Method, device and equipment for determining cash replenishing strategy of teller machine, medium and program product

Abstract

The invention provides a teller machine money adding strategy determination method and device, electronic equipment, a medium and a computer program product, and can be applied to the technical field of artificial intelligence. The method for determining the cash replenishing strategy of the teller machine comprises the following steps: acquiring the withdrawal amount of the teller machine every day in the first p days; according to the daily withdrawal amount, the current-day teller machine environment information and the current-day external environment information in the previous p days, the predicted withdrawal amount of the current day is obtained by using a pre-constructed withdrawal amount prediction model, the current-day teller machine environment information and the current-day external environment information; according to the predicted withdrawal amount of the current day, the total withdrawal amount of the future m days is calculated; determining a cash replenishing strategy according to the total withdrawal amount in the future m days and the acquired inventory of the teller machine; wherein the withdrawal amount prediction model comprises an autoregression model and an XGBoost model, and the autoregression model obtains a to-be-predicted withdrawal amount of the current day; the XGBoost model obtains the compensation error withdrawal amount of the to-be-predicted withdrawal amount, and the predicted withdrawal amount is the sum of the to-be-predicted withdrawal amount and the compensation error withdrawal amount.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and more specifically, to a method, device, electronic device, medium and computer program product for determining a cash adding strategy for an ATM. Background Art

[0002] The bank's cash deposit and withdrawal services can be handled at bank branches and self-service ATMs. Today, self-service ATMs are found in communities with a relatively high concentration of people, greatly facilitating users who need to use cash. Setting up self-service ATMs in the community means that users do not have to make a separate trip to a branch and can choose to handle cash deposit and withdrawal services at a self-service ATM.

[0003] With the rapid development of mobile payment, the frequency of cash use has begun to decrease. Some self-service ATMs have a large amount of cash inventory for a long time, which wastes resources and increases operating costs. Therefore, some banks choose to reduce the cash inventory of self-service ATMs. Cash is still an indispensable part of people's daily life, especially in communities and towns with a large elderly population. During holidays or banquets, and when the Social Security Bureau issues pensions, people use cash more frequently than on weekdays. If the cash inventory of self-service ATMs is reduced blindly, the cash inventory in the self-service ATMs will not be able to meet the users' cash withdrawal needs, leaving some customers with no money to withdraw and having to go to nearby business outlets or other self-service ATMs to handle withdrawal services, which will reduce customer service satisfaction.

[0004] If the bank does not reduce the cash inventory of the self-service deposit and withdrawal machines, for the bank cash center, the cash in the self-service deposit and withdrawal machines is part of the bank's cash inventory and part of the cash for business use. If the cash inventory of the self-service deposit and withdrawal machines is kept high for a long time, it will not only increase the bank's operating risks, but also increase the operating costs caused by the bank's frequent cash additions. At the same time, the large amount of cash inventory in the cash box of the self-service deposit and withdrawal machine will cause a waste of cash resources for business use. Therefore, how to determine a cash addition strategy for ATMs that can both meet user needs and reduce the cash operating costs of self-service deposit and withdrawal machines has become a technical problem that technicians in this field need to solve urgently. Summary of the invention

[0005] In view of this, the present disclosure provides a method, device, electronic device, computer-readable storage medium and computer program product for determining an ATM cash replenishment strategy that can both meet user withdrawal requirements and reduce the cash operating costs of the ATM.

[0006] One aspect of the present disclosure provides a method for determining a cash replenishment strategy for an ATM, comprising: obtaining the daily withdrawal amount of the ATM in the previous p days, the ATM environment information of the day, and the external environment information of the day, wherein p is an integer greater than or equal to 1; obtaining a predicted withdrawal amount of the day according to the daily withdrawal amount in the previous p days, the ATM environment information of the day, and the external environment information of the day using a pre-built withdrawal amount prediction model; calculating the total withdrawal amount of the next m days according to the predicted withdrawal amount of the day, wherein the day is the first day of the future, and m is an integer greater than or equal to 1; and calculating the total withdrawal amount of the next m days according to the predicted withdrawal amount of the day. The total withdrawal amount of the day and the obtained inventory amount of the ATM are used to determine the banknote replenishment strategy; wherein the withdrawal amount prediction model includes an autoregressive model and an XGBoost model, and the autoregressive model obtains the proposed predicted withdrawal amount of the day according to the withdrawal amount of each day in the previous p days by using the autoregressive model parameters; the XGBoost model obtains the compensated error withdrawal amount of the proposed predicted withdrawal amount according to the ATM environment information of the day and the external environment information of the day by using the XGBoost model, and the predicted withdrawal amount is the sum of the proposed predicted withdrawal amount and the compensated error withdrawal amount.

[0007] According to the method for determining the cash addition strategy of the ATM according to the embodiment of the present disclosure, the daily withdrawal amount of the ATM in the previous p days, the ATM environment information of the day and the external environment information of the day are obtained; according to the daily withdrawal amount in the previous p days, the ATM environment information of the day and the external environment information of the day, the withdrawal amount prediction model constructed in advance is used to obtain the predicted withdrawal amount of the day. Among them, the withdrawal amount prediction model includes an autoregression model and an XGBoost model, and the autoregression model obtains the proposed predicted withdrawal amount of the day according to the daily withdrawal amount in the previous p days using the autoregression model parameters; the XGBoost model obtains the compensation error withdrawal amount of the proposed predicted withdrawal amount according to the ATM environment information of the day and the external environment information of the day using the XGBoost model, and the predicted withdrawal amount is the sum of the proposed predicted withdrawal amount and the compensation error withdrawal amount; according to the predicted withdrawal amount of the day, the total withdrawal amount of the next m days is calculated; according to the total withdrawal amount of the next m days and the obtained inventory of the ATM, the cash addition strategy can be determined. The present disclosure uses the fusion of the XGBoost model and the autoregressive model to construct a withdrawal amount prediction model that does not require too much configuration and specific preprocessing to obtain a more accurate prediction effect, and can also be good at combining trend inference results, further improving the prediction accuracy. Since the prediction of the withdrawal amount is more accurate, the banknote addition strategy determined by the withdrawal amount is more accurate. Therefore, the banknote addition strategy determined by the present disclosure can both meet the user's withdrawal amount requirements and reduce the cash operating costs of the ATM.

[0008] In some embodiments, the step of pre-constructing a withdrawal amount prediction model includes: constructing n training samples and g verification samples, wherein each of the training samples and each of the verification samples includes p+1 time series data, each of the time series data includes ATM environment information, external environment information, a withdrawal amount label and a timestamp, and n and g are both integers greater than or equal to 1; according to the n training samples, training the autoregressive model parameters of the autoregressive model and the XGBoost model parameters of the XGBoost model, and using the trained autoregressive model parameters and the trained XGBoost model parameters as the training model parameters of the withdrawal amount prediction model; verifying the training model parameters according to the g verification samples; if the verification passes, applying the training model parameters as the model parameters of the withdrawal amount prediction model.

[0009] In some embodiments, according to the n training samples, the autoregressive model parameters of the autoregressive model and the XGBoost model parameters of the XGBoost model are trained, and the steps of using the trained autoregressive model parameters and the trained XGBoost model parameters as the training model parameters of the withdrawal amount prediction model include: according to the withdrawal amount labels and timestamps of the n training samples, using the autoregressive model to determine the influence coefficient of the withdrawal amount of each day in the previous p days on the withdrawal amount predicted on the day; using the influence coefficient of the withdrawal amount of each day in the previous p days on the withdrawal amount predicted on the day as the autoregressive model. The invention relates to a method for predicting the withdrawal amount of a cash flow prediction model by using the trained autoregressive model parameters; according to the daily predicted withdrawal amount values ​​corresponding to the timestamps in the n training samples obtained by using the trained autoregressive model parameters, the ATM environment information, external environment information, withdrawal amount labels and timestamps of the n training samples, the XGBoost model is used to determine the error influence degree of the daily ATM environment information and external environment information on the predicted withdrawal amount values; the error influence degree is used as the XGBoost model parameters; the trained autoregressive model parameters and the trained XGBoost model parameters are used as the training model parameters of the withdrawal amount prediction model.

[0010] In some embodiments, the step of verifying the training model parameters based on g verification samples includes: inputting each of the g verification samples into the withdrawal amount prediction model, and obtaining g prediction values ​​using the training model parameters; subtracting the withdrawal amount of each verification sample on the p+1th day from the prediction value of the verification sample to obtain a prediction error value for each verification sample; normalizing the g prediction error values ​​to obtain a comprehensive value; when the comprehensive value is less than or equal to a set threshold, the verification is passed.

