Method and device for determining inter-bank fund flow and computer program product

Abstract

The invention provides an inter-bank fund flow determination method and apparatus, and a computer program product. The method comprises the steps of obtaining current inter-bank fund flow feature data; the current inter-row fund flow feature data is input into a corresponding inter-row fund flow determination model, a corresponding model output result is obtained, and the inter-row fund flow determination model comprises a first inter-row fund flow determination model, a second inter-row fund flow determination model and a third inter-row fund flow determination model; and carrying out weighted combination on the output results of the plurality of models to obtain a target inter-row fund flow. According to the method, the problems that inter-row fund flow fluctuation characteristics are not fully considered and the prediction precision of the inter-row fund flow is low due to direct discarding of residual data in the prior art are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of inter-bank fund flow prediction, and in particular to a method for determining inter-bank fund flow, a determining device, a computer-readable storage medium, and a computer program product. Background Art

[0002] In the financial industry, interbank fund flows involve the transfer of funds between multiple banks and financial institutions. The accurate prediction of their flow direction and quantity is of great significance for risk management, fund scheduling and investment decisions. However, due to the complexity, uncertainty and dynamics of interbank fund flows, traditional statistical methods and single machine learning algorithms often fail to meet the prediction requirements. These methods often fail to accurately capture the complex changing trends of interbank fund flows. Since interbank fund flows involve complex interactions between multiple financial institutions and enterprises, existing methods often fail to fully consider the impact of multiple factors on interbank fund flows, including market demand, economic environment, policies and regulations, etc., resulting in inaccurate prediction results and inability to fully grasp the dynamic changes of fund flows. Therefore, improving the prediction accuracy of interbank fund flows is an urgent problem to be solved by current technology. Traditional interbank fund flow prediction methods are mainly based on statistics and time series analysis, such as regression analysis, ARIMA model, etc., but these methods often have high requirements on data and are not effective in predicting complex interbank fund flows. With the development of machine learning algorithms, such as decision trees, neural networks, support vector machines, etc., they have gradually been applied to the construction of interbank fund flow prediction. However, a single machine learning algorithm often cannot meet the prediction requirements.

[0003] At present, the prediction of interbank payment fund flows mainly relies on traditional statistical models, such as time series analysis and regression analysis. However, these models are mainly based on historical data for prediction, but the complexity and dynamics of the financial market limit the prediction accuracy of these models. The patents of existing fund flow prediction methods mainly have the following deficiencies: (1) Lack of accuracy: The interbank payment fund flow prediction methods in existing patent technologies are often limited by the limitations of a single model or algorithm and cannot provide highly accurate prediction results. For example, the single ARIMA model order determination does not use the grid search method, but determines the p value and q value by observing the autocorrelation diagram and the partial autocorrelation diagram, and errors are difficult to avoid. The characteristic value of LGB is selected from the interbank transaction details, and the feature importance is relatively small. The waveform of the Prophet model prediction result is relatively flat and the fluctuation range is small. LSTM is suitable for data analysis with multiple data dimensions and large data volume. At present, the use of LSTM for single-dimensional time series analysis cannot give full play to the advantages of the model. (2) Insufficient information utilization: The interbank payment fund flow prediction methods in existing patent technologies fail to make full use of relevant external data and indicators during the model construction process. These external data and indicators may include market changes, economic indicators, etc., which can provide a more comprehensive information basis for the forecasting model, but existing technologies fail to effectively integrate these data, resulting in a decrease in the comprehensiveness and accuracy of the forecast results. Summary of the invention

[0004] The main purpose of the present application is to provide a method for determining inter-bank fund flows, a determination device, a computer-readable storage medium and a computer program product, so as to at least solve the problem that the prior art does not fully consider the fluctuation characteristics of inter-bank fund flows and directly discards residual data, resulting in low prediction accuracy of inter-bank fund flows.

[0005] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a method for determining inter-bank capital flow is provided, comprising: obtaining current inter-bank capital flow characteristic data, wherein the current inter-bank capital flow characteristic data is divided into time series characteristic data, residual sequence characteristic data and characteristic values; inputting the current inter-bank capital flow characteristic data into a corresponding inter-bank capital flow determination model to obtain a corresponding model output result, wherein the inter-bank capital flow determination model comprises a first inter-bank capital flow determination model, a second inter-bank capital flow determination model and a third inter-bank capital flow determination model, wherein the first inter-bank capital flow determination model is a convergence model obtained by inputting sample time series characteristic data into an ARIMA model for training, the second inter-bank capital flow determination model is a convergence model obtained by inputting sample residual sequence characteristic data into an LSTM model for training, and the third inter-bank capital flow determination model is a convergence model obtained by inputting sample characteristic values ​​into an LGB model for training; performing weighted combination on the output results of multiple models to obtain a target inter-bank capital flow.

