CN116308472A

CN116308472A - Transaction amount prediction method, device, equipment and storage medium of bank equipment

Info

Publication number: CN116308472A
Application number: CN202211478088.9A
Authority: CN
Inventors: 王攀
Original assignee: Ping An Bank Co Ltd
Current assignee: Ping An Bank Co Ltd
Priority date: 2022-11-23
Filing date: 2022-11-23
Publication date: 2023-06-23

Abstract

The invention discloses a transaction amount prediction method, a device, equipment and a storage medium of bank equipment, wherein the method acquires equipment type, historical transaction data and abnormal transaction amount frequency of equipment to be predicted; determining the window length of historical transaction data of the equipment to be predicted according to the abnormal frequency of the transaction amount of the equipment to be predicted and the equipment type of the equipment to be predicted; grouping historical transaction data according to window length to obtain a plurality of groups of transaction data, and obtaining a target transaction reference quantity of equipment to be predicted according to the historical transaction quantity in each group of transaction data; determining and obtaining a target transaction amount of equipment to be predicted according to the target transaction reference amount and the predicted transaction amount; the method and the system for predicting the transaction amount based on the bank equipment have the advantages that the predicted transaction amount is corrected through the target transaction reference amount, the target transaction amount is obtained, the window length is determined based on the equipment type of the equipment to be predicted and the abnormal frequency of the transaction amount, and meanwhile historical experience and type factors are considered, so that the accuracy of the transaction amount prediction of the bank equipment is improved.

Description

Transaction amount prediction method, device, equipment and storage medium of bank equipment

Technical Field

The invention relates to the technical field of data prediction, in particular to a transaction amount prediction method, device and equipment of banking equipment and a storage medium.

Background

With the continuous development of economic level, more and more customers access property at banks, ATM machines are used as daily self-service funds transaction devices for customers, and have become an indispensable resource for daily operation of banks. Banking personnel typically need to add a schedule to the ATM machine number based on the daily transaction amount requirements. However, the prediction accuracy is not high in the prior art mainly by manual estimation.

Disclosure of Invention

The embodiment of the invention provides a transaction amount prediction method, device and equipment of banking equipment and a storage medium, so as to improve the accuracy of the transaction amount prediction of the banking equipment.

In one aspect, an embodiment of the present invention provides a transaction amount prediction method for a banking device, including:

acquiring equipment type, historical transaction data and transaction amount anomaly frequency of equipment to be predicted;

predicting transaction amount according to the historical transaction data to obtain predicted transaction amount of the equipment to be predicted;

determining the window length of historical transaction data of the equipment to be predicted according to the abnormal frequency of the transaction amount of the equipment to be predicted and the equipment type of the equipment to be predicted;

Grouping the historical transaction data according to the window length to obtain a plurality of groups of transaction data;

obtaining target transaction reference quantity of the equipment to be predicted according to the historical transaction quantity in each group of transaction data;

and determining and obtaining the target transaction amount of the equipment to be predicted according to the target transaction reference amount and the predicted transaction amount.

In another aspect, an embodiment of the present invention provides a transaction amount predicting apparatus for a banking device, including:

the acquisition module is used for acquiring the equipment type, the historical transaction data and the abnormal transaction amount frequency of the equipment to be predicted;

the first prediction module is used for predicting the transaction amount according to the historical transaction data to obtain the predicted transaction amount of the equipment to be predicted;

the window determining module is used for determining and obtaining the window length of the historical transaction data of the equipment to be predicted according to the abnormal frequency of the transaction amount of the equipment to be predicted and the equipment type of the equipment to be predicted;

the grouping module is used for grouping the historical transaction data according to the window length to obtain a plurality of groups of transaction data;

the reference quantity determining module is used for obtaining the target transaction reference quantity of the equipment to be predicted according to the historical transaction quantity in each group of transaction data;

And the second prediction module is used for determining and obtaining the target transaction amount of the equipment to be predicted according to the target transaction reference amount and the predicted transaction amount.

On the other hand, the embodiment of the invention provides transaction amount prediction equipment of banking equipment, which comprises a memory and a processor; the memory stores an application program, and the processor is configured to run the application program in the memory to perform the operations in the transaction amount prediction method of the banking device.

In another aspect, an embodiment of the present invention provides a storage medium storing a plurality of instructions adapted to be loaded by a processor to perform the steps in the transaction amount prediction method of banking devices described above.

The embodiment of the invention acquires the equipment type, historical transaction data and abnormal transaction amount frequency of equipment to be predicted; predicting transaction amount according to the historical transaction data to obtain predicted transaction amount of equipment to be predicted; determining the length of a data window for obtaining historical transaction of the equipment to be predicted according to the abnormal frequency of the transaction amount of the equipment to be predicted and the equipment type of the equipment to be predicted; grouping historical transaction data according to window length to obtain a plurality of groups of transaction data, and obtaining a target transaction reference quantity of equipment to be predicted according to the historical transaction quantity in each group of transaction data; determining and obtaining a target transaction amount of equipment to be predicted according to the target transaction reference amount and the predicted transaction amount; according to the embodiment of the invention, the predicted transaction amount obtained based on the historical transaction data is corrected through the target transaction reference amount of the equipment to be predicted, the target transaction amount is obtained, the window length is determined based on the equipment type and the abnormal transaction amount frequency of the equipment to be predicted, and meanwhile, the historical experience and type factors are considered, so that the accuracy of the window length is improved, the accuracy of determining the target transaction reference amount of the equipment to be predicted according to the window length is further improved, and the accuracy of predicting the transaction amount of the bank equipment is further improved.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the description of the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

Fig. 1 is an application scenario schematic diagram of a transaction amount prediction method of banking equipment provided by an embodiment of the present invention;

fig. 2 is a flow chart of a transaction amount prediction method of a banking device according to an embodiment of the present invention;

fig. 3 is a schematic structural diagram of a transaction amount predicting device of a banking apparatus according to an embodiment of the present invention;

fig. 4 is a schematic structural diagram of a transaction amount predicting device of a banking device according to an embodiment of the present invention.

Detailed Description

The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to fall within the scope of the invention.

As described in the background art, the conventional transaction amount prediction method of a banking device mainly sets a corresponding window length based on a historical transaction amount of the device, predicts a daily transaction amount of the device based on the window length and the historical transaction amount, and determines an amount of data to be added to the device based on the predicted daily transaction amount. However, the window length of all the bank devices is set to the same window length without considering the device type of the device when the window length is set, so that deviation exists in the predicted transaction amount, and the accuracy of the transaction amount prediction of the bank devices is reduced.

Based on the above, in order to improve the accuracy of the transaction amount prediction of the banking equipment, the embodiment of the invention provides a transaction amount prediction method of the banking equipment, which is used for determining the window length based on the equipment type and the abnormal frequency of the transaction amount, and simultaneously considering the historical experience and the type factor, so that the accuracy of determining the target transaction reference amount of the equipment to be predicted according to the length of the sliding window is improved, and the accuracy of correcting the predicted transaction amount obtained based on the historical transaction data through the target transaction reference amount of the equipment to be predicted is improved, and the accuracy of the transaction amount prediction of the banking equipment is improved.

As shown in fig. 1, fig. 1 is an application scenario schematic diagram of a transaction amount prediction method of a banking device according to an embodiment of the present invention, where the application scenario includes a transaction server, at least one banking device, and a database. The banking equipment can be terminal equipment in various forms, such as a POS machine, an ATM (automatic teller machine) or a banking terminal equipment. The transaction server may comprise a transaction server of one or more banks.