[0011] In some embodiments, the step of pre-constructing a withdrawal amount prediction model also includes: according to the withdrawal amount of each day in the previous p days, the ATM environment information of the day and the external environment information of the day, using the pre-constructed withdrawal amount prediction model to obtain the predicted withdrawal amount of the day, constructing a new training sample with the predicted withdrawal amount of the day, the ATM environment information of the day and the external environment information of the day, and updating n training samples with the new training sample; according to the updated n training samples, updating the autoregressive model parameters of the autoregressive model and the XGBoost model parameters of the XGBoost model, and using the updated autoregressive model parameters and the updated XGBoost model parameters as new training model parameters of the withdrawal amount prediction model; verifying the new training model parameters according to the pre-constructed r verification samples, wherein r is an integer greater than or equal to 1; if the verification passes, applying the new training model parameters as the model parameters of the withdrawal amount prediction model.

[0012] In some embodiments, the step of determining the note adding strategy based on the total withdrawal amount in the next m days and the obtained inventory amount of the ATM includes: when the total withdrawal amount in the next m days is less than or equal to the inventory amount, determining the note adding strategy as: no note adding is required; when the total withdrawal amount in the next m days is greater than the inventory amount, determining the note adding strategy as: note adding is required, and the note adding amount is the total withdrawal amount in the next m days minus the inventory amount.

[0013] In some embodiments, the ATM has an inventory upper limit value. When the total withdrawal amount in the next m days minus the inventory amount is less than or equal to the inventory upper limit value, the banknote adding strategy is determined as: additional banknotes are needed, and the additional amount is the total withdrawal amount in the next m days minus the inventory amount; when the total withdrawal amount in the next m days minus the inventory amount is greater than the inventory upper limit value, the banknote adding strategy is determined as: additional banknotes are needed, and the additional amount is the inventory upper limit value minus the inventory amount.

[0014] In some embodiments, the method for determining the ATM cash replenishment strategy also includes: when the inventory cannot meet the withdrawal demand, a pop-up window guide information selection button is displayed, wherein the guide information selection button is a button for selecting whether to accept the guidance; in response to a click operation on the button for selecting to accept the guidance, the addresses of other ATMs and / or bank branches closest to the ATM are displayed.

[0015] In some embodiments, the method for determining the ATM cash replenishment strategy also includes: in response to a click operation on a button for selecting to accept guidance, displaying a reservation number button, and in response to a click operation on the reservation number button, sending the number information to the reserved bank branch.

[0016] Another aspect of the present disclosure provides a device for determining a cash replenishment strategy for an ATM, including: an acquisition module, the acquisition module is used to execute acquisition of the daily withdrawal amount of the ATM in the previous p days, the ATM environment information of the day and the external environment information of the day, wherein p is an integer greater than or equal to 1; a prediction module, the prediction module is used to execute, based on the daily withdrawal amount in the previous p days, the ATM environment information of the day and the external environment information of the day, using a pre-built withdrawal amount prediction model to obtain the predicted withdrawal amount of the day; a calculation module, the calculation module is used to execute, based on the predicted withdrawal amount of the day, calculation of the total withdrawal amount for the next m days, wherein the day is the first day of the future and m is a large number. An integer equal to 1; a determination module, wherein the determination module is used to determine a banknote adding strategy based on the total withdrawal amount of the future m days and the acquired inventory amount of the ATM; wherein the withdrawal amount prediction model includes an autoregressive model and an XGBoost model, wherein the autoregressive model obtains a proposed predicted withdrawal amount for the day based on the withdrawal amount of each day in the previous p days using autoregressive model parameters; wherein the XGBoost model obtains a compensated error withdrawal amount for the proposed predicted withdrawal amount based on the ATM environment information of the day and the external environment information of the day using the XGBoost model, wherein the predicted withdrawal amount is the sum of the proposed predicted withdrawal amount and the compensated error withdrawal amount.

[0017] Another aspect of the present disclosure provides an electronic device, comprising one or more processors and one or more memories, wherein the memories are used to store executable instructions, and when the executable instructions are executed by the processors, the above-mentioned method is implemented.

[0018] Another aspect of the present disclosure provides a computer-readable storage medium storing computer-executable instructions, which are used to implement the above method when executed.

[0019] Another aspect of the present disclosure provides a computer program product, comprising a computer program, wherein the computer program comprises computer executable instructions, and the instructions are used to implement the method as described above when executed. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The above and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0021] Figure 1 An exemplary system architecture to which the method and apparatus according to an embodiment of the present disclosure can be applied is schematically shown;

[0022] Figure 2 A flowchart of a method for determining a cash adding strategy for an ATM according to an embodiment of the present disclosure is schematically shown;

[0023] Figure 3 A flowchart schematically shows the steps of pre-building a withdrawal amount prediction model according to an embodiment of the present disclosure;

[0024] Figure 4 A flowchart schematically shows the steps of training the autoregressive model parameters of the autoregressive model and the XGBoost model parameters of the XGBoost model according to n training samples, and using the trained autoregressive model parameters and the trained XGBoost model parameters as the training model parameters of the withdrawal amount prediction model according to an embodiment of the present disclosure;

[0025] Figure 5 A flowchart schematically shows the steps of pre-building a withdrawal amount prediction model according to an embodiment of the present disclosure;

[0026] Figure 6 A flowchart schematically shows the steps of verifying training model parameters according to g verification samples according to an embodiment of the present disclosure;

[0027] Figure 7 A flowchart schematically shows the steps of determining a banknote adding strategy according to the total withdrawal amount in the next m days and the acquired inventory amount of the ATM according to an embodiment of the present disclosure;

[0028] Figure 8 A flowchart schematically shows the steps of determining a banknote adding strategy according to the total withdrawal amount in the next m days and the acquired inventory amount of the ATM according to an embodiment of the present disclosure;

[0029] Fig. 9 A flowchart of a method for determining a cash adding strategy for an ATM according to an embodiment of the present disclosure is schematically shown;

[0030] Fig.10 A block diagram of a device for determining a banknote adding strategy for an ATM according to an embodiment of the present disclosure is schematically shown;

[0031] Fig.11 A block diagram of an electronic device according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0032] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0033] In the technical solution of the present disclosure, the user information (including but not limited to user personal information, user image information, user device information, such as location information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with relevant laws, regulations and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0034] In the scenario of using personal information for automated decision-making, the methods, devices, and systems provided by the embodiments of the present disclosure provide users with corresponding operation portals for users to choose to agree or reject the automated decision-making results; if the user chooses to reject, the expert decision-making process will be entered. The expression "automated decision-making" here refers to the activity of automatically analyzing and evaluating a person's behavioral habits, interests and hobbies, or economic, health, credit status, etc. through computer programs, and making decisions. The expression "expert decision-making" here refers to the activity of making decisions by people who specialize in a certain field, have specialized experience, knowledge and skills, and have reached a certain level of professionalism.

[0035] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0036] When using expressions such as "at least one of A, B or C, etc.", it should generally be interpreted in accordance with the meaning of the expression generally understood by those skilled in the art (for example, "a system having at least one of A, B or C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.). The terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, features defined as "first" and "second" may explicitly or implicitly include one or more of the said features.

[0037] The bank's cash deposit and withdrawal services can be handled at bank branches and self-service ATMs. Today, self-service ATMs are found in communities with a relatively high concentration of people, greatly facilitating users who need to use cash. Setting up self-service ATMs in the community means that users do not have to make a separate trip to a branch and can choose to handle cash deposit and withdrawal services at a self-service ATM.

[0038] With the rapid development of mobile payment, the frequency of cash use has begun to decrease. Some self-service ATMs have a large amount of cash inventory for a long time, which wastes resources and increases operating costs. Therefore, some banks choose to reduce the cash inventory of self-service ATMs. Cash is still an indispensable part of people's daily life, especially in communities and towns with a large elderly population. During holidays or banquets, and when the Social Security Bureau issues pensions, people use cash more frequently than on weekdays. If the cash inventory of self-service ATMs is reduced blindly, the cash inventory in the self-service ATMs will not be able to meet the users' cash withdrawal needs, leaving some customers with no money to withdraw and having to go to nearby business outlets or other self-service ATMs to handle withdrawal services, which will reduce customer service satisfaction.