[0006] Optionally, before inputting the current inter-bank funds flow characteristic data into the corresponding inter-bank funds flow determination model to obtain the corresponding model output result, the method further includes: obtaining multiple groups of historical inter-bank funds flow characteristic data, the historical inter-bank funds flow characteristic data including the sample time series characteristic data, the sample residual sequence characteristic data and the sample characteristic value, the sample time series characteristic data including the periodic sequence and the trend sequence; a training step, respectively inputting the sample time series characteristic data, the sample residual sequence characteristic data and the sample characteristic value in each group of the historical inter-bank funds flow characteristic data into the corresponding predetermined training model for iterative training to obtain the corresponding training model. The predetermined training model includes the ARIMA model, the LSTM model and the LGB model; a testing step, testing all the training output results by a testing method to obtain a test result, and the testing method at least includes a residual white noise test and a root mean square error; when the test result does not meet the model accuracy requirement, the training step and the testing step are repeated at least once in sequence until the test result meets the model accuracy requirement or the training number reaches the maximum number of iterations, and the first inter-bank fund flow determination model, the second inter-bank fund flow determination model and the third inter-bank fund flow determination model are obtained respectively.

[0007] Optionally, multiple groups of historical inter-bank funds flow characteristic data are obtained, including: obtaining historical inter-bank funds flow data, the historical inter-bank funds flow data including inter-bank transaction details data and macroeconomic data within a historical period, the inter-bank transaction details data including at least transaction time, transaction amount and transaction direction, the macroeconomic data including at least macroeconomic factors, market environment factors, business cycle characteristics, intra-bank plans, key regions and holiday characteristics; performing preprocessing operations on the macroeconomic data in the historical inter-bank funds flow data to obtain the sample characteristic values, the preprocessing operations including at least outlier processing, missing value filling, normalization processing and outlier processing; using the STL decomposition method to extract time series characteristics and residual series characteristics in the inter-bank transaction details data to obtain the sample time series characteristic data and the residual series characteristic data, respectively, the time series characteristics including periodic information and trend information.

[0008] Optionally, a plurality of model output results are weightedly combined to obtain a target interbank funds flow, including: inputting the output result of the first model and the output result of the second model into the combination scheme to obtain a first interbank funds flow, the output result of the first model is the output result of the first interbank funds flow determination model, and the output result of the second model is the output result of the second interbank funds flow determination model; inputting the first interbank funds flow and the output result of the third model into the combination scheme to obtain the target interbank funds flow, the output result of the third model is the output result of the third interbank funds flow determination model.

[0009] Optionally, before inputting the output results of the first model and the output results of the second model into a combination scheme to obtain the first inter-bank funds flow, the method also includes: determining the combination scheme, the combination scheme being a scheme for obtaining the inter-bank funds flow by adding the first product and the second product, the first product being the product of a first weight coefficient and a first prediction result, the first prediction result being the output result of the first model, the first model being any one of the inter-bank funds flow determination models, the first weight coefficient being the weight coefficient corresponding to the first model, the second product being the product of a second weight coefficient and a second prediction result, the second prediction result being the output result of the second model, the second model being any one of the inter-bank funds flow determination models other than the first model, the second weight coefficient being the weight coefficient corresponding to the second model, and the sum of the first weight coefficient and the second weight coefficient being equal to 1.

[0010] Optionally, before determining the combination scheme, the method also includes: inputting all the training output results output by the first inter-bank funds flow determination model and the first sample inter-bank funds flow into a univariate linear regression model to obtain the weight coefficient corresponding to the first inter-bank funds flow determination model, and the first sample inter-bank funds flow is the sample inter-bank funds flow corresponding to the sample time series feature data; inputting all the training output results output by the second inter-bank funds flow determination model and the second sample inter-bank funds flow into the univariate linear regression model to obtain the weight coefficient corresponding to the second inter-bank funds flow determination model, and the second sample inter-bank funds flow is the sample inter-bank funds flow corresponding to the sample residual sequence feature data; inputting all the training output results output by the third inter-bank funds flow determination model and the third sample inter-bank funds flow into the univariate linear regression model to obtain the weight coefficient corresponding to the third inter-bank funds flow determination model, and the third sample inter-bank funds flow is the sample inter-bank funds flow corresponding to the sample characteristic value.