The transaction server may be an independent server, or may be a server network or a server cluster formed by servers, for example, a computer, a network host, a single network server, a plurality of network server sets, or a cloud server formed by a plurality of servers. Wherein the Cloud server is composed of a large number of computers or web servers based on Cloud Computing (Cloud Computing).

The transaction server may be a server that provides background support for banking applications. A large number of computer or web server components (Cloud Computing) may be composed of one or more functional units.

For example, as shown in FIG. 1, the transaction server includes an interface unit, a data unit, and a processing unit.

The interface unit is used for carrying out data interaction with at least one bank device, sending a data request to the bank device, and acquiring transaction data and device identification returned by the at least one bank device based on the data request.

The data unit is used for writing transaction data of at least one bank device into the database after being associated with the device identification, and inquiring the database according to the device identification of the bank device when receiving the transaction amount prediction instruction to obtain the device type, the historical transaction data and the abnormal frequency of the transaction amount.

The processing unit is used for predicting the transaction amount according to the historical transaction data to obtain the predicted transaction amount of the equipment to be predicted, determining the window length of the historical transaction data of the equipment to be predicted according to the abnormal frequency of the transaction amount of the equipment to be predicted and the equipment type of the equipment to be predicted, and grouping the historical transaction data according to the window length to obtain a plurality of groups of transaction data; obtaining target transaction reference quantity of equipment to be predicted according to the historical transaction quantity in each group of transaction data; and determining and obtaining the target transaction amount of the equipment to be predicted according to the target transaction reference amount and the predicted transaction amount.

The databases may be Redis databases (full name: remote Dictionary Server, chinese: remote data service), SQL (full name: structured Query Language, chinese: structured query language) databases, or other types of databases. The database is used for storing various data, such as historical transaction data of each bank terminal, equipment type, abnormal transaction amount frequency and the like.

The at least one banking device may upload historical transaction data to the transaction server upon receiving a data request sent by the transaction server, wherein the data request is for requesting the banking device for the transaction data and device identification of the current day.

In some embodiments of the invention, the at least one banking device is connected to the transaction server via a communications network, which in some embodiments may be a wired network or a wireless network.

In some embodiments of the present invention, the wired or wireless network described above uses standard communication techniques and/or protocols. The network may be the Internet, but may be any network including, but not limited to, a wide area network, a metropolitan area network, a regional network, a third generation partnership project (3rd Generation Partnership Project,3GPP), a long term evolution (Long Term Evolution LTE), a worldwide interoperability for microwave access (Worldwide Interoperability for Microwave Access WiMAX), or a computer network communication based on the TCP/IP protocol family (TCP/IPProtocol Suite TCP/IP), the user datagram protocol (User Datagram Protocol UDP), or the like.

As shown in fig. 2, fig. 2 is a flow chart of a transaction amount prediction method of a banking device according to an embodiment of the present invention, where the transaction amount prediction method of the banking device shown in fig. 1 may be applied to a transaction server to implement prediction of a transaction amount of the banking device, and specifically the transaction amount prediction method of the banking device includes steps 201 to 206:

And 201, acquiring the equipment type, historical transaction data and abnormal transaction amount frequency of equipment to be predicted.

Historical transaction data includes, but is not limited to, flowing data of the device to be predicted over a period of time, historical inflow maximum, historical outflow maximum, rolling stock, and the like. The flow data comprises historical inflow data and historical outflow data of equipment to be predicted; the historical inflow maximum value characterization is that when data transaction is executed on equipment to be predicted, the data inflow is marked as a positive value, the data outflow is marked as a negative value, the data amount of each transaction in the past period is added one by one, and the maximum positive value is the historical inflow maximum value; the historical outflow maximum value is obtained by adding the data amount of each transaction in a period of time in the past when the data transaction is executed on the equipment to be predicted, wherein the data inflow mark is a positive value, the data outflow mark is a negative value, and the maximum negative value is the historical outflow maximum value; the rolling stock is the difference between the amount of data flowing out of the data and the amount of data flowing in of the data over a period of time while the data transaction is being performed on the equipment to be predicted.

Wherein the past period of time may be the past 1 day, the past 2 days, the past 1 week, the past 1 month, or the past 1 year.

Transaction volume anomalies include, but are not limited to, transaction services that fail to provide data in due to an amount of data in the device to be predicted being greater than or equal to a preset upper data volume limit, and transaction services that fail to provide data out due to an amount of data in the device to be predicted being less than or equal to a preset lower data volume limit. It will be appreciated that the abnormal frequency of transaction amount includes a frequency of transaction service incapable of providing data inflow and a frequency of transaction service incapable of providing data outflow.

Further, in some embodiments, in the process of providing data transaction, when the bank device detects that the bank device cannot provide the transaction service into which data flows or cannot provide the transaction service from which data flows, the transaction server reports the abnormal information to the transaction server, the transaction server stores the received abnormal information and the device identifier of the bank device in a database in association, and when the transaction server receives the transaction amount prediction instruction, the transaction server queries the database according to the device identifier of the device to be predicted, and obtains the abnormal frequency of the transaction amount of the device to be predicted in a past period of time. Wherein the anomaly information includes anomaly types including, but not limited to, transaction volume anomalies.

Further, in other embodiments, historical transaction data of each bank device in the period may be periodically acquired, according to the running water data in the historical transaction data of the bank device in the period, the data amount of the data in the period and the data amount of the data out-flowing are counted, if the data amount of the data in the period is greater than the data amount of the data out-flowing or the data amount of the data in the period is less than the data amount of the data out-flowing, the abnormal transaction amount of the bank device is determined, the abnormal transaction amount number of the bank device is recorded, and the abnormal transaction amount frequency of each bank device is associated with the device identifier of the bank device. And when the transaction server executes transaction amount prediction, acquiring abnormal frequency of the transaction amount associated with the equipment identifier according to the equipment identifier of the equipment to be predicted. Wherein the period may be 1 day, 3 days, 1 week, etc. It can be appreciated that when the data amount of the incoming data is greater than the data amount of the outgoing data in the period, the data amount reaches the upper limit in the period, which is often caused by the bank device not providing the transaction service of the incoming data; when the data amount of the incoming data is smaller than the data amount of the outgoing data in the period, the banking device is often caused by the fact that the transaction service of the outgoing data cannot be provided when the data amount is smaller than or equal to the lower limit in the period.

In some embodiments of the invention, the device types include, but are not limited to, banking internal devices, banking outsourcing devices, remote devices, and near devices, among others. The bank internal equipment characterization transaction server can directly access the bank equipment, the bank outsourcing equipment characterization transaction server needs to access the bank equipment by means of the proxy node, the remote equipment characterization bank equipment and the bank website where the transaction server is located are larger than or equal to a preset distance threshold, and the close equipment characterization bank equipment and the bank website where the transaction server is located are smaller than the preset distance threshold. It can be understood that, for the internal equipment of the bank and the close-range equipment, the external equipment of the bank is inconvenient to operate due to the fact that historical transaction data are not easy to operate and the data amount of the equipment is not convenient to operate, and the remote equipment is inconvenient to operate when the data amount of the equipment is increased, larger window data are required to be set for predicting the transaction data of a plurality of subsequent days, so that the problem of abnormal transaction amount caused by untimely increase of the data amount when the data amount of the external equipment of the bank and the remote equipment reaches the lower limit is avoided.

And 202, predicting the transaction amount according to the historical transaction data to obtain the predicted transaction amount of the equipment to be predicted.