[0039] If the bank does not reduce the cash inventory of the self-service deposit and withdrawal machines, for the bank cash center, the cash in the self-service deposit and withdrawal machines is part of the bank's cash inventory and part of the cash for business use. If the cash inventory of the self-service deposit and withdrawal machines is kept high for a long time, it will not only increase the bank's operating risks, but also increase the operating costs caused by the bank's frequent cash additions. At the same time, the large amount of cash inventory in the cash box of the self-service deposit and withdrawal machine will cause a waste of cash resources for business use. Therefore, how to determine a cash addition strategy for ATMs that can both meet user needs and reduce the cash operating costs of self-service deposit and withdrawal machines has become a technical problem that technicians in this field need to solve urgently.

[0040] The embodiments of the present disclosure provide a method, device, electronic device, computer-readable storage medium and computer program product for determining a cash addition strategy for an ATM. The method for determining a cash addition strategy for an ATM includes: obtaining the daily withdrawal amount of the ATM in the previous p days, the ATM environment information of the day and the external environment information of the day; and obtaining the predicted withdrawal amount of the day by using a pre-constructed withdrawal amount prediction model based on the daily withdrawal amount in the previous p days, the ATM environment information of the day and the external environment information of the day. Among them, the withdrawal amount prediction model includes an autoregressive model and an XGBoost model. The autoregressive model uses the autoregressive model parameters to obtain the proposed predicted withdrawal amount for the day based on the daily withdrawal amount in the previous p days; the XGBoost model uses the XGBoost model to obtain the compensated error withdrawal amount of the proposed predicted withdrawal amount based on the ATM environment information and the external environment information of the day. The predicted withdrawal amount is the sum of the proposed predicted withdrawal amount and the compensated error withdrawal amount; based on the predicted withdrawal amount of the day, the total withdrawal amount for the next m days is calculated; based on the total withdrawal amount for the next m days and the obtained ATM inventory, the banknote replenishment strategy can be determined.

[0041] It should be noted that the ATM cash loading strategy determination method, device, electronic device, computer-readable storage medium and computer program product disclosed in the present invention can be used in the field of artificial intelligence technology, and can also be used in any field other than the field of artificial intelligence technology, such as the field of financial technology technology. The field of the present invention is not limited here.

[0042] Figure 1 The exemplary system architecture 100 to which the method, device, electronic device, computer-readable storage medium and computer program product for determining a cash addition strategy of an ATM according to an embodiment of the present disclosure can be applied is schematically shown. It should be noted that: Figure 1 What is shown is merely an example of a system architecture to which the embodiments of the present disclosure can be applied, in order to help those skilled in the art understand the technical content of the present disclosure, but it does not mean that the embodiments of the present disclosure cannot be used in other devices, systems, environments or scenarios.

[0043] like Figure 1 As shown, the application scenario 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, and a server 105. The network 104 is used to provide a medium for a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103, and the server 105. The network 104 may include various connection types, such as wired, wireless communication links, or optical fiber cables, etc.

[0044] The user can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only for example).

[0045] The first terminal device 101, the second terminal device 102, and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0046] The server 105 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0047] It should be noted that the method for determining the banknote adding strategy of the teller provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the device for determining the banknote adding strategy of the teller provided in the embodiment of the present disclosure can generally be arranged in the server 105. The method for determining the banknote adding strategy of the teller provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Accordingly, the device for determining the banknote adding strategy of the teller provided in the embodiment of the present disclosure can also be arranged in a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105.

[0048] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0049] The following will be based on Figure 1 The scene described by Figure 2~Figure 9 The method for determining a cash replenishment strategy for an ATM in an embodiment of the present disclosure is described in detail.

[0050] Figure 2 The flowchart of the method for determining the cash adding strategy of an ATM according to an embodiment of the present disclosure is schematically shown.

[0051] like Figure 2 As shown, the method for determining a cash adding strategy for an ATM in this embodiment includes operations S210 to S240.

[0052] In operation S210, the daily withdrawal amount of the ATM in the previous p days, the ATM environment information of the day, and the external environment information of the day are obtained, where p is an integer greater than or equal to 1.

[0053] In operation S220, the withdrawal amount of each day in the previous p days, the ATM environment information of the day and the external environment information of the day are used to obtain the predicted withdrawal amount of the day using the pre-built withdrawal amount prediction model. The withdrawal amount prediction model includes an autoregressive model and an XGBoost model. The autoregressive model obtains the proposed predicted withdrawal amount of the day based on the withdrawal amount of each day in the previous p days using the autoregressive model parameters; the XGBoost model obtains the compensated error withdrawal amount of the proposed predicted withdrawal amount based on the ATM environment information of the day and the external environment information of the day using the XGBoost model. The predicted withdrawal amount is the sum of the proposed predicted withdrawal amount and the compensated error withdrawal amount.

[0054] It is understandable that the historical daily withdrawal amount can be obtained from the back-end system of the ATM, and y is used here.t represents the predicted withdrawal amount for the day, y t-1 Indicates the withdrawal amount of the previous day,

[0055] y t-2 represents the withdrawal amount in the previous two days, ..., y t-i represents the withdrawal amount of the previous i days, ..., y t-p Represents the withdrawal amount of the previous p days. The autoregressive model can obtain the predicted withdrawal amount through formula (1). Based on the predicted withdrawal amount of the day, the ATM environment information of the day and the external environment information of the day, the XGBoost model can be used to calculate the compensation error withdrawal amount, which is expressed as

[0056] The compensation error withdrawal amount represents the compensation error withdrawal amount obtained by considering the error of the ATM environment information and the external environment information when predicting the withdrawal amount to be predicted. The working process of the withdrawal amount prediction model can be expressed by formula (2).

[0057] (1)

[0058] (2)

[0059] Among them, c represents a constant term, , , …, , …, It represents the coefficient of influence of the daily withdrawal amount in the previous p days on the predicted withdrawal amount of the current day.

[0060] Among them, the ATM environment information includes but is not limited to the inventory of ATMs, the number of ATMs within a radius of L1 kilometers, the types of ATMs within a radius of L2 kilometers, the number of bank outlets within a radius of L3 kilometers, the upper limit of the inventory of ATMs, the per capita deposit amount of ATMs, and the number of ATMs issued on behalf of ATMs, etc. The external environment information includes but is not limited to holidays, working days, quarters, weather, the number of communities within a radius of L4 kilometers, traffic conditions, the number of residents within a radius of L5 kilometers, the age of residents, the flow of people, the number of offices within a radius of L6 kilometers, the school district conditions within a radius of L7 kilometers, the commercial supporting conditions within a radius of L8 kilometers, and the number of hospitals within a radius of L9 kilometers, etc. L1~L9 are all greater than 0.

[0061] The ATM environment information and external environment information are obtained through preliminary analysis of the ATM backend data, and we know that the cash inventory and withdrawal volume of the ATMs have periodic fluctuations. Through communication with the bank cash center and bank branches, we know the operating characteristics of the ATMs, and at the same time draw a portrait of the users who use the ATMs, capture the characteristics of these users, analyze the indicator variables related to the ATM withdrawal volume, and preliminarily determine the variable indicators related to the withdrawal volume. These variables are mainly determined based on preliminary analysis of previous data, communication with staff and actual observation.

[0062] As a possible implementation method, Figure 3 As shown, the steps of pre-building a withdrawal amount prediction model include operations S310 to S340.

[0063] In operation S310, n training samples and g verification samples are constructed, wherein each training sample and each verification sample includes p+1 time series data, each time series data includes ATM environment information, external environment information, withdrawal amount label and timestamp, and n and g are both integers greater than or equal to 1.

[0064] Here, for the convenience of description, the ATM environment information and external environment information are represented as x variables, where x is an integer greater than or equal to 1. The time series data can be understood as continuous data with a timestamp. Taking the present disclosure as an example, p+1 time series data can be {October 1, 24, variable 1, variable 2, …, variable j, …, variable x, withdrawal amount}, {October 2, 24, variable 1, variable 2, …, variable j, …, variable x, withdrawal amount}, {October 3, 24, variable 1, variable 2, …, variable j, …, variable x, withdrawal amount}, {October 4, 24, variable 1, variable 2, …, variable j, …, variable x, withdrawal amount}, {October 5, 24, variable 1, variable 2, …, variable j, …, variable x, withdrawal amount} and so on. The above October 1, 2024 to October 5, 2024 is only for illustration, and the p+1 time series data can be the p+1 data in any continuous time period.