[0011] Optionally, after determining the combination scheme, the method also includes: obtaining a first accuracy and a second accuracy, the first accuracy being the accuracy of the output result of the first model, and the second accuracy being the accuracy of the output result of the second model; when the first accuracy is greater than the second accuracy, increasing the first weight coefficient and decreasing the second weight coefficient; when the first accuracy is less than the second accuracy, decreasing the first weight coefficient and increasing the second weight coefficient.

[0012] According to another aspect of the present application, a device for determining inter-bank capital flow is provided, the device comprising: a first acquisition unit, used to acquire current inter-bank capital flow feature data, the current inter-bank capital flow feature data is divided into time series feature data, residual sequence feature data and feature values; an input-output unit, used to input the current inter-bank capital flow feature data into a corresponding inter-bank capital flow determination model to obtain a corresponding model output result, the inter-bank capital flow determination model comprises a first inter-bank capital flow determination model, a second inter-bank capital flow determination model and a third inter-bank capital flow determination model, the first inter-bank capital flow determination model is a convergence model obtained by inputting sample time series feature data into an ARIMA model for training, the second inter-bank capital flow determination model is a convergence model obtained by inputting sample residual sequence feature data into an LSTM model for training, and the third inter-bank capital flow determination model is a convergence model obtained by inputting sample feature values ​​into an LGB model for training; a combination unit, used to weightedly combine the output results of a plurality of the models to obtain a target inter-bank capital flow.

[0013] According to another aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute any one of the methods described.

[0014] According to another aspect of the present application, a computer program product is provided, comprising computer instructions, wherein when the computer instructions are executed by a processor, any one of the methods described above is implemented.

[0015] Applying the technical solution of the present application, in the method for determining interbank fund flow, first, obtain the current interbank fund flow characteristic data, which is divided into time series characteristic data, residual sequence characteristic data and characteristic value; then, input the current interbank fund flow characteristic data into the corresponding interbank fund flow determination model to obtain the corresponding model output result, the interbank fund flow determination model includes the first interbank fund flow determination model, the second interbank fund flow determination model and the third interbank fund flow determination model, the first interbank fund flow determination model is a convergence model obtained by inputting the sample time series characteristic data into the ARIMA model for training, the second interbank fund flow determination model is a convergence model obtained by inputting the sample residual sequence characteristic data into the LSTM model for training, and the third interbank fund flow determination model is a convergence model obtained by inputting the sample characteristic value into the LGB model for training; finally, perform weighted combination on the output results of multiple models to obtain the target interbank fund flow. According to the fluctuation characteristics of interbank cash flow, the present application combines the traditional machine learning algorithm and the deep neural network algorithm, gives full play to the advantages of each algorithm, and constructs multiple interbank fund flow determination models, thereby improving the prediction accuracy. Substitute the historical prediction results and historical actual values ​​of each inter-bank fund flow determination model into the univariate linear regression model training to determine the weight coefficient of each inter-bank fund flow determination model. Not only is the weight coefficient automatically adjusted, but the adjustment of the weight coefficient is also made more reasonable and accurate. Thereby, according to the accuracy of the prediction results of each inter-bank fund flow determination model, the proportion of the prediction results in the final prediction results is automatically adjusted, further improving the accuracy of the prediction. The present application solves the problem that the prior art does not fully consider the fluctuation characteristics of inter-bank fund flows and directly discards the residual data, resulting in low prediction accuracy of inter-bank fund flows. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A hardware structure block diagram of a mobile terminal for executing a method for determining an inter-bank fund flow provided in an embodiment of the present application is shown;

[0017] Figure 2 A schematic diagram of a flow chart of a method for determining an inter-bank fund flow provided according to an embodiment of the present application is shown;

[0018] Figure 3 A schematic flow chart of an optional method for determining inter-bank fund flow provided according to an embodiment of the present application is shown;

[0019] Figure 4 A training flow chart of an inter-bank funds flow determination model provided according to an embodiment of the present application is shown;

[0020] Figure 5 A schematic diagram of the structure of an inter-bank funds flow management system provided according to an embodiment of the present application is shown;

[0021] Figure 6 A structural block diagram of a device for determining an inter-bank fund flow provided according to an embodiment of the present application is shown.