The preset transaction amount characterizes an increased amount of data required by the device to be predicted to perform the transaction over a future period of time. The future time period may be a future day, a future three days, a future week, etc., which is not particularly limited in the embodiment of the present invention.

In some embodiments of the present invention, historical transaction data may be input to a preset prediction model to predict the transaction amount, so as to obtain a predicted transaction amount of the device to be predicted.

Further, in some embodiments, the predictive model may be a Long Short term memory (Long Short TermMemory, LSTM) based predictive model. For example, the predicted transaction amount of the device to be predicted is obtained by predicting the transaction amount according to the historical transaction data based on a prediction model of BiLSTM.

Further, in other embodiments, the predictive model may be a machine learning based predictive model. For example, the predictive model may be XGBoost, lightGBM, logistic regression, random forest based predictive model.

Further, in other embodiments, the predictive model may be a convolutional neural (Convolutional Neural Networks, CNN) based predictive model. For example, a predictive model based on deconvolution neural networks (De-Convolutional Networks, DN), deep neural networks (Deep Neural Networks, DNN), deep convolutional inverse graph networks (Deep Convolutional Inverse Graphics Networks, DCIGN), region-based convolutional networks (Region-based Convolutional Networks, RCNN), region-based fast convolutional networks (Faster Region-based Convolutional Networks, faster RCNN).

In other embodiments of the present invention, statistical analysis may be performed on the historical transaction data to obtain statistical features of the historical transaction data, and transaction amount prediction may be performed according to the statistical features of the historical transaction data to obtain a predicted transaction amount of the device to be predicted. Wherein the statistical features include, but are not limited to, average, maximum, minimum, median, mode, range, etc.

Further, in some embodiments, statistical analysis may be performed on the historical transaction data to obtain statistical features of the historical transaction data, and the transaction amount is predicted according to the statistical features to obtain a predicted transaction amount of the device to be predicted. For example, the statistical feature may be set as an initial predicted transaction amount, a preset predicted number of days is acquired, and a product of the initial predicted transaction amount and the preset predicted number of days is set as the predicted transaction amount; or, according to the statistical feature, the pre-stored coefficient data is queried to obtain a target coefficient corresponding to the data range where the statistical feature is located, the product of the target coefficient and the extremely bad is set as the initial predicted transaction amount, the preset predicted days are obtained, and the product of the initial predicted transaction amount and the preset predicted days is set as the predicted transaction amount. The pre-stored coefficient data comprises a plurality of statistical characteristics, a plurality of data ranges of each statistical characteristic and coefficients corresponding to each data range.

Further, in other embodiments, statistical analysis may be performed on the historical transaction data to obtain statistical features of the historical transaction data, and the statistical features are input into a preset prediction model to perform transaction amount prediction, so as to obtain a predicted transaction amount of the device to be predicted. The prediction model may be a neural network-based prediction model, such as an LSTM-based prediction model, or a machine learning-based prediction model, such as a logistic regression-based prediction model or a random forest-based prediction model.

And 203, determining the window length of historical transaction data of the equipment to be predicted according to the abnormal frequency of the transaction amount of the equipment to be predicted and the equipment type of the equipment to be predicted.

The window length is used for quantifying a future time period to be predicted by the device to be predicted, for example, when the window length is 1 day, namely, the target transaction amount of the device to be predicted in the future day needs to be determined; when the window length is 5 days, the target transaction amount of the device to be predicted in the next five days needs to be determined.

The window length of the historical transaction data of the equipment to be predicted is determined according to the abnormal frequency of the transaction amount of the equipment to be predicted and the equipment type of the equipment to be predicted.

In some embodiments of the present invention, a preset window length may be obtained, a correction parameter is determined according to the abnormal frequency of the transaction amount of the device to be predicted and the device type of the device to be predicted, and a historical transaction data window length is obtained according to the preset window length and the correction parameter.

Further, in some embodiments, the device type of the device to be predicted may be converted to obtain a device type parameter, and the correction parameter may be determined according to a sum or an average of the device type parameter and the abnormal frequency of the transaction amount. For example, the sum or average of the device type parameter and the transaction amount abnormality frequency may be determined as the correction parameter; and the prestored parameter data can be queried according to the numerical range where the sum of the equipment type parameter and the abnormal frequency of the transaction amount or the average value is located, so as to obtain the correction parameter. The pre-stored parameter data comprises a plurality of numerical value ranges and correction parameters corresponding to the numerical value ranges.

Further, in other embodiments, the first correction parameter may be determined according to the abnormal frequency of the transaction amount of the device to be predicted, the second correction parameter may be determined according to the type of the device to be predicted, and the sum, the arithmetic average or the weighted average value of the first correction parameter and the second correction parameter may be used as the correction parameter.

In other embodiments of the present invention, the historical transaction data window length may be determined from pre-stored window data based on the frequency of transaction volume anomalies for the device to be predicted and the device type of the device to be predicted. For example, the target window length data corresponding to the equipment type can be obtained from the pre-stored window data according to the equipment type of the equipment to be predicted, and the target window length data can be queried according to the abnormal frequency of the transaction amount of the equipment to be predicted, so as to obtain the historical transaction data window length of the equipment to be predicted. The pre-stored window data comprise window length data corresponding to various equipment types; the target window length data comprises a plurality of abnormal transaction amount frequencies and open lengths corresponding to the abnormal transaction amount frequencies.

204, grouping the historical transaction data according to the window length to obtain a plurality of groups of transaction data.

And 205, obtaining target transaction reference quantity of the equipment to be predicted according to the historical transaction quantity in each group of transaction data.

The target transaction reference quantity is used for quantifying abnormal conditions of the transaction quantity of the equipment to be predicted in the past period of time so as to determine whether the equipment to be predicted frequently generates abnormal transaction quantity in the past period of time.

In some embodiments, the target transaction reference for the device to be predicted may be derived from statistical features of historical transaction amounts in each set of transaction data. For example, statistics of historical transaction amounts in each set of transaction data include, but are not limited to, the range, maximum, minimum, and average of the historical transaction amounts in each set of transaction data.

And 206, determining and obtaining the target transaction amount of the equipment to be predicted according to the target transaction reference amount and the predicted transaction amount.

The target transaction amount characterizes an amount of data required by the device to be predicted to perform the transaction over a number of days of a future window length.

In some embodiments, an average of the target transaction reference amount and the predicted transaction amount may be determined as the target transaction amount for the device to be predicted.

In other embodiments, the maximum of the target transaction reference amount and the predicted transaction amount may be set as the target transaction amount of the device to be predicted.

In other embodiments, a weighted sum of the target transaction reference amount and the predicted transaction amount may be set as the target transaction amount for the device to be predicted.

In some embodiments of the present invention, after determining the target transaction amount for the device to be predicted, the amount of data that the device to be predicted needs to increase may be determined based on the target transaction amount for the device to be predicted. For example, the target transaction amount of the device to be predicted is set as the amount of data that the device to be predicted needs to increase, or the difference between the target transaction amount of the device to be predicted and the current remaining amount of data of the device to be predicted is set as the amount of data that the device to be predicted needs to increase.

In other embodiments of the present invention, after determining the target transaction amount for the device to be predicted, a remaining data amount for the device to be predicted is predicted based on historical transaction data for the device to be predicted, and a trip plan report for the device to be predicted is determined based on the remaining data amount and the target transaction amount.