[0065] Here, p+1 time series data are used as 1 training sample, and n training samples are constructed to input into the withdrawal amount prediction model. By adjusting the size of n, the accuracy of the determination of the autoregressive model parameters and the XGBoost model parameters can be adjusted. The larger the number of samples, the higher the accuracy of the determination of the autoregressive model parameters and the XGBoost model parameters. However, considering the resource cost and time cost, a certain trade-off can be made between the number of samples and the accuracy of the model parameters to maintain a balance between various factors.

[0066] In operation S320, the autoregressive model parameters of the autoregressive model and the XGBoost model parameters of the XGBoost model are trained according to the n training samples, and the trained autoregressive model parameters and the trained XGBoost model parameters are used as the training model parameters of the withdrawal amount prediction model.

[0067] As some possible implementation methods, such as Figure 4 As shown, operation S320 trains the autoregressive model parameters of the autoregressive model and the XGBoost model parameters of the XGBoost model according to n training samples, and uses the trained autoregressive model parameters and the trained XGBoost model parameters as the training model parameters of the withdrawal amount prediction model, including operations S321 to S325.

[0068] In operation S321, based on the withdrawal amount labels and timestamps of the n training samples, the autoregressive model is used to determine the influence coefficient of the withdrawal amount of each day in the previous p days on the predicted withdrawal amount of the current day. It can be understood that in each training sample, the withdrawal amount of the p+1th day is known, and the withdrawal amount of each day in the previous p days is known. The withdrawal amount of the p+1th day is substituted into the proposed predicted withdrawal amount in formula (1), and the withdrawal amount of each day in the previous p days is substituted into yt-1~yt-p in formula (1), and the influence coefficient of the withdrawal amount of each day in the previous p days on the predicted withdrawal amount of the current day can be determined.

[0069] In operation S322, the coefficient of influence of the daily withdrawal amount in the previous p days on the predicted withdrawal amount of the current day is used as an autoregressive model parameter.

[0070] In operation S323, the error influence degree of the daily ATM environment information and external environment information on the expected withdrawal amount value is determined by using the XGBoost model according to the daily expected withdrawal amount value corresponding to the timestamp in the n training samples obtained by using the trained autoregressive model parameters, the ATM environment information, external environment information, withdrawal amount label and timestamp of the n training samples. The error influence degree can be represented by formula (3).

[0071] (3)

[0072] Among them, in formula (3) It can be obtained by subtracting the predicted withdrawal amount from the daily withdrawal amount label. Indicates when Minimum The value of represents the predicted withdrawal value of the ith training sample among n training samples, represents the withdrawal amount label of the p+1th day of the ith training sample among the n training samples, Represents a penalty term, which is used to prevent overfitting of the model. The penalty term is proportional to the number of variables x (in the above, the ATM environment information and external environment information are represented as x variables, where x is an integer greater than or equal to 1).

[0073] In operation S324, the error influence degree is used as an XGBoost model parameter.

[0074] In operation S325, the trained autoregressive model parameters and the trained XGBoost model parameters are used as training model parameters of the withdrawal amount prediction model.

[0075] Operations S321 to S325 can facilitate the steps of training the autoregressive model parameters of the autoregressive model and the XGBoost model parameters of the XGBoost model according to n training samples, and using the trained autoregressive model parameters and the trained XGBoost model parameters as the training model parameters of the withdrawal amount prediction model.

[0076] In operation S330, the training model parameters are verified based on the g verification samples.

[0077] In operation S340, if the verification is passed, the training model parameters are applied as model parameters of the withdrawal amount prediction model.

[0078] Operations S310 to S340 can facilitate the implementation of the steps of pre-constructing a withdrawal amount prediction model. The withdrawal amount prediction model pre-constructed by operations S310 to S340 does not need to update the prediction model parameters within a period of time. The period of time can be understood as one month, one year, or the service life of the withdrawal amount prediction model. There is no limitation on the period of time here. After the period of time, if the model parameters of the withdrawal amount prediction model need to be updated, the training samples and the verification samples can be reconstructed. According to the reconstructed training samples and the verification samples, operations S310 to S340 are performed to obtain the model parameters of the retrained withdrawal amount prediction model, thereby obtaining a new withdrawal amount prediction model.

[0079] As other possible implementations of the present disclosure, operations S310 to S340 may be a withdrawal amount prediction model pre-built before the first prediction of the withdrawal amount. After the withdrawal amount prediction model built by operations S310 to S340 is put into use for the first time, Figure 5 As shown, the step of pre-building a withdrawal amount prediction model may also include operations S350 to S380.

[0080] In operation S350, based on the daily withdrawal amount in the previous p days, the ATM environment information of the day and the external environment information of the day, a pre-constructed withdrawal amount prediction model is used to obtain the predicted withdrawal amount of the day, and then a new training sample is constructed with the predicted withdrawal amount of the day, the ATM environment information of the day, and the external environment information of the day, and the new training sample is used to update n training samples.

[0081] In operation S360, the autoregressive model parameters of the autoregressive model and the XGBoost model parameters of the XGBoost model are updated according to the updated n training samples, and the updated autoregressive model parameters and the updated XGBoost model parameters are used as new training model parameters of the withdrawal amount prediction model.

[0082] In operation S370, the new training model parameters are verified based on r pre-constructed verification samples, where r is an integer greater than or equal to 1.

[0083] In operation S380, if the verification passes, the newly trained model parameters are applied as model parameters of the withdrawal amount prediction model.

[0084] Through operations S350 to S380, new training samples and new verification samples can be obtained based on the new predicted data every day. The model parameters of the withdrawal amount prediction model can be continuously updated based on the new training samples and verification samples, thereby continuously updating the withdrawal amount prediction model. The continuously updated withdrawal amount prediction model can keep pace with the times, meet real-time prediction needs, and improve prediction accuracy.

[0085] According to some embodiments of the present disclosure, Figure 6 As shown, operation S330 is a step of verifying the training model parameters according to g verification samples, including operations S331 to S334.

[0086] In operation S331, each of the g verification samples is input into the withdrawal amount prediction model, and g prediction values ​​are obtained by using the training model parameters.

[0087] In operation S332, the withdrawal amount of each verification sample on the p+1th day is subtracted from the predicted value of the verification sample to obtain the prediction error value of each verification sample.

[0088] In operation S333, the g prediction error values ​​are normalized to obtain a comprehensive value.

[0089] In operation S334, when the comprehensive value is less than or equal to the set threshold, the verification is passed.

[0090] Operations S331 to S334 can facilitate the step of verifying the training model parameters based on g verification samples.

[0091] According to some embodiments of the present disclosure, operation S330 is a step of verifying the training model parameters according to g verification samples, and further includes operation S335 and operation S336.

[0092] In operation S335 , when the comprehensive value is less than or equal to the set threshold, the verification fails.

[0093] In operation S336, when the verification fails, operations S310 to S330 are repeatedly performed until the verification passes.

[0094] In operation S230, the total withdrawal amount for the next m days is calculated based on the predicted withdrawal amount for the current day, where the current day is the first day in the future and m is an integer greater than or equal to 1.

[0095] Here, the total withdrawal amount for the next m days is equal to the sum of the withdrawal amounts for each of the m days.

[0096] In operation S240, a banknote replenishment strategy is determined according to the total withdrawal amount in the next m days and the acquired inventory amount of the teller machine.

[0097] As an implementable approach, Figure 7 As shown, operation S240 is a step of determining a banknote adding strategy according to the total withdrawal amount in the next m days and the acquired inventory amount of the teller machine, including operation S241 and operation S242.

[0098] In operation S241, when the total withdrawal amount in the next m days is less than or equal to the inventory amount, the banknote adding strategy is determined as: no banknote adding is required.

[0099] In operation S242, when the total withdrawal amount in the next m days is greater than the inventory amount, the banknote replenishment strategy is determined as: banknote replenishment is required, and the amount of banknote replenishment is the total withdrawal amount in the next m days minus the inventory amount.

[0100] Operation S241 and operation S242 can facilitate the step of determining the banknote adding strategy according to the total withdrawal amount in the next m days and the acquired inventory amount of the teller machine.

[0101] Furthermore, the ATM has an upper limit on inventory. Figure 8 As shown, operation S240 is a step of determining a banknote adding strategy according to the total withdrawal amount in the next m days and the acquired inventory amount of the teller machine, including operation S243 and operation S244.