[0022] The above drawings include the following reference numerals:

[0023] 102, processor; 104, memory; 106, transmission device; 108, input and output devices. DETAILED DESCRIPTION

[0024] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0025] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0026] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0027] As introduced in the background technology, the inter-bank payment fund flow prediction method in the prior art fails to make full use of relevant external data and indicators in the process of model construction. These external data and indicators may include market changes, economic indicators, etc., which can provide a more comprehensive information basis for the prediction model, but the prior art fails to effectively integrate these data, resulting in a decrease in the comprehensiveness and accuracy of the prediction results. In order to solve the problem that the prior art does not fully consider the fluctuation characteristics of inter-bank fund flows and directly discards residual data, resulting in low prediction accuracy of inter-bank fund flows, the embodiments of the present application provide a method for determining an inter-bank fund flow, a determination device, a computer-readable storage medium and a computer program product.

[0028] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0029] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 1 is a hardware structure block diagram of a mobile terminal for executing a method for determining inter-bank fund flow according to an embodiment of the present invention. Figure 1 As shown, the mobile terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data, wherein the mobile terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the mobile terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0030] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the method for determining the inter-bank fund flow in the embodiment of the present invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, the above method is implemented. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the above network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof. The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0031] In this embodiment, a method for determining inter-bank fund flows running on a mobile terminal, a computer terminal or a similar computing device is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0032] Figure 2 is a flow chart of a method for determining inter-bank fund flows according to an embodiment of the present application. Figure 2 As shown, the method comprises the following steps:

[0033] Step S201, obtaining current inter-bank fund flow characteristic data, the current inter-bank fund flow characteristic data is divided into time series characteristic data, residual sequence characteristic data and characteristic value.

[0034] Specifically, the current inter-bank transaction details data and macroeconomic data are collected in real time, and the inter-bank transaction details data and macroeconomic data are pre-processed to obtain the current inter-bank fund flow characteristic data.

[0035] Step S202, input the above-mentioned current inter-bank capital flow feature data into the corresponding inter-bank capital flow determination model to obtain the corresponding model output result, the above-mentioned inter-bank capital flow determination model includes a first inter-bank capital flow determination model, a second inter-bank capital flow determination model and a third inter-bank capital flow determination model, the above-mentioned first inter-bank capital flow determination model is a convergence model obtained by inputting the sample time series feature data into the ARIMA model for training, the above-mentioned second inter-bank capital flow determination model is a convergence model obtained by inputting the sample residual sequence feature data into the LSTM model for training, and the above-mentioned third inter-bank capital flow determination model is a convergence model obtained by inputting the sample feature value into the LGB model for training.

[0036] Specifically, after obtaining the current inter-bank funds flow characteristic data, the time series characteristic data is input into the first inter-bank funds flow determination model, the residual sequence characteristic data is input into the second inter-bank funds flow determination model and the characteristic value is input into the third inter-bank funds flow determination model to obtain the model output results of each model. The function and effect of this method is that it can comprehensively utilize the advantages of different models to improve the prediction accuracy and stability of inter-bank funds flow.

[0037] Step S203, performing weighted combination on the output results of multiple models to obtain the target inter-bank capital flow.

[0038] Specifically, since each model has its own unique characteristics and scope of application, there may be correlations between different models. The weighted combination can make up for the shortcomings of a single model, improve the overall forecasting effect, and further improve the accuracy of the forecast. Therefore, the use of multi-model weighted combination method can effectively improve the forecast accuracy of inter-bank fund flows, which plays an important role in risk control and fund management.

[0039] In addition, the whole process processing method is as follows Figure 3 As shown, the original sample, that is, the set of all historical data, is obtained, the original sample is preprocessed to extract the features with high importance to construct a training sample, and multiple preset training models are trained based on the training samples to obtain the above-mentioned inter-bank funds flow determination model, including the first inter-bank funds flow determination model, the second inter-bank funds flow determination model and the third inter-bank funds flow determination model, and the weighted number method is used to combine multiple trained prediction models to obtain the final prediction result, and finally the trained and combined prediction model can be applied to the actual inter-bank funds flow data for actual prediction.