The clearing plan comprises whether to increase the data volume or recycle the data volume of the equipment to be predicted, and the data volume which needs to be increased when the data volume is increased and the data volume which needs to be recycled when the data volume is recycled. The data volume recovery refers to recovering the data volume in the equipment to be predicted to a banking website, namely reducing the data volume in the equipment to be predicted.

Further, in some embodiments, the historical transaction data of the device to be predicted may be input into a preset data amount prediction model to perform data amount prediction, so as to obtain the remaining data amount of the device to be predicted. Wherein the data amount prediction model is similar to the above prediction model. For example, the historical actual data amount of the device to be predicted in each window length time period can be input into a preset data amount prediction model to perform data amount prediction, so that the residual data amount of the device to be predicted is obtained.

Further, in some embodiments, the remaining data amount may be compared with the target transaction amount, and if the remaining data amount is greater than or equal to the target transaction amount, it is determined that the data amount needs to be recovered for the device to be predicted, and the data amount needs to be recovered is determined according to the remaining data amount and the target transaction amount; if the residual data volume is smaller than the target transaction volume, determining that the data volume of the equipment to be predicted needs to be increased, and determining the data volume needing to be increased according to the residual data volume and the target transaction volume. The difference between the remaining data amount and the target transaction amount may be set as the data amount to be recovered, or the difference between the sum of the remaining data amount and the target transaction amount and the preset upper limit of the data amount may be determined as the data amount to be recovered.

According to the transaction amount prediction method for the bank equipment, the window length is determined based on the equipment type and the abnormal frequency of the transaction amount, meanwhile, the historical experience and the type factor are considered, the accuracy of the window length is improved, and the accuracy of determining the target transaction reference amount of the equipment to be predicted according to the length of the sliding window is improved, so that the accuracy of correcting the predicted transaction amount obtained based on the historical transaction data through the target transaction reference amount of the equipment to be predicted is improved, and the accuracy of predicting the transaction amount of the bank equipment is improved.

In some embodiments of the present invention, in order to better distinguish the device types of the banking device, relationship data between the device and the type may be pre-established, and after determining the device identifier of the device to be predicted, the relationship data between the device and the type may be queried to determine the device type of the device to be predicted, where the relationship data between the device and the type includes a plurality of device identifiers and the device type associated with each device identifier.

In some embodiments of the present invention, clustering may be performed according to basic data of each bank device to obtain a device type of each bank device, and the device type and the device identifier of each bank device are stored in an associated manner to obtain relationship data of the device and the type. Specifically, the device-type relationship data management method includes:

(1) And according to the basic data of each device, performing data cleaning and aggregation on the basic data to generate basic aggregation data centering on each device. The basic data comprises business attribute data, deployment environment attribute data and space-time association attribute data.

(2) And carrying out classification screening and clustering treatment on the basic aggregate data to obtain the type of the attribution of each device, and associating the device identifier of each device with the type of the attribution of the device to obtain the relationship data of the device and the type.

Further, in some embodiments, the business attribute data may be obtained through an in-line system of the bank. The business attribute data includes, but is not limited to, a type of a banking website to which the device belongs, a hierarchy of the banking website to which the device belongs, an upper limit of data volume of the device, an area to which the device belongs, a business type to which the device belongs, and the like. The service type of the equipment comprises internal service and external service, the equipment is characterized as bank external equipment when the service type of the equipment is external service, and the equipment is characterized as bank internal equipment when the service type of the equipment is internal service.

Further, in some embodiments, deployment environment attribute data may be obtained through in-line systems and web crawlers of a bank. The deployment environment attribute data includes, but is not limited to, the economic type of the city in which the equipment is located, the population scale of the city in which the equipment is located, the position coordinates of the equipment deployment, the location type of the equipment deployment, the people flow density of the surrounding environment of the equipment deployment, and the like.

Further, in some embodiments, the spatio-temporal correlation attribute data characterizes the temporal or spatial correlation attribute data of the device. May be acquired through an in-line system and a map service system. Wherein the time-space association attribute data includes, but is not limited to, the number of devices at the same location, the average transaction amount of the devices at the same location, the distance between the devices and the banking outlets, and the like.

Further, in some embodiments the in-line system of the bank includes a plurality of devices and the underlying data for each device. Wherein the in-line system distinguishes each device according to its device identification.

In some embodiments of the present invention, after obtaining the basic data, since the basic data contains a large amount of noise, the basic data is directly used for clustering, which affects the accuracy of the final device type. Therefore, after the basic data is obtained, the embodiment of the invention generates the initial basic aggregation data based on the basic data, and performs data cleaning and aggregation on the initial basic aggregation data.

Further, in some embodiments, data cleansing includes, but is not limited to, culling duplicate data in the initial base aggregate data and repairing missing data, outlier data, and outlier data in the initial base aggregate data. For example, the missing data may be filled with a near-four-week mean or median; according to the normal distribution 3 sigma principle, data with the standard deviation three times of that of the initial basic aggregate Data can be regarded as noise Data, and the noise Data is subjected to smoothing processing by adopting a Smooth Data function; the outlier data can be determined through the upper edge and the lower edge of the box diagram, and the outlier data is filled in a front-back four-week average filling mode.

Further, in some embodiments, after data cleansing is completed, the cleansed base data for each device may be aggregated according to the device identification for each device, generating base aggregate data centered on each device.

Further, in some embodiments, clustering may be performed on the basic aggregate data according to a K-means algorithm, to obtain a type to which each device belongs. Specifically, the method comprises the steps of performing feature engineering processing on basic aggregate data to generate multi-dimensional feature data of each device, and clustering through a K-mean algorithm according to most of the feature data of each device to obtain the attribution type of each device. Among them, feature engineering processes include, but are not limited to, feature construction, feature selection, feature extraction, and normalization processes.

Further, in some embodiments, after obtaining the device type of each device, the device type of each device and the device identifier of the device are stored in an associated manner, so as to obtain the relationship data of the device and the type.

In some embodiments of the present invention, in determining the device identifier of the device to be predicted, the historical transaction data and the abnormal transaction amount frequency of the device to be predicted may be determined according to step 201, and the preset relationship data between the device and the type may be queried according to the device identifier, so as to obtain the device type of the device to be predicted.

In some embodiments of the present invention, after obtaining the historical transaction data of the device to be predicted, the historical transaction data may be input to a preset prediction model to predict the transaction amount according to step 202, so as to obtain the predicted transaction amount of the device to be predicted.

Further, in some embodiments, statistical analysis may be performed on historical transaction data to obtain statistical characteristics of the historical transaction data, the historical transaction data is input into a preset prediction model to predict transaction amounts to obtain initial predicted transaction amounts of equipment to be predicted, the predicted transaction amounts of the equipment to be predicted are obtained according to the statistical characteristics and the initial predicted transaction amounts, the prediction model is exemplified by GBoost-based model, the median of data amounts of historical outflow data of the statistical characteristics, average is performed according to current flowing data, current flowing data of the previous day and current flowing data of the previous two days in the historical transaction data of the equipment to be predicted to obtain average flowing data of the equipment to be predicted on three days, the average flowing data of the three days is set as current flowing data, the smooth flowing data of the previous day and the smooth flowing data of the previous two days are determined according to the method to obtain the smooth flowing data of the equipment to be predicted on each day in the past period, the smooth flowing data of the equipment to be predicted on each day is input into the preset prediction model to obtain the current flowing data of the current flowing data, and the initial flowing data of the current flowing data is set as the predicted flowing data of the current; and obtaining the predicted transaction amount of the equipment to be predicted according to the median of the data amount of the historical outflow data and the initial predicted transaction amount.