[0102] In operation S243, when the total withdrawal amount in the next m days minus the inventory amount is less than or equal to the inventory upper limit, the banknote replenishment strategy is determined as: banknote replenishment is required, and the amount of banknote replenishment is the total withdrawal amount in the next m days minus the inventory amount.

[0103] In operation S244, when the total withdrawal amount in the next m days minus the inventory amount is greater than the inventory upper limit, the note adding strategy is determined as: note adding is required, and the note adding amount is the inventory upper limit minus the inventory amount.

[0104] Operations S243 and S244 can facilitate the step of determining a banknote adding strategy based on the total withdrawal amount in the next m days and the acquired inventory amount of the ATM when the ATM has an inventory upper limit.

[0105] According to the method for determining the cash addition strategy of the ATM according to the embodiment of the present disclosure, the daily withdrawal amount of the ATM in the previous p days, the ATM environment information of the day and the external environment information of the day are obtained; according to the daily withdrawal amount in the previous p days, the ATM environment information of the day and the external environment information of the day, the withdrawal amount prediction model constructed in advance is used to obtain the predicted withdrawal amount of the day. Among them, the withdrawal amount prediction model includes an autoregression model and an XGBoost model, and the autoregression model obtains the proposed predicted withdrawal amount of the day according to the daily withdrawal amount in the previous p days using the autoregression model parameters; the XGBoost model obtains the compensation error withdrawal amount of the proposed predicted withdrawal amount according to the ATM environment information of the day and the external environment information of the day using the XGBoost model, and the predicted withdrawal amount is the sum of the proposed predicted withdrawal amount and the compensation error withdrawal amount; according to the predicted withdrawal amount of the day, the total withdrawal amount of the next m days is calculated; according to the total withdrawal amount of the next m days and the obtained inventory of the ATM, the cash addition strategy can be determined. The present disclosure uses the fusion of the XGBoost model and the autoregressive model to construct a withdrawal amount prediction model that does not require too much configuration and specific preprocessing to obtain a more accurate prediction effect, and can also be good at combining trend inference results, further improving the prediction accuracy. Since the prediction of the withdrawal amount is more accurate, the banknote addition strategy determined by the withdrawal amount is more accurate. Therefore, the banknote addition strategy determined by the present disclosure can both meet the user's withdrawal amount requirements and reduce the cash operating costs of the ATM.

[0106] According to some embodiments of the present disclosure, Fig. 9 As shown, the method for determining a cash adding strategy for an ATM further includes operation S001 and operation S002.

[0107] In operation S001, when the inventory cannot meet the withdrawal demand, a guidance information selection button is displayed in a pop-up window, wherein the guidance information selection button is a button for selecting whether to accept the guidance.

[0108] In operation S002, in response to a click operation on a button for selecting to receive guidance, addresses of other teller machines and / or bank branches closest to the teller machine are displayed.

[0109] Thus, the user can be prompted with the addresses of other ATMs and / or bank branches closest to the ATM where the user is located, saving the user the effort of searching on his own and thus improving user satisfaction.

[0110] According to some embodiments of the present disclosure, the method for determining a cash adding strategy for an ATM further includes operation S003.

[0111] In operation S003, in response to the click operation of the button for selecting to accept the guidance, a button for making a reservation is displayed, and in response to the click operation of the button for making a reservation, the number information is sent to the bank branch for making the reservation. In this way, it is possible to make a reservation for the number of the bank branch to be visited in advance, reduce the waiting time of the user in the queue after arriving at the bank branch, and further improve the user satisfaction.

[0112] In the existing implementation of banknote adding operation, the banknote adding operation route is fixed, and whether the ATM needs to add banknotes, how much banknotes need to be added, the distance between the ATM and the cash management center, etc. are not considered, which easily leads to waste of resources and high cost of adding banknotes. According to some embodiments of the present disclosure, the banknote adding strategy can also include optimization of the banknote adding route, that is, according to the withdrawal amount in the next m days, the inventory of the ATM, the upper limit of the banknote adding of the ATM, the transportation volume of the banknote truck and the distribution of the ATM, the genetic algorithm is used to optimize the banknote adding operation route and reduce the operation cost of the banknote adding operation.

[0113] The following describes in detail a method for determining a cash adding strategy for an ATM according to an embodiment of the present disclosure. It is worth noting that the following description is merely an exemplary description and is not a specific limitation of the present disclosure.

[0114] The ATM cash replenishment solution based on withdrawal volume forecast is intended to further improve customer service quality while reducing the cash operating costs of bank ATMs and improving the efficiency of capital and human resources utilization.

[0115] This solution mainly predicts the withdrawal volume of the ATM in the next few days to determine whether the ATM needs to add banknotes, how much banknotes need to be added, and determine the optimal banknote adding route, so as to meet the customer's cash withdrawal needs while reducing the operating costs of the bank's cash center. The specific steps include the following.

[0116] Indicator selection and data collection: In the early stage, we used time series and other methods to conduct a preliminary analysis of the original data of ATMs, and learned that the cash inventory and withdrawal volume of ATMs have periodic fluctuations. For example, on pension payment days and holidays, the frequency of use of ATMs is significantly higher than on weekdays, and the withdrawal volume is also higher than on weekdays. There are even cases where ATMs have no cash inventory. For example, during the hot summer period, the frequency of use of ATMs is low, and the cash inventory in the machines is high. In addition, we communicated with the bank's cash center and outlets to understand the operating characteristics of ATMs, and at the same time drew portraits of customers using ATMs, captured the characteristics of these customers, analyzed the indicator variables related to the withdrawal volume of ATMs, and preliminarily determined the indicator variables related to the withdrawal volume.

[0117] The indicator variables include internal indicator variables and external indicator variables. The internal indicator variables include but are not limited to the inventory of ATMs, the number of ATMs within a radius of L kilometers, the types of ATMs within a radius of L kilometers, the number of bank outlets within a radius of L kilometers, the upper limit of the inventory of ATMs, the per capita deposit amount of ATMs, and the number of ATMs issued on behalf of ATMs, etc. The external indicator variables include but are not limited to holidays, working days, quarters, weather, the number of communities within a radius of L kilometers, traffic conditions, the number of residents within a radius of L kilometers, the age of residents, the flow of people, the number of offices within a radius of L kilometers, the school district conditions within a radius of L kilometers, the commercial supporting conditions within a radius of L kilometers, and the number of hospitals within a radius of L kilometers, etc.

[0118] The indicator variables are internal and external indicator variables that may affect the withdrawal volume of ATMs in the next few days. These indicator variables are mainly determined based on preliminary analysis of previous data, communication with staff and actual observation. The indicator variable is the independent variable x in the construction of this model, and the withdrawal volume on the day is the dependent variable y.

[0119] Data preprocessing: Due to various uncertain factors such as manual input and machine equipment, data may be missing or inaccurate during the data collection process, which seriously affects the accuracy and effectiveness of data analysis and modeling results. Therefore, data cleaning is a crucial link in data preprocessing, and the quality of cleaned data largely determines the accuracy and effectiveness of data modeling results.

[0120] For the data in the bank database, the amount of data that can be captured is huge, and the proportion of missing values ​​is very small. The amount of information in the data with missing values ​​hardly affects the construction of the overall model. For external data, the data quality varies due to various reasons. Therefore, this solution uses mean interpolation to deal with internal data missing, k-nearest neighbor interpolation to deal with external data missing, or directly deletes indicator variables with data missing exceeding a certain proportion. For data with incorrect format content, this solution converts the data with incorrect format content and deletes abnormal and logically incorrect data.

[0121] After cleaning, the data set is complete. If the data set is used directly for modeling, not only will the modeling cost be high, but the model accuracy will be low, and the amount of information in the data cannot be fully extracted. Therefore, after data cleaning, the data set needs to be feature processed. This solution performs One-Hot encoding on categorical data, normalization on continuous data, and ordered encoding on ordered data. At the same time, correlation verification and factor analysis are used to extract the characteristics of the above indicator variables to reduce the dimension of the independent variables and improve the efficiency of model prediction. The data set is divided into a training set and a validation set in a ratio of 7:3.