[0040] In this embodiment, in the determination method of interbank fund flow, first, obtain the current interbank fund flow characteristic data, which is divided into time series characteristic data, residual sequence characteristic data and characteristic value; then, input the current interbank fund flow characteristic data into the corresponding interbank fund flow determination model to obtain the corresponding model output result, the interbank fund flow determination model includes the first interbank fund flow determination model, the second interbank fund flow determination model and the third interbank fund flow determination model, the first interbank fund flow determination model is a convergence model obtained by inputting the sample time series characteristic data into the ARIMA model for training, the second interbank fund flow determination model is a convergence model obtained by inputting the sample residual sequence characteristic data into the LSTM model for training, and the third interbank fund flow determination model is a convergence model obtained by inputting the sample characteristic value into the LGB model for training; finally, perform weighted combination on the output results of multiple models to obtain the target interbank fund flow. According to the fluctuation characteristics of interbank cash flow, this application combines the traditional machine learning algorithm and the deep neural network algorithm, gives full play to the advantages of each algorithm, and constructs multiple interbank fund flow determination models, thereby improving the prediction accuracy. Substitute the historical prediction results and historical actual values ​​of each inter-bank fund flow determination model into the univariate linear regression model training to determine the weight coefficient of each inter-bank fund flow determination model. Not only is the weight coefficient automatically adjusted, but the adjustment of the weight coefficient is also made more reasonable and accurate. Thereby, according to the accuracy of the prediction results of each inter-bank fund flow determination model, the proportion of the prediction results in the final prediction results is automatically adjusted, further improving the accuracy of the prediction. The present application solves the problem that the prior art does not fully consider the fluctuation characteristics of inter-bank fund flows and directly discards the residual data, resulting in low prediction accuracy of inter-bank fund flows.

[0041] In order to enable those skilled in the art to more clearly understand the technical solution of the present application, the implementation process of the method for determining the inter-bank fund flow of the present application will be described in detail below in conjunction with specific embodiments.

[0042] In order to improve the prediction accuracy, in an optional implementation manner, before the above step S202, the method further includes:

[0043] Step S301, obtaining multiple groups of historical inter-bank fund flow characteristic data, the historical inter-bank fund flow characteristic data including the sample time series characteristic data, sample residual sequence characteristic data and sample characteristic value, the sample time series characteristic data including periodic sequence and trend sequence;

[0044] Step S302, a training step, inputting the sample time series feature data, the sample residual sequence feature data and the sample feature value in each group of the historical inter-bank fund flow feature data into a corresponding predetermined training model for iterative training to obtain a corresponding training output result, wherein the predetermined training model includes the ARIMA model, the LSTM model and the LGB model;

[0045] Step S303, a testing step, testing all the above training output results by a testing method to obtain a test result, wherein the above testing method at least includes residual white noise test and root mean square error;

[0046] Step S304, when the above-mentioned test results do not meet the model accuracy requirements, repeat the above-mentioned training steps and the above-mentioned test steps in sequence at least once until the above-mentioned test results meet the above-mentioned model accuracy requirements or the number of training times reaches the maximum number of iterations, and obtain the above-mentioned first inter-bank funds flow determination model, the above-mentioned second inter-bank funds flow determination model and the above-mentioned third inter-bank funds flow determination model respectively.

[0047] In the above embodiment, Figure 4A flow chart for training a model for determining interbank capital flow provided in an embodiment of the present application combines a variety of prediction models to give full play to the advantages of each model and make up for the shortcomings, thereby improving the accuracy of prediction. The ARIMA model is suitable for predicting interbank capital flows for the periodicity and trend of time series, and the LSTM model is suitable for predicting irregular time series, so the LSTM model is used to predict interbank capital flows for irregular residuals. The LGB model can calculate the feature importance of each eigenvalue through an algorithm, thereby automatically selecting eigenvalues ​​that contribute greatly to the prediction results, and is suitable for predicting data with many eigenvalues. Therefore, the LGB model is used to train macroeconomic data and predict interbank capital flows. Therefore, multiple groups of historical interbank capital flow feature data are first obtained, including sample time series feature data, sample residual sequence feature data, and sample eigenvalues, wherein the sample time series feature data includes a periodic sequence and a trend sequence. The sample time series feature data in each group of historical interbank capital flow feature data is input into the ARIMA model to be trained, the sample residual sequence feature data is input into the LSTM model to be trained, and the LGB model to be trained with the sample eigenvalue is iteratively trained. The training output results of the model are tested by residual white noise test, calculation of root mean square error, etc. If the test result does not meet the model accuracy requirements, the training and test steps are repeated until the test result meets the requirements or the number of iterations reaches the maximum number of iterations. Finally, the first inter-bank fund flow determination model, the second inter-bank fund flow determination model and the third inter-bank fund flow determination model are obtained.