In some embodiments of the present invention, when the predicted transaction amount is obtained, a preset window length of the device to be predicted may be obtained, and the target transaction amount is obtained by predicting the transaction amount by the preset window length.

In some embodiments of the present invention, in order to improve accuracy of a target transaction amount, the embodiment of the present invention sets a window length of a device to be predicted according to a device type of the device to be predicted and abnormal frequency of the transaction amount, groups historical transaction data of the device to be predicted according to the window length, obtains a plurality of groups of transaction data, and obtains a target transaction reference amount of the device to be predicted according to the historical transaction amount in each group of transaction data.

Further, in some embodiments, the window length of the device history transaction data to be predicted may be determined according to the method in step 203.

Further, in other embodiments, to improve accuracy of the window length, a history increasing frequency of the transaction object of the device to be predicted may be obtained, respective corresponding window correction parameters are determined according to the history increasing frequency, the device type and the abnormal frequency of the transaction amount, a target window correction parameter is obtained according to the respective corresponding window correction parameters of the history increasing frequency, the device type and the abnormal frequency of the transaction amount, and the window length of the device to be predicted is obtained according to the preset window length and the target window correction parameter. The method for determining the window correction parameters corresponding to the history increasing frequency, the equipment type and the transaction amount abnormal frequency is similar to the method for determining the window correction parameters in step 203.

Further, in other embodiments, to improve accuracy of the window length, a historical increasing frequency of the transaction object of the device to be predicted may be obtained, an initial window length is determined according to the abnormal frequency of the transaction amount of the device to be predicted and the historical increasing frequency of the transaction object of the device to be predicted, a target window correction parameter is determined according to the device type of the device to be predicted, and the window length of the device to be predicted is obtained according to the initial window length and the target window correction parameter. Specifically, the window length determining method of the device to be predicted includes steps a1 to a4:

step a1, obtaining historical increasing frequency of a transaction object of equipment to be predicted and total abnormal times of transaction amounts of all banking equipment.

Wherein the historical increasing frequency of the transaction object characterizes frequent cases of increasing the data amount in the device to be predicted in the past period of time, for example, when the historical increasing frequency is 1, the data amount in the device to be predicted needs to be increased every day in the past period of time, namely, the data amount in the device to be predicted is frequent, and when the historical increasing frequency is 14, the data amount in the device to be predicted needs to be increased every 14 days in the past period of time, namely, the data amount in the device to be predicted is not frequent; the determination may be based on a ratio of the number of increases in the amount of data in the device to be predicted over the past period of time to the length of time over the past period of time. Wherein the transaction object is for the device to be predicted to perform a transaction including, but not limited to, data volume, data flow, goods, financial products, money, and the like. Wherein the financial products include financial products such as funds, insurance, and the like.

The total number of transaction amount anomalies refers to the total number of transaction amount anomalies for all banking devices in the bank over a period of time. In some embodiments, abnormal frequency of transaction amounts of all bank devices in the bank can be obtained through an in-line system.

And a2, determining an initial window length of historical transaction data of the equipment to be predicted according to the ratio of the transaction amount anomaly frequency of the equipment to be predicted in the total transaction amount anomaly frequency and the historical increasing frequency of the transaction object of the equipment to be predicted.

Further, in some embodiments, the window coefficient may be obtained according to the ratio of the abnormal frequency of the transaction amount of the device to be predicted in the total abnormal frequency of the transaction amount and the historical increasing frequency of the transaction object of the predicting device, and the pre-stored relationship data of the window coefficient and the length may be queried to obtain the initial window length of the device to be predicted. The window coefficient and length relation data comprises a plurality of window coefficients and window lengths corresponding to the window coefficients. In some embodiments, the sum, product, average, or weighted sum of the duty ratio in the total number of transaction amount anomalies according to the transaction amount anomaly frequency of the device to be predicted and the historical increase frequency of the transaction object of the predicting device may be set as the window coefficient.

Further, in other embodiments, it may be determined whether the device to be predicted is a device that is prone to occurrence of a transaction amount abnormality according to a ratio of the transaction amount abnormality frequency of the device to be predicted to the total number of transaction amount abnormalities; if the equipment to be predicted is equipment with abnormal transaction amount, acquiring a preset increasing frequency, and determining the initial window length according to the preset increasing frequency and the historical increasing frequency; if the equipment to be predicted is not equipment which is easy to cause abnormal transaction amount, determining the initial window length according to the historical increasing frequency.

In some embodiments, if the device to be predicted is a device that is prone to abnormal transaction amount, the average, the sum, or the maximum of the preset increase frequency and the historical increase frequency may be determined as the initial window length; if the device to be predicted is not a device that is prone to transaction volume anomalies, the historical increasing frequency may be determined as the initial window length.

Taking as an example that the maximum value of the preset increasing frequency and the historical increasing frequency is determined as the initial window length, the method for determining the initial window length specifically includes:

(1) And if the duty ratio of the abnormal frequency of the transaction amount of the equipment to be predicted in the total abnormal frequency of the transaction amount is larger than or equal to a preset proportion threshold value, setting the maximum value of the preset increasing frequency and the historical increasing frequency as the initial window length of the historical transaction data of the equipment to be predicted.

(2) If the duty ratio of the abnormal frequency of the transaction amount of the equipment to be predicted in the total abnormal frequency of the transaction amount is smaller than a preset proportion threshold value, determining the initial window length of the historical transaction data of the equipment to be predicted according to the historical increasing frequency.

In some embodiments, the ratio of the transaction amount anomaly frequency of the device to be predicted in the total transaction amount anomaly times may be compared with a preset proportion threshold, and if the ratio of the transaction amount anomaly frequency of the device to be predicted in the total transaction amount anomaly times is greater than or equal to the preset proportion threshold, which indicates that the device to be predicted is a device with a transaction amount anomaly easily occurring, the maximum value of the preset increase frequency and the historical increase frequency is set as the initial window length of the device to be predicted; if the ratio of the abnormal frequency of the transaction amount of the equipment to be predicted in the total abnormal frequency of the transaction amount is smaller than the preset proportion threshold value, which indicates that the equipment to be predicted is not equipment with abnormal transaction amount easily, setting the historical increasing frequency of the transaction object of the equipment to be predicted as the initial window length of the equipment to be predicted.

For example, taking the example that the historical increasing frequency is 1, the preset increasing frequency is 14 and the preset proportion threshold value is 5%, when the ratio of the abnormal frequency of the transaction amount of the equipment to be predicted in the total abnormal times of the transaction amount is greater than or equal to 5%, the initial window length is set to be 14, namely, the data amount in the equipment to be predicted is increased by predicting the transaction amount of 14 days in the future, so that the database in the equipment to be predicted can meet the future transaction, and the abnormal times of the transaction amount of the equipment to be predicted in the future transaction are reduced; when the ratio of the abnormal frequency of the transaction amount of the equipment to be predicted in the total abnormal frequency of the transaction amount is less than 5%, the initial window length is set to be 1, namely, the transaction amount of the equipment to be predicted in the open day is predicted, and only the transaction amount in the open day is predicted to meet the transaction of the equipment to be predicted because the equipment to be predicted is not easy to generate the abnormal transaction amount.

In some embodiments, the historical increasing frequency may be determined as an initial window length of historical transaction data for the device to be predicted; in other embodiments, the preset window parameter may also be obtained, and the product of the preset open length and the historical increasing frequency is set to be the initial window length of the historical transaction data of the device to be predicted.