[0122] Model building: In recent years, machine learning algorithms based on gradient iteration have attracted more and more attention from scholars. They are favored by scholars for their advantages of obtaining satisfactory results without too much configuration and specific preprocessing. XGBoost model algorithm is one of them. However, XGBoost model also has weaknesses, because tree-based models cannot technically infer higher eigenvalues ​​than those in the training data, that is, they cannot predict values ​​outside the observed interval. The seasonal time series model is less affected by the dynamic behavior of the data and is better at inferring trends, which makes up for the weaknesses of the XGBoost model. Therefore, this scheme will adopt a hybrid scheme of XGBoost model and seasonal multivariate regression model to predict the withdrawal volume of ATMs in the next three days, so as to determine whether additional banknotes are needed and how much additional banknotes are needed.

[0123] First, based on the preliminary analysis, we know that the withdrawal volume of ATMs has temporal seasonality. We use the time series method to extract the seasonal trend information in the withdrawal volume y and construct the AR autoregressive model as follows.

[0124]

[0125] in, is today's withdrawal amount, is the withdrawal amount yesterday, is the withdrawal amount of the previous day, c represents the constant term, and p represents the order of autoregression. Dependent variable Error after seasonal information extraction.

[0126] Regression diagnosis is performed on the above AR autoregressive model, and the Lagrange multiplier method is used to test the effectiveness of the model.

[0127] Secondly, the error term of the autoregressive model contains other information related to the independent variable x. This solution uses the XGBoost model to further extract With the information about the withdrawal amount, the XGBoost model is constructed as follows.

[0128]

[0129] Indicates when When the value is minimum, Predicted value,

[0130] The first term is the prediction residual, and the second term is the penalty term. The penalty term is to prevent the model from overfitting.

[0131] Perform regression diagnosis on the above XGBoost model, and under the premise of meeting the algorithm's assumptions, comprehensively screen out the cash business forecasting model with the best performance through evaluation indicators such as F1 score, model accuracy, ROC curve, and AUC value. , in order to ensure the accuracy and stability of the model.

[0132] Finally, the autoregressive XGBoost model of this solution is as follows.

[0133] The above model is trained using the training set to obtain the optimal autoregressive XGBoost model for this solution. At the same time, the model is tested using the validation set. The autoregressive XGBoost model is further improved through the test results, so that the model of this solution is optimal and the withdrawal amount prediction is more accurate.

[0134] Direct application of the model: Use the constructed autoregressive XGBoost model to predict the ATM withdrawal volume y ̂_(t+1), y ̂_(t+2), and y ̂_(t+3) in the next three days, that is, the total withdrawal volume of ATMs in the next three days is y ̂.

[0135] y ̂=y ̂_(t+1)+y ̂_(t+2)+y ̂_(t+3)

[0136] And compare it with the cash inventory z of the ATM on that day. The judgment conditions are as follows.

[0137]

[0138] When the cash inventory z of the ATM on the day is greater than or equal to the total withdrawal amount y ̂ in the next three days, there is no need to add money. When z is less than y ̂, it is necessary to add money. Therefore, this solution is set to add money w.

[0139] w=y ̂-z

[0140] Since the cash inventory upper limit u varies among ATM models and manufacturers, there may be multiple schemes with different amounts of cash added for the cash adding operation. The specific cash adding schemes are as follows.

[0141]

[0142] Where v represents the current cash inventory balance in the ATM. When the sum of the added banknote quantity w and the cash inventory balance v is less than or equal to the cash inventory upper limit u, the added banknote quantity can be set to w. When the sum of the added banknote quantity w and the cash inventory balance v is greater than the cash inventory upper limit u, the added banknote quantity can be set to uv, that is, the ATM inventory is filled up.

[0143] In the case where the sum of the amount of added banknotes w and the cash inventory balance v is greater than the upper limit of the cash inventory u, it is obvious that the cash inventory in the ATM cannot meet the customer's cash withdrawal needs. At this time, it is necessary to link other ATMs and bank branches to guide customers who cannot meet their cash withdrawal needs to other ATMs and bank branches. The first guidance method is to display the cash inventory of the nearest ATM in the form of a bar chart on the ATM. The use of a bar chart instead of directly using a digital method is mainly due to the issue of security risks, and the geographical location of the available ATMs is marked at the same time. In addition, the nearest outlets are linked to customers who cannot meet their withdrawal needs. The screen of the ATM will display "Do you need to withdraw or deposit money from the nearest outlet? If necessary, please click the appointment button below to link the nearby outlets to queue up for number collection", solving the cumbersome problem of customers having to take numbers after arriving at the outlets. While meeting customers' cash withdrawal needs, the efficiency of cash resource utilization is improved to help banks achieve high quality.

[0144] Indirect application of the model: In the above ATM cash refilling scheme based on the forecast of cash withdrawal volume in the next three days, the distance between the ATMs and the cash refilling operation is not taken into account, as well as the problem that some areas require fewer cash refilling machines and some areas require more cash refilling machines. Therefore, this scheme can further use genetic algorithms to optimize the cash refilling operation route. Areas with less cash refilling can dispatch nearby cash refilling vehicles that need cash refilling operations to complete the operation together, thereby improving the efficiency of cash refilling operations and reducing the cost of cash refilling operations.

[0145] The ATM cash replenishment solution based on withdrawal volume forecast is conducive to meeting customers' cash deposit and withdrawal needs at one time; improving the utilization efficiency of bank cash resources; reducing the operating costs of bank ATMs; and reducing unnecessary resource waste in cash truck dispatch.

[0146] Based on the above-mentioned ATM cash adding strategy determination method, the present disclosure also provides an ATM cash adding strategy determination device. Fig.10 The device for determining a cash adding strategy for an ATM is described in detail.

[0147] Fig.10 The structural block diagram of the device for determining a cash adding strategy for an ATM according to an embodiment of the present disclosure is schematically shown.

[0148] The device 10 for determining a cash adding strategy for an ATM includes an acquisition module 1 , a prediction module 2 , a calculation module 3 and a determination module 4 .

[0149] The acquisition module 1 is used to execute acquisition of the daily withdrawal amount of the ATM in the previous p days, the ATM environment information of the day and the external environment information of the day, wherein p is an integer greater than or equal to 1.

[0150] Prediction module 2 is used to execute the predicted withdrawal amount of the day based on the daily withdrawal amount in the previous p days, the ATM environment information of the day and the external environment information of the day, using a pre-built withdrawal amount prediction model, wherein the withdrawal amount prediction model includes an autoregression model and an XGBoost model, the autoregression model uses the autoregression model parameters to obtain the proposed predicted withdrawal amount of the day based on the daily withdrawal amount in the previous p days; the XGBoost model uses the XGBoost model to obtain the compensated error withdrawal amount of the proposed predicted withdrawal amount based on the ATM environment information of the day and the external environment information of the day, and the predicted withdrawal amount is the sum of the proposed predicted withdrawal amount and the compensated error withdrawal amount.

[0151] The calculation module 3 is used to calculate the total withdrawal amount for the next m days based on the predicted withdrawal amount for the current day, where the current day is the first day of the future and m is an integer greater than or equal to 1.

[0152] The determination module 4 is used to determine the banknote adding strategy according to the total withdrawal amount in the next m days and the acquired inventory amount of the teller machine.

[0153] According to some embodiments of the present disclosure, the ATM cash replenishment strategy determination device also includes a pre-built module, which is used to pre-build a withdrawal amount prediction model. The pre-built module includes a first determination unit, a second determination unit and a first application unit.

[0154] The first determination unit is used to determine the error term by using the XGBoost model according to the obtained n first training samples, where n is an integer greater than or equal to 1.

[0155] The second determination unit is used to determine the influence coefficient of the daily withdrawal amount in the previous p days on the predicted withdrawal amount of the day based on the obtained h second training samples and the error term using the autoregressive model, where h is an integer greater than or equal to 1.

[0156] The first application unit is used to apply the influence coefficient and error term of the daily withdrawal amount in the previous p days on the predicted withdrawal amount of the current day as prediction model parameters to the autoregressive model to obtain a withdrawal amount prediction model.

[0157] According to some embodiments of the present disclosure, the first determination unit includes a first acquisition element, a first output element and a first determination element.

[0158] The first acquisition element is used to acquire n time series data as n first training samples, wherein each time series data includes x variables, a withdrawal amount label and a timestamp, the timestamp is in days, and x is an integer greater than or equal to 1.

[0159] The first output element is used to input n first training samples into the XGBoost model, and use the output of the XGBoost model for each first training sample as a predicted value of the withdrawal amount of each first training sample.