[0048] In order to ensure the quality and integrity of the data, in an optional implementation manner, the above step S301 includes:

[0049] Step S3011, obtaining historical inter-bank fund flow data, the historical inter-bank fund flow data including inter-bank transaction details data and macroeconomic data within a historical period, the inter-bank transaction details data at least including transaction time, transaction amount and transaction direction, the macroeconomic data at least including macroeconomic factors, market environment factors, business cycle characteristics, internal bank plans, key regions and holiday characteristics;

[0050] Step S3012, performing a preprocessing operation on the macroeconomic data in the historical inter-bank fund flow data to obtain the sample characteristic value, wherein the preprocessing operation at least includes abnormal value processing, missing value filling, normalization processing and outlier processing;

[0051] Step S3013, using the STL decomposition method to extract the time series characteristics and residual series characteristics in the above-mentioned inter-bank transaction details data, and obtain the above-mentioned sample time series characteristic data and the above-mentioned residual series characteristic data respectively, and the above-mentioned time series characteristics include period information and trend information.

[0052] In the above embodiment, Figure 5 This is a schematic diagram of the structure of the inter-bank funds flow management system. Figure 5 The data collection module collects historical inter-bank transaction details and historical macroeconomic data. The historical inter-bank transaction details include transaction time, transaction amount, transaction direction and other information. The historical macroeconomic data includes macroeconomic factors (such as GDP, inflation rate, unemployment rate, etc.), market environment factors (such as stock market index, bond market index, foreign exchange market index, etc.), business cycle characteristics (such as the beginning and end of the year, the beginning and end of the month, tax period, etc.), internal plans, key regions and holiday characteristics (such as short holidays, Spring Festival, etc.). The collected historical inter-bank transaction details and historical macroeconomic data are saved to the AI ​​database through the interface server. Figure 4 The data preprocessing module in the above-mentioned macroeconomic data in the above-mentioned historical interbank fund flow data performs preprocessing operations to obtain the above-mentioned sample characteristic values. The above-mentioned preprocessing operations at least include outlier processing, missing value filling, normalization processing and outlier processing to ensure the quality and integrity of the data and eliminate the impact of the dimension on the prediction results. Interbank fund flow fluctuations are seasonal, cyclical and trendy. It is necessary to predict future interbank fund flows by analyzing the laws contained in the historical interbank fund flow fluctuations. Therefore, the STL decomposition method is used to decompose and extract the time series characteristics and residual series characteristics existing in the above-mentioned interbank transaction details data to obtain the above-mentioned sample time series characteristic data and the above-mentioned residual series characteristic data.

[0053] In order to avoid the limitations and deviations that may exist in a single model, in an optional implementation, the above step S203 includes:

[0054] Step S2031, inputting the output result of the first model and the output result of the second model into the combination scheme to obtain the first inter-bank capital flow, wherein the output result of the first model is the output result of the first inter-bank capital flow determination model, and the output result of the second model is the output result of the second inter-bank capital flow determination model;

[0055] Step S2032, input the above-mentioned first inter-bank funds flow and the output result of the third model into the above-mentioned combination scheme to obtain the above-mentioned target inter-bank funds flow, and the output result of the above-mentioned third model is the output result of the above-mentioned third inter-bank funds flow determination model.

[0056] In the above embodiment, when obtaining the final prediction result, the output results of each model are not simply averaged. The output results of the first model and the output results of the second model are used as input, and a combination scheme is used to obtain the first inter-bank funds flow. The output results of the first inter-bank funds flow and the third model are input into the combination scheme to obtain the output results of the determination model of the target inter-bank funds flow. By combining the output results of different models, the prediction results of multiple models can be comprehensively considered to improve the accuracy and stability of the inter-bank funds flow prediction. By combining the output results of different models, more comprehensive and integrated inter-bank funds flow prediction information can be obtained to help decision makers better formulate funds management strategies and risk control measures.