And a3, inquiring relation data between the predicted type and the window correction parameters according to the equipment type of the equipment to be predicted, and obtaining a target window correction coefficient corresponding to the equipment type of the equipment to be predicted.

The target window correction coefficient is used for determining the growth rate of the initial window length; the relationship data between the type and the window correction parameters comprises a plurality of device types and window correction coefficients corresponding to each device type.

In some embodiments of the present invention, the requirements of devices of different device types on the window length are considered to be different, so that the embodiments of the present invention determine the target window correction coefficient according to the device type of the device to be predicted.

In some embodiments, the target window correction coefficient corresponding to the device type of the device to be predicted may be obtained according to the device type of the device to be predicted and according to the method for determining the window correction parameter in step 203.

And a4, determining the window length of the historical transaction data of the equipment to be predicted according to the target window correction coefficient and the initial window length.

Further, in some embodiments the product of the target window correction factor and the initial window length may be set to the window length of the historical transaction data of the device to be predicted.

Further, in other embodiments, the window correction length may be obtained according to a product of the target window correction coefficient and the initial window length, and the sum of the initial window length and the window correction length is set as the window length of the device to be predicted. For example, when the target window correction factor is 10%, characterization requires that the window length be obtained by the initial window length (1+10%).

In some embodiments of the present invention, after determining the window length, dividing equally according to the window length according to the time length of the historical transaction data of the device to be predicted, to obtain sub-historical transaction data of each time period and each time period, and setting the sub-historical transaction data of each time period as a group of transaction data. For example, when the time length of the historical transaction data is 6 days and the window length is 3, the historical transaction data is divided into two groups, each group containing three days of sub-historical transaction data.

In other embodiments of the present invention, after determining the window length, the sub-historical transaction data in the window length period from the historical transaction data of the device to be predicted may be a set of transaction data with the day as a starting point, and the sub-historical transaction data in the window length period from the historical transaction data of the device to be predicted may be a set of transaction data with the day before the day as a new starting point, until all days of the past period are traversed, and the historical transaction data is divided into multiple sets of transaction data. For example, when the time length of the historical transaction data is 6 days and the window length is 3, the sub-historical transaction data corresponding to each of the sixth day, the fifth day and the fourth day is taken as the first group of transaction data, the sub-historical transaction data corresponding to each of the fifth day, the fourth day and the third day is taken as the second group of transaction data, the sub-historical transaction data corresponding to each of the fourth day, the third day and the second day is taken as the third group of transaction data, and the sub-historical transaction data corresponding to each of the third day, the second day and the first day is taken as the fourth group of transaction data, so that four groups of transaction data are obtained.

In some embodiments of the present invention, after obtaining multiple sets of transaction data, the average value of the set of transaction data may be calculated for each set of transaction data, and according to the average value of each set of transaction data, the difference between the maximum average value and the minimum average value in the multiple sets of transaction data is determined, and the difference between the maximum average value and the minimum average value in the multiple sets of transaction data is set as the target transaction reference amount of the device to be predicted.

In other embodiments of the present invention, after obtaining a plurality of sets of transaction data, for each set of transaction data, a difference between a maximum transaction amount and a minimum transaction amount in the set of transaction data is determined, the difference between the maximum transaction amount and the minimum transaction amount in the set of transaction data is set to be a range of the set of transaction data, and the maximum range is set as a target transaction reference amount of the device to be predicted based on the range of each set of transaction data.

In other embodiments of the present invention, after obtaining multiple sets of transaction data, for each set of transaction data, determining a difference between a maximum transaction amount and a minimum transaction amount in the set of transaction data, setting the difference between the maximum transaction amount and the minimum transaction amount in the set of transaction data as a range of the set of transaction data, counting the range of each set of transaction data to obtain a range statistics feature, and setting the range statistics feature as a target transaction reference amount of a device to be predicted. The range statistics include, but are not limited to, median, mode, difference between maximum and minimum, etc.

In some embodiments of the present invention, considering that a banking device may have the same type of continuous transaction in a past period of time, that is, the banking device may have the same type of continuous transaction in the future, it is ensured that the data amount in the banking device can satisfy the data amount of the same type of continuous transaction in the future, and when the target transaction amount is predicted, the target transaction reference amount needs to be determined according to the transaction amount of the same type of continuous transaction in the past period of time of the device to be predicted and each group of transaction data, thereby improving the accuracy of the target transaction data. Specifically, the method for determining the target transaction reference quantity comprises the following steps of b 1-b 2:

Step b1, determining the statistical characteristics of each group of transaction data. Wherein the statistical features include average historical transaction amount and historical transaction amount tolerance values.

And b2, determining target transaction reference quantity of the equipment to be predicted according to the statistical characteristics of each group of transaction data.

In some embodiments, the historical transaction amount for each set of transaction data may be smoothed according to the window length to obtain an average historical transaction amount for each set of transaction data. For example, the average historical transaction amount for each set of transaction data is obtained by smoothing the historical transaction amount/window length in each set of transaction data.

In other embodiments, for each set of transaction data, a maximum historical transaction amount for a unit duration and a minimum historical transaction amount for the unit duration in the set of transaction data are determined, and a difference between the maximum historical transaction amount and the minimum historical transaction amount is set as a historical transaction amount tolerance value. Wherein the unit time period may be 1 day.

In some embodiments, the maximum, average, median, mode, or weighted average of the statistical characteristics of the sets of transaction data may be determined as a target transaction reference for the device to be predicted.

In other embodiments, to improve accuracy of transaction amount prediction, it may be determined whether the device to be predicted has continuous transactions of the same type within the unit duration, and when the device to be predicted has continuous transactions of the same type within the unit duration, determining the target transaction reference amount of the device to be predicted according to the transaction amounts of the continuous transactions of the same type within the unit duration and statistical features of each set of transaction data, specifically, step b2 includes:

(1) Determining whether the equipment to be predicted has continuous transactions of the same type in the unit duration according to the historical transaction data;

(2) If the equipment to be predicted has continuous transactions of the same type in the unit duration, acquiring the total transaction amount of the continuous transactions of the same type, and acquiring a target transaction reference amount of the equipment to be predicted according to the statistical characteristics of each group of transaction data and the total transaction amount.

In some embodiments, the same type of continuous transaction includes a continuous inflow transaction and a continuous outflow transaction.

In some embodiments, it may be determined from sub-historical transaction data for each day over a period of time whether there is a continuous outflow transaction or a continuous inflow transaction for the device to be predicted for a unit of time; if each day in the past period does not have a continuous outflow transaction or a continuous inflow transaction, determining that the device to be predicted does not have the same type of continuous transaction in the unit duration; if there is a continuous outflow transaction or a continuous inflow transaction for at least one day over a period of time, it is determined that the device to be predicted has the same type of continuous transaction for the unit duration.

Further, in some embodiments, if the device to be predicted has the same type of continuous transaction in the unit duration, the date on which the same type of continuous transaction exists in the unit duration and the transaction amount of the continuous transaction on the date may be determined, and the transaction amount of the continuous transaction on each date may be counted to obtain the transaction total amount of the same type of continuous transaction.

In other embodiments, whether there is a continuous outgoing transaction or a continuous incoming transaction per day in the time range in which the set of transaction data is located may be determined by sub-historical transaction data per day in each set of transaction data, and if there is no continuous outgoing transaction or continuous incoming transaction per day in the time range in which each set of transaction data is located, it is determined that the device to be predicted does not have a continuous transaction of the same type for a unit duration; if the target group transaction data of the continuous outflow transaction or the continuous inflow transaction exists in a certain day in the plurality of groups of transaction data, the device to be predicted is determined to exist in the same type of continuous transaction in the unit duration.