[0160] The first determination element is used to take the withdrawal amount label of each first training sample as the true value of the withdrawal amount, and determine the error term based on the true value, predicted value and penalty term of the n first training samples, wherein the penalty term is proportional to the number of variables x.

[0161] According to some embodiments of the present disclosure, the second determining unit includes a second acquiring element and a second determining element.

[0162] The second acquisition element is used to use p+1 time series data as one second training sample to acquire h second training samples.

[0163] The second determination element is used to input the withdrawal amount and error term of p+1 time series data of each second training sample in h second training samples into the autoregressive model to determine the influence coefficient of the daily withdrawal amount in the previous p days on the predicted withdrawal amount of the day.

[0164] According to some embodiments of the present disclosure, the pre-built module further includes a first updating unit, a second updating unit and a third updating unit.

[0165] The first updating unit is used to obtain the predicted withdrawal amount of the day based on the daily withdrawal amount in the previous p days using a pre-built withdrawal amount prediction model, and then update the predicted withdrawal amount of the day with n first training samples, and update the error term based on the updated n first training samples.

[0166] The second updating unit is used to obtain the predicted withdrawal amount for the day based on the daily withdrawal amount in the previous p days using a pre-built withdrawal amount prediction model, and then update the predicted withdrawal amount for the day with h second training samples; based on the updated h second training samples and the updated error term, update the coefficient of influence of the daily withdrawal amount in the previous p days on the predicted withdrawal amount for the day.

[0167] The third updating unit is used to apply the updated influence coefficient of the daily withdrawal amount in the previous p days on the predicted withdrawal amount of the day and the updated error term as prediction model parameters to the autoregressive model to obtain a withdrawal amount prediction model.

[0168] According to some embodiments of the present disclosure, the determination module includes a first auxiliary determination unit and a second auxiliary unit.

[0169] The first auxiliary determination unit is used to determine the banknote adding strategy as: no banknote adding is required when the total withdrawal amount in the next m days is less than or equal to the inventory amount.

[0170] The second auxiliary determination unit is used to determine the banknote adding strategy as follows: when the total withdrawal amount in the next m days is greater than the inventory amount, banknote adding is required, and the amount of banknote adding is the total withdrawal amount in the next m days minus the inventory amount.

[0171] According to some embodiments of the present disclosure, the determination module includes a third auxiliary determination unit and a fourth auxiliary unit. The teller machine has an inventory upper limit value.

[0172] The third auxiliary determination unit is used to determine the banknote adding strategy as: need to add banknotes, and the amount of banknotes added is the total withdrawal amount in the next m days minus the inventory amount when the total withdrawal amount in the next m days minus the inventory amount is less than or equal to the inventory upper limit.

[0173] The fourth auxiliary unit is used to determine the banknote adding strategy as follows: banknote adding is required, and the banknote adding amount is the inventory upper limit minus the inventory amount when the total withdrawal amount in the next m days minus the inventory amount is greater than the inventory upper limit.

[0174] According to some embodiments of the present disclosure, the device for determining a cash adding strategy for an ATM further includes a pop-up module and a display module.

[0175] The pop-up window module is used to display a pop-up window guidance information selection button when the inventory cannot meet the withdrawal demand, where the guidance information selection button is a button for selecting whether to accept the guidance.

[0176] The display module is used for displaying the addresses of other ATMs and / or bank branches closest to the ATM in response to a click operation on a button for selecting to accept guidance.

[0177] According to some embodiments of the present disclosure, the device for determining a cash replenishment strategy for an ATM further includes a reservation module.

[0178] The reservation module is used to display a reservation number button in response to a click operation on a button for selecting to accept guidance, and to send the number information to the reserved bank branch in response to a click operation on the reservation number button.

[0179] According to the device 10 for determining the cash addition strategy of the ATM according to the embodiment of the present disclosure, the daily withdrawal amount of the ATM in the previous p days, the ATM environment information of the day and the external environment information of the day are obtained; according to the daily withdrawal amount in the previous p days, the ATM environment information of the day and the external environment information of the day, the withdrawal amount prediction model constructed in advance is used to obtain the predicted withdrawal amount of the day. Among them, the withdrawal amount prediction model includes an autoregression model and an XGBoost model, and the autoregression model obtains the proposed predicted withdrawal amount of the day according to the daily withdrawal amount in the previous p days by using the autoregression model parameters; the XGBoost model obtains the compensation error withdrawal amount of the proposed predicted withdrawal amount according to the ATM environment information of the day and the external environment information of the day by using the XGBoost model, and the predicted withdrawal amount is the sum of the proposed predicted withdrawal amount and the compensation error withdrawal amount; according to the predicted withdrawal amount of the day, the total withdrawal amount of the next m days is calculated; according to the total withdrawal amount of the next m days and the obtained inventory of the ATM, the cash addition strategy can be determined. The present disclosure uses the fusion of the XGBoost model and the autoregressive model to construct a withdrawal amount prediction model that does not require too much configuration and specific preprocessing to obtain a more accurate prediction effect, and can also be good at combining trend inference results, further improving the prediction accuracy. Since the prediction of the withdrawal amount is more accurate, the banknote addition strategy determined by the withdrawal amount is more accurate. Therefore, the banknote addition strategy determined by the present disclosure can both meet the user's withdrawal amount requirements and reduce the cash operating costs of the ATM.

[0180] In addition, according to an embodiment of the present disclosure, any multiple modules among the acquisition module 1, the prediction module 2, the calculation module 3 and the determination module 4 can be combined into one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module.

[0181] According to an embodiment of the present disclosure, at least one of the acquisition module 1, the prediction module 2, the calculation module 3 and the determination module 4 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application-specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or can be implemented in any one of the three implementation methods of software, hardware and firmware, or in an appropriate combination of any of them.

[0182] Alternatively, at least one of the acquisition module 1, the prediction module 2, the calculation module 3 and the determination module 4 may be at least partially implemented as a computer program module, and when the computer program module is executed, the corresponding function may be executed.

[0183] Fig.11 A block diagram of an electronic device suitable for implementing the above method according to an embodiment of the present disclosure is schematically shown.

[0184] like Fig.11 As shown, the electronic device 900 according to an embodiment of the present disclosure includes a processor 901, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 902 or a program loaded from a storage part 908 to a random access memory (RAM) 903. The processor 901 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or a related chipset and / or a dedicated microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 901 may also include an onboard memory for caching purposes. The processor 901 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0185] In RAM 903, various programs and data required for the operation of electronic device 900 are stored. Processor 901, ROM 902 and RAM 903 are connected to each other via bus 904. Processor 901 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 902 and / or RAM 903. It should be noted that the program can also be stored in one or more memories other than ROM 902 and RAM 903. Processor 901 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in the one or more memories.

[0186] According to an embodiment of the present disclosure, the electronic device 900 may further include an input / output (I / O) interface 905, which is also connected to the bus 904. The electronic device 900 may further include one or more of the following components connected to the input / output (I / O) interface 905: an input portion 906 including a keyboard, a mouse, etc.; an output portion 907 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 908 including a hard disk, etc.; and a communication portion 909 including a network interface card such as a LAN card, a modem, etc. The communication portion 909 performs communication processing via a network such as the Internet. A drive 910 is also connected to the input / output (I / O) interface 905 as needed. A removable medium 911, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 910 as needed, so that a computer program read therefrom is installed into the storage portion 908 as needed.

[0187] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present disclosure is implemented.

[0188] According to an embodiment of the present disclosure, a computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus or a device. For example, according to an embodiment of the present disclosure, a computer-readable storage medium may include the ROM 902 and / or RAM 903 described above and / or one or more memories other than ROM 902 and RAM 903.

[0189] The embodiment of the present disclosure also includes a computer program product, which includes a computer program, and the computer program includes a program code for executing the method shown in the flowchart. When the computer program product is run in a computer system, the program code is used to enable the computer system to implement the method provided by the embodiment of the present disclosure. When the computer program is executed by the processor 901, the above-mentioned functions defined in the system / device of the embodiment of the present disclosure are executed. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0190] In one embodiment, the computer program may be based on a tangible storage medium such as an optical storage device, a magnetic storage device, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of a signal on a network medium, and downloaded and installed through the communication part 909, and / or installed from a removable medium 911. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0191] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 909, and / or installed from the removable medium 911. When the computer program is executed by the processor 901, the above functions defined in the system of the embodiment of the present disclosure are performed. According to the embodiment of the present disclosure, the system, device, means, module, unit, etc. described above can be implemented by a computer program module.

