data, the daily historical workload corresponding to each group of training data and the hybrid model, the bank branch workload prediction model is trained; wherein the hybrid model includes multiple original models; the loss function used to determine whether the model training is terminated is constructed based on the average waiting time of customers; during the model training process, the calculated average waiting time of customers is compared with a set threshold, and if the average waiting time of customers is less than the set threshold, the model training is stopped; Wherein, the loss function is: Among them, w q represents the average waiting time of customers, 1 / λ represents the average time interval of customer arrival, μ represents the maximum number of customers served by a bank branch per unit time, and L s Represents the average demand for the service desk.
2. The method according to claim 1, characterized in that Each original model included in the hybrid model is obtained by screening multiple initial models, and the specific screening steps include: Based on the network point training feature data, the single-day historical workload corresponding to the network point training feature data, and each initial model, training is performed to obtain each model to be screened; Based on the network point training feature data and each model to be screened, obtaining a prediction value corresponding to each model to be screened; Based on the predicted value corresponding to each model to be screened, the single-day historical workload corresponding to each model to be screened, and the residual sum of squares formula, the residual sum of squares corresponding to each model to be screened is calculated; Based on the residual sum of squares corresponding to each model to be screened and the Akaike information criterion formula, the AIC value corresponding to each model to be screened is calculated; Based on the AIC value corresponding to each model to be screened, the original model included in the mixed model is screened out; wherein the original model uniquely corresponds to one initial model.
3. The method according to claim 2, characterized in that The original models included in the mixed model are screened out based on the AIC value corresponding to each model to be screened, including: Compare the AIC values corresponding to each model to be screened and obtain the minimum AIC value; Calculate the difference between each other AIC value and the minimum AIC value to obtain the difference corresponding to each other AIC value; wherein each other AIC value refers to each AIC value except the minimum AIC value among the AIC values corresponding to each model to be screened; According to the difference corresponding to each other AIC value and the model weight formula, the model weight of the to-be-screened model corresponding to each other AIC value is calculated; wherein the model weight formula is preset; According to the model weights of the models to be screened corresponding to the other AIC values, the original model included in the hybrid model is obtained.
4. The method according to claim 3, characterized in that The model weight formula is: Among them, w i Indicates the model weight of the model to be screened corresponding to the i-th other AIC value, ΔAIC i represents the i-th other AIC value, e represents a natural constant, R represents the total number of models to be screened corresponding to other AIC values, i is a positive integer, and i is less than or equal to R.
5. The method according to any one of claims 1 to 4, characterized in that: The obtaining of bank branch data includes: The basic data of the bank branch, the characteristic data of the bank branch, other data of the bank branch and the historical data of the bank branch within a preset time period are obtained as the bank branch data.
6. A bank branch teller deployment device, characterized in that: include: The first acquisition module is used to acquire bank branch data; A first acquisition module is used to perform feature engineering processing on the bank branch data to obtain branch feature data; A prediction module, used to obtain a daily predicted workload based on the branch feature data and a bank branch workload prediction model; wherein the bank branch workload prediction model is obtained by training based on bank branch training data; A determination module, used to determine the number of bank branch tellers required based on the daily predicted workload; A second acquisition module, used to acquire the bank branch training data; A second acquisition module is used to perform feature engineering processing on the bank branch training data to obtain branch training feature data; wherein the branch training feature data includes multiple groups of training data, and each group of training data is labeled with a corresponding single-day historical workload; The first training module is used to train and obtain the workload prediction model of the bank branch based on each group of training data, the single-day historical workload corresponding to each group of training data, and the hybrid model; wherein the hybrid model includes multiple original models; the loss function used to determine whether the model training is terminated is constructed based on the average waiting time of customers; during the model training process, the calculated average waiting time of customers is compared with a set threshold, and if the average waiting time of customers is less than the set threshold, the model training is stopped; Wherein, the loss function is: Among them, w q represents the average waiting time of customers, 1 / λ represents the average time interval of customer arrival, μ represents the maximum number of customers served by a bank branch per unit time, and L s Represents the average demand for the service desk.
7. The device according to claim 6, characterized in that Also includes: A second training module is used to train each model to be screened based on the network point training feature data, the single-day historical workload corresponding to the network point training feature data, and each initial model; A third obtaining module is used to obtain a prediction value corresponding to each to-be-screened model based on the network point training feature data and each to-be-screened model; The first calculation module is used to calculate the residual sum of squares corresponding to each to-be-screened model based on the predicted value corresponding to each to-be-screened model, the single-day historical workload corresponding to each to-be-screened model, and the residual sum of squares formula; The second calculation module is used to calculate the AIC value corresponding to each model to be screened based on the residual sum of squares corresponding to each model to be screened and the Akaike information criterion formula; The screening module is used to screen out the original model included in the mixed model based on the AIC value corresponding to each model to be screened; wherein the original model uniquely corresponds to one initial model.
8. The device according to claim 7, characterized in that The screening module comprises: The comparison unit is used to compare the AIC values corresponding to the models to be screened and obtain the minimum AIC value; A first calculation unit is used to calculate the difference between each other AIC value and the minimum AIC value to obtain the difference corresponding to each other AIC value; wherein each other AIC value refers to each AIC value except the minimum AIC value among the AIC values corresponding to each model to be screened; The second calculation unit calculates the model weight of the to-be-screened model corresponding to each other AIC value according to the difference corresponding to each other AIC value and the model weight formula; wherein the model weight formula is preset; The obtaining unit is used to obtain the original model included in the hybrid model according to the model weights of the to-be-screened models corresponding to each other AIC value.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.