Further, in other embodiments, if the device to be predicted has the same type of continuous transaction within the unit duration, a date on which the same type of continuous transaction exists in one day in the target group transaction data and a transaction amount of the continuous transaction on the date are obtained, the transaction amounts of the continuous transactions on each date are summarized, the transaction amounts of the same type of continuous transaction in the target group transaction data are obtained, and the transaction amounts of the same type of continuous transaction in each target group transaction data are summarized to obtain a transaction total amount of the same type of continuous transaction.

Further, in some embodiments, the statistical characteristics of each set of transaction data and the maximum value of the total amount of transactions may be set as a target transaction reference amount for the device to be predicted; in other embodiments, the statistical characteristics of the sets of transaction data and the average of the total amount of transactions may be set as target transaction references for the device to be predicted.

In some embodiments of the present invention, if the device to be predicted does not have the same type of continuous transaction within the unit time, the statistical characteristics of each set of transaction data determine a target transaction reference quantity of the device to be predicted. For example, when the statistical feature is an average historical transaction amount, a maximum value of the average historical transaction amount of each set of transaction data may be set as a target transaction reference amount of the device to be predicted. When the statistical feature is a historical transaction amount limit value, the historical transaction amount limit value of each set of transaction data may be set as a target transaction reference amount of the device to be predicted. When the statistical feature is an average historical transaction amount, a difference between a maximum value of the average historical transaction amount and a minimum value of the average historical transaction amount of each set of transaction data may be set as a target transaction reference amount of the device to be predicted.

In some embodiments of the present invention, after determining the target transaction reference amount, a maximum value between the target transaction reference amount and the predicted transaction amount may be set as the target transaction amount of the device to be predicted, specifically including: comparing the target transaction reference quantity with the predicted transaction quantity; if the target transaction reference quantity is larger than or equal to the predicted transaction quantity, setting the target transaction reference quantity as the target transaction quantity of the equipment to be predicted; if the target transaction reference quantity is smaller than the predicted transaction quantity, the predicted transaction quantity is set as the target transaction quantity of the equipment to be predicted.

In other embodiments of the present invention, after determining the target transaction reference amount, weights corresponding to the target transaction reference amount and the predicted transaction amount may be obtained, a weighted average of the target transaction reference amount and the predicted transaction amount is determined according to the target transaction reference amount and the predicted transaction amount, and the weighted average of the target transaction reference amount and the predicted transaction amount is set as the target transaction amount of the device to be predicted.

According to the transaction amount prediction method of the bank equipment, the window length is determined based on the equipment type and the abnormal frequency of the transaction amount, meanwhile, the historical experience and the type factor are considered, the accuracy of the window length is improved, and the accuracy of determining the target transaction reference amount of the equipment to be predicted according to the length of the sliding window is improved, so that the accuracy of correcting the predicted transaction amount obtained based on the historical transaction data through the target transaction reference amount of the equipment to be predicted is improved, and the accuracy of predicting the transaction amount of the bank equipment is improved.

In order to better implement the transaction amount predicting method of the banking device provided by the embodiment of the present invention, on the basis of the transaction amount predicting method of the banking device, the embodiment of the present invention provides a transaction amount predicting device of the banking device, as shown in fig. 3, fig. 3 is a schematic structural diagram of the transaction amount predicting device of the banking device provided by the embodiment of the present invention, where the transaction amount predicting device of the banking device includes:

the window determining module is used for determining the window length of historical transaction data of the equipment to be predicted according to the abnormal frequency of the transaction amount of the equipment to be predicted and the equipment type of the equipment to be predicted;

the reference quantity determining module is used for obtaining a target transaction reference quantity of the equipment to be predicted according to the historical transaction quantity in each group of transaction data;

In some embodiments of the present invention, the window determining module is configured to:

acquiring historical increasing frequency of a transaction object of equipment to be predicted and total transaction amount abnormality times of all banking equipment;

determining an initial window length of historical transaction data of the equipment to be predicted according to the ratio of the transaction amount anomaly frequency of the equipment to be predicted in the total transaction amount anomaly frequency and the historical increasing frequency of the transaction object of the equipment to be predicted;

inquiring relation data between the predicted type and window correction parameters according to the equipment type of the equipment to be predicted, and obtaining a target window correction coefficient corresponding to the equipment type of the equipment to be predicted;

and determining the window length of the historical transaction data of the equipment to be predicted according to the target window correction coefficient and the initial window length.

In some embodiments of the invention, a window module is configured to:

if the duty ratio of the abnormal frequency of the transaction amount of the equipment to be predicted in the total abnormal frequency of the transaction amount is larger than or equal to a preset proportion threshold value, determining the initial window length of the historical transaction data of the equipment to be predicted according to the preset increasing frequency and the historical increasing frequency;

if the duty ratio of the abnormal frequency of the transaction amount of the equipment to be predicted in the total abnormal frequency of the transaction amount is smaller than a preset proportion threshold value, determining the initial window length of the historical transaction data of the equipment to be predicted according to the historical increasing frequency.

In some embodiments of the invention, the reference quantity determination module is configured to:

determining the statistical characteristics of each group of transaction data, wherein the statistical characteristics comprise average historical transaction quantity and historical transaction quantity extreme values;

and determining target transaction reference quantity of the equipment to be predicted according to the statistical characteristics of each group of transaction data.

determining whether the equipment to be predicted has continuous transactions of the same type in the unit duration according to the historical transaction data;

if the equipment to be predicted has continuous transactions of the same type in the unit duration, acquiring the total transaction amount of the continuous transactions of the same type, and obtaining a target transaction reference amount of the equipment to be predicted according to the statistical characteristics of each group of transaction data and the total transaction amount.

In some embodiments of the invention, the acquisition module is configured to:

according to the basic data of each device, carrying out data cleaning and aggregation on the basic data to generate basic aggregation data centering on each device; the basic data comprises service attribute data, deployment environment attribute data and real control associated attribute data;

clustering the basic aggregation data to obtain the type of each equipment, and associating the equipment identifier of each equipment with the type of the equipment to obtain the relationship data of the equipment and the type;

And acquiring the equipment identification of the equipment to be predicted, and determining the equipment type of the equipment to be predicted according to the relation data of the equipment and the type.

In some embodiments of the invention, the first prediction module is configured to:

and inputting the historical transaction data into a preset prediction model to predict the transaction amount, so as to obtain the predicted transaction amount of the equipment to be predicted.

According to the transaction amount prediction device of the bank equipment, the window length is determined based on the equipment type and the abnormal frequency of the transaction amount, meanwhile, the historical experience and the type factor are considered, the accuracy of the window length is improved, so that the accuracy of determining the target transaction reference amount of the equipment to be predicted according to the length of the sliding window is improved, the accuracy of correcting the predicted transaction amount obtained based on the historical transaction data through the target transaction reference amount of the equipment to be predicted is improved, and the accuracy of predicting the transaction amount of the bank equipment is improved.