[0192] According to an embodiment of the present disclosure, the program code for executing the computer program provided by the embodiment of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect through the Internet).

[0193] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0194] It will be appreciated by those skilled in the art that the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways, even if such combinations or combinations are not explicitly described in the present disclosure. In particular, without departing from the spirit and teachings of the present disclosure, the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways. All of these combinations and / or combinations fall within the scope of the present disclosure.

[0195] The embodiments of the present disclosure are described above. However, these embodiments are only for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described above, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. Without departing from the scope of the present disclosure, those skilled in the art may make a variety of substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. A method for determining a cash adding strategy for an ATM, characterized in that: include: Obtain the daily withdrawal amount of the ATM in the previous p days, the ATM environment information of the day, and the external environment information of the day, where p is an integer greater than or equal to 1; According to the daily withdrawal amount in the previous p days, the ATM environment information of the day and the external environment information of the day, a withdrawal amount prediction model constructed in advance is used to obtain the predicted withdrawal amount of the day; Calculate the total withdrawal amount for the next m days based on the predicted withdrawal amount for the current day, where the current day is the first day of the future and m is an integer greater than or equal to 1; Determine a banknote replenishment strategy according to the total withdrawal amount in the next m days and the acquired inventory amount of the ATM; Among them, the withdrawal amount prediction model includes an autoregressive model and an XGBoost model. The autoregressive model uses the autoregressive model parameters to obtain the proposed predicted withdrawal amount for the day based on the daily withdrawal amount in the previous p days; the XGBoost model uses the XGBoost model to obtain the compensated error withdrawal amount of the proposed predicted withdrawal amount based on the ATM environment information of the day and the external environment information of the day. The predicted withdrawal amount is the sum of the proposed predicted withdrawal amount and the compensated error withdrawal amount.

2. The method for determining a cash adding strategy for an ATM according to claim 1, characterized in that: The steps to pre-build a withdrawal volume prediction model include: Construct n training samples and g verification samples, wherein each of the training samples and each of the verification samples includes p+1 time series data, each of the time series data includes ATM environment information, external environment information, withdrawal amount label and timestamp, and n and g are both integers greater than or equal to 1; According to the n training samples, the autoregressive model parameters of the autoregressive model and the XGBoost model parameters of the XGBoost model are trained, and the trained autoregressive model parameters and the trained XGBoost model parameters are used as training model parameters of the withdrawal amount prediction model; Verify the training model parameters according to g verification samples; If the verification is passed, the training model parameters are applied as model parameters of the withdrawal amount prediction model.

3. The method for determining a cash adding strategy for an ATM according to claim 2, characterized in that: The step of training the autoregressive model parameters of the autoregressive model and the XGBoost model parameters of the XGBoost model according to the n training samples, and using the trained autoregressive model parameters and the trained XGBoost model parameters as the training model parameters of the withdrawal amount prediction model comprises: According to the withdrawal amount labels and timestamps of the n training samples, using an autoregressive model, determine the coefficient of influence of the daily withdrawal amount in the previous p days on the predicted withdrawal amount of the current day; The influence coefficient of the daily withdrawal amount in the previous p days on the predicted withdrawal amount of the current day is used as the autoregressive model parameter; According to the daily predicted withdrawal amount values ​​corresponding to the timestamps in the n training samples obtained by using the trained autoregressive model parameters, the ATM environment information, external environment information, withdrawal amount labels and timestamps of the n training samples, using the XGBoost model, determine the degree of error influence of the daily ATM environment information and external environment information on the predicted withdrawal amount values; Using the error impact degree as the XGBoost model parameter; The trained autoregressive model parameters and the trained XGBoost model parameters are used as the training model parameters of the withdrawal amount prediction model.

4. The method for determining a cash adding strategy for an ATM according to claim 2, characterized in that: The step of verifying the training model parameters according to g verification samples comprises: Input each of the g verification samples into the withdrawal amount prediction model, and obtain g prediction values ​​using the training model parameters; Subtract the withdrawal amount of each verification sample on the p+1th day from the predicted value of the verification sample to obtain the prediction error value of each verification sample; Normalizing the g prediction error values ​​to obtain a comprehensive value; When the comprehensive value is less than or equal to the set threshold, the verification passes.

5. The method for determining a cash adding strategy for an ATM according to claim 2, characterized in that: The steps of pre-building a withdrawal volume prediction model also include: According to the daily withdrawal amount in the previous p days, the ATM environment information of the day and the external environment information of the day, after obtaining the predicted withdrawal amount of the day by using the pre-built withdrawal amount prediction model, the predicted withdrawal amount of the day, the ATM environment information of the day and the external environment information of the day are constructed into a new training sample, and the new training sample is used to update n training samples; According to the updated n training samples, the autoregressive model parameters of the autoregressive model and the XGBoost model parameters of the XGBoost model are updated, and the updated autoregressive model parameters and the updated XGBoost model parameters are used as new training model parameters of the withdrawal amount prediction model; Verify the new training model parameters according to r pre-constructed verification samples, where r is an integer greater than or equal to 1; If the verification is passed, the newly trained model parameters are applied as model parameters of the withdrawal amount prediction model.

6. The method for determining a cash adding strategy for an ATM according to any one of claims 1 to 5, characterized in that: The step of determining a banknote adding strategy according to the total withdrawal amount in the next m days and the acquired inventory amount of the ATM includes: When the total withdrawal amount in the next m days is less than or equal to the inventory amount, the banknote adding strategy is determined as: no banknote adding is required; When the total withdrawal amount in the future m days is greater than the inventory amount, the banknote adding strategy is determined as: additional banknotes are needed, and the additional amount is the total withdrawal amount in the future m days minus the inventory amount.

7. The method for determining a cash adding strategy for an ATM according to claim 6, characterized in that: The ATM has an upper limit of inventory. When the total withdrawal amount in the next m days minus the inventory amount is less than or equal to the inventory upper limit, the banknote adding strategy is determined as: banknote adding is required, and the banknote adding amount is the total withdrawal amount in the next m days minus the inventory amount; When the total withdrawal amount in the next m days minus the inventory amount is greater than the inventory upper limit value, the note adding strategy is determined as: note adding is required, and the note adding amount is the inventory upper limit value minus the inventory amount.

8. The method for determining a cash adding strategy for an ATM according to claim 7, characterized in that: Also includes: When the inventory cannot meet the withdrawal demand, a pop-up window will be displayed with a guidance information selection button, wherein the guidance information selection button is a button for selecting whether to accept the guidance; In response to a click operation on a button for selecting to accept guidance, addresses of other teller machines and / or bank branches closest to the teller machine are displayed.

9. The method for determining a cash adding strategy for an ATM according to claim 8, characterized in that: Also includes: In response to a click operation on a button for selecting to accept guidance, a reservation number button is displayed, and in response to a click operation on the reservation number button, the number information is sent to the reserved bank branch.

10. A device for determining a cash adding strategy for an ATM, characterized in that: include: An acquisition module, the acquisition module is used to execute acquisition of the daily withdrawal amount of the ATM in the previous p days, the ATM environment information of the day and the external environment information of the day, wherein p is an integer greater than or equal to 1; A prediction module, the prediction module is used to obtain the predicted withdrawal amount of the day according to the withdrawal amount of each day in the previous p days, the ATM environment information of the day and the external environment information of the day, using a pre-built withdrawal amount prediction model; A calculation module, the calculation module is used to calculate the total withdrawal amount of the next m days according to the predicted withdrawal amount of the current day, wherein the current day is the first day of the future, and m is an integer greater than or equal to 1; A determination module, the determination module is used to determine a banknote adding strategy according to the total withdrawal amount in the next m days and the acquired inventory amount of the ATM; Among them, the withdrawal amount prediction model includes an autoregressive model and an XGBoost model. The autoregressive model uses the autoregressive model parameters to obtain the proposed predicted withdrawal amount for the day based on the daily withdrawal amount in the previous p days; the XGBoost model uses the XGBoost model to obtain the compensated error withdrawal amount of the proposed predicted withdrawal amount based on the ATM environment information of the day and the external environment information of the day. The predicted withdrawal amount is the sum of the proposed predicted withdrawal amount and the compensated error withdrawal amount.

11. An electronic device, comprising: one or more processors; a storage device for storing one or more computer programs, It is characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 9.

12. 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 9 are implemented.

13. A computer program product, comprising a computer program, 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 9 are implemented.

Images