The embodiment of the invention also provides a transaction amount prediction device of the banking device, as shown in fig. 4, which shows a schematic structural diagram of the transaction amount prediction device of the banking device according to the embodiment of the invention, specifically:

The transaction amount predicting device of the banking device may include one or more processors 401 of a processing core, one or more memories 402 of a computer readable storage medium, a power supply 403, and an input unit 404, etc. It will be appreciated by those skilled in the art that the configuration of the transaction amount predicting device of the banking device shown in fig. 4 does not constitute a limitation of the transaction amount predicting device of the banking device, and may include more or less components than illustrated, or may combine certain components, or may be a different arrangement of components. Wherein:

the processor 401 is a control center of the transaction amount predicting device of the banking device, connects respective parts of the transaction amount predicting device of the entire banking device using various interfaces and lines, and performs various functions and processing data of the transaction amount predicting device of the banking device by running or executing software programs and/or modules stored in the memory 402 and calling data stored in the memory 402, thereby performing overall monitoring of the transaction amount predicting device of the banking device. Optionally, processor 401 may include one or more processing cores; preferably, the processor 401 may integrate an application processor and a modem processor, wherein the application processor mainly processes an operating system, a user interface, an application program, etc., and the modem processor mainly processes wireless communication. It will be appreciated that the modem processor described above may not be integrated into the processor 401.

The memory 402 may be used to store software programs and modules, and the processor 401 executes various functional applications and data processing by executing the software programs and modules stored in the memory 402. The memory 402 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program (such as a sound playing function, an image playing function, etc.) required for at least one function, and the like; the storage data area may store data created from the use of the transaction amount predicting device of the banking device, and the like. In addition, memory 402 may include high-speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other volatile solid-state storage device. Accordingly, the memory 402 may also include a memory controller to provide the processor 401 with access to the memory 402.

The transaction amount predicting device of the banking device further comprises a power supply 403 for supplying power to each component, and preferably, the power supply 403 may be logically connected to the processor 401 through a power management system, so that functions of managing charging, discharging, power consumption management and the like are realized through the power management system. The power supply 403 may also include one or more of any of a direct current or alternating current power supply, a recharging system, a power failure detection circuit, a power converter or inverter, a power status indicator, and the like.

The transaction amount predicting device of the banking device may further comprise an input unit 404, which input unit 404 may be used for receiving input numerical or character information and generating keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control.

Although not shown, the transaction amount prediction device of the banking device may further include a display unit or the like, which is not described herein. In particular, in this embodiment, the processor 401 in the transaction amount predicting device of the banking device loads executable files corresponding to the processes of one or more application programs into the memory 402 according to the following instructions, and the processor 401 executes the application programs stored in the memory 402, so as to implement various functions as follows:

predicting transaction amount according to the historical transaction data to obtain predicted transaction amount of equipment to be predicted;

grouping historical transaction data according to window length to obtain a plurality of groups of transaction data;

Obtaining target transaction reference quantity of equipment to be predicted according to the historical transaction quantity in each group of transaction data;

Those of ordinary skill in the art will appreciate that all or a portion of the steps of the various methods of the above embodiments may be performed by instructions, or by instructions controlling associated hardware, which may be stored in a computer-readable storage medium and loaded and executed by a processor.

To this end, an embodiment of the present invention provides a storage medium having stored therein a plurality of instructions that can be loaded by a processor to perform the steps in any of the transaction amount prediction methods of banking devices provided by the embodiment of the present invention. For example, the instructions may perform the steps of:

The specific implementation of each operation above may be referred to the previous embodiments, and will not be described herein.

Wherein the storage medium may include: read Only Memory (ROM), random access Memory (RAM, random Access Memory), magnetic or optical disk, and the like.

The instructions stored in the storage medium can execute the steps in the transaction amount prediction method of any kind of banking equipment provided by the embodiment of the present invention, so that the beneficial effects that can be achieved by the transaction amount prediction method of any kind of banking equipment provided by the embodiment of the present invention can be achieved, which are detailed in the previous embodiments and are not described herein.

The foregoing describes in detail the transaction amount prediction method, apparatus, device and storage medium of a banking device provided by the embodiments of the present invention, and specific examples are applied to illustrate the principles and embodiments of the present invention, and the foregoing examples are only used to help understand the method and core idea of the present invention; meanwhile, as those skilled in the art will have variations in the specific embodiments and application scope in light of the ideas of the present invention, the present description should not be construed as limiting the present invention.

Claims

1. A transaction amount prediction method for a banking apparatus, the method comprising:

2. The method for predicting transaction amount of a banking device as claimed in claim 1, wherein said determining a window length for obtaining historical transaction data of the device to be predicted based on the abnormal frequency of transaction amount of the device to be predicted and the type of device of the device to be predicted includes:

acquiring historical increasing frequency of a transaction object of the equipment to be predicted and total transaction amount abnormality times of all bank equipment;

Determining an initial window length of historical transaction data of the equipment to be predicted according to the duty ratio of the transaction amount anomaly frequency of the equipment to be predicted in the total transaction amount anomaly times and the historical increasing frequency of the transaction objects of the equipment to be predicted;

3. The method for predicting transaction amount of a banking device as claimed in claim 2, wherein the determining an initial window length for obtaining historical transaction data of the device to be predicted according to a ratio of the transaction amount anomaly frequency of the device to be predicted to the total number of transaction amount anomalies and a historical increasing frequency of a transaction object of the predicting device includes:

if the duty ratio of the abnormal frequency of the transaction amount of the equipment to be predicted in the total abnormal frequency of the transaction amount is larger than or equal to a preset proportion threshold value, determining the initial window length of the historical transaction data of the equipment to be predicted according to a preset increasing frequency and the historical increasing frequency;

4. The method for predicting transaction amount of a banking device as claimed in claim 1, wherein the obtaining the target transaction reference amount of the device to be predicted based on the historical transaction amount in each set of transaction data includes:

determining a statistical feature of each set of the transaction data; the statistical features include average historical transaction amount and historical transaction amount extreme difference

And determining target transaction reference quantity of the equipment to be predicted according to the statistical characteristics of the transaction data of each group.

5. The method for predicting transaction amount of a banking device as claimed in claim 4, wherein said determining a target transaction reference amount of the device to be predicted based on statistical characteristics of each set of the transaction data includes:

determining whether the equipment to be predicted has continuous transactions of the same type in unit duration according to the historical transaction data;

and if the equipment to be predicted has continuous transactions of the same type in the unit duration, acquiring the transaction total amount of the continuous transactions of the same type, and acquiring a target transaction reference amount of the equipment to be predicted according to the statistical characteristics of the transaction data of each group and the transaction total amount.

6. The transaction amount predicting method of a banking device as claimed in claim 1, wherein before the device type of the device to be predicted, the historical transaction data, and the transaction amount abnormality frequency are acquired, the method includes:

according to the basic data of each device, carrying out data cleaning and aggregation on the basic data to generate basic aggregation data centering on each device; the basic data comprise business attribute data, deployment environment attribute data and real control associated attribute data;

clustering the basic aggregate data to obtain the type of the equipment attribution, and associating the equipment identifier of the equipment with the type of the equipment attribution to obtain the relation data of the equipment and the type;

the obtaining the equipment type of the equipment to be predicted comprises the following steps:

7. The transaction amount predicting method of a banking device according to any one of claims 1 to 6, wherein predicting the transaction amount according to the historical transaction data, obtaining the predicted transaction amount of the device to be predicted includes:

8. A transaction amount predicting apparatus of a banking device, characterized by comprising:

9. A transaction amount predicting device for a banking device, comprising a memory and a processor; the memory stores an application program, and the processor is configured to execute the application program in the memory to perform the operations in the transaction amount prediction method of the banking device according to any one of claims 1 to 7.

10. A storage medium storing a plurality of instructions adapted to be loaded by a processor to perform the steps in the transaction amount prediction method of a banking device as claimed in any one of claims 1 to 7.