[0057] In order to improve the overall prediction effect of multiple models, in an optional implementation, before the above step S2031, the method further includes:

[0058] Step S401, determine the above-mentioned combination scheme, which is a scheme for adding the first product and the second product to obtain the inter-bank fund flow, the above-mentioned first product is the product of the first weight coefficient and the first prediction result, the above-mentioned first prediction result is the output result of the first model, the above-mentioned first model is any one of the above-mentioned inter-bank fund flow determination models, the above-mentioned first weight coefficient is the weight coefficient corresponding to the above-mentioned first model, the above-mentioned second product is the product of the second weight coefficient and the second prediction result, the above-mentioned second prediction result is the output result of the second model, the above-mentioned second model is any one of the above-mentioned inter-bank fund flow determination models except the above-mentioned first model, the above-mentioned second weight coefficient is the weight coefficient corresponding to the above-mentioned second model, and the sum of the above-mentioned first weight coefficient and the above-mentioned second weight coefficient is equal to 1.

[0059] In the above embodiment, before combining the output results of multiple trained inter-bank fund flow determination models by using the weighted number method, it is necessary to determine a combination scheme, that is, the combination scheme of the inter-bank fund flow prediction result (that is, the above target fund flow) is the inter-bank fund flow prediction result = λ1×the prediction result of the first model (that is, the model output result of the first model) + λ2×the prediction result of the second model (that is, the model output result of the second model). Among them, λ1 is the weight coefficient of the first model, λ2 is the weight coefficient of the second model, and the output results of multiple models are combined in pairs in sequence to obtain the final prediction result. Of course, the above combination scheme is an optional combination scheme provided by the present invention, and those skilled in the art can also adopt other more efficient combination schemes (for example, machine learning algorithms) or inefficient combination methods to achieve the purpose, for example, the target inter-bank fund flow = λ1×the prediction result of the first model (that is, the model output result of the first model) + λ2×the prediction result of the second model (that is, the model output result of the second model) + λ3×the prediction result of the third model (that is, the model output result of the third model), and the third model is the inter-bank fund flow determination model remaining except the above first model and the above second model. The use of a combination of solutions can reduce the risk of single model forecasts and avoid the possible limitations and biases of a single model, thereby improving the reliability of inter-bank cash flow forecasts.

[0060] In order to improve the accuracy of the weight coefficient, in an optional implementation manner, before the above step S401, the method further includes:

[0061] Step S501, inputting all training output results output by the first inter-bank funds flow determination model and the first sample inter-bank funds flow into a univariate linear regression model to obtain the weight coefficient corresponding to the first inter-bank funds flow determination model, wherein the first sample inter-bank funds flow is the sample inter-bank funds flow corresponding to the sample time series feature data;

[0062] Step S502, inputting all the above training output results and the second sample inter-bank fund flow output by the above second inter-bank fund flow determination model into the above univariate linear regression model to obtain the above weight coefficient corresponding to the above second inter-bank fund flow determination model, wherein the above second sample inter-bank fund flow is the sample inter-bank fund flow corresponding to the above sample residual sequence feature data;

[0063] Step S503, input all the above-mentioned training output results and the third sample inter-bank funds flow output by the above-mentioned third inter-bank funds flow determination model into the above-mentioned univariate linear regression model, and obtain the above-mentioned weight coefficient corresponding to the above-mentioned third inter-bank funds flow determination model, and the above-mentioned third sample inter-bank funds flow is the sample inter-bank funds flow corresponding to the above-mentioned sample characteristic value.

[0064] In the above embodiment, the weight coefficients in the combination scheme are not directly set weight coefficients, in order to reduce the subjectivity of direct setting, but the historical prediction results of each model (i.e. the above training output results) and the historical actual values ​​(i.e. the above sample inter-bank fund flow) are substituted into the univariate linear regression model training to determine the weight coefficients of each model. Because the contribution of each model to accuracy varies and changes over time, if it is simply averaged, the accuracy of the prediction results will be reduced. Therefore, the historical prediction results and historical actual values ​​of each model are substituted into the univariate linear regression model training, and the weight coefficients are obtained in this way. According to the accuracy of the prediction results of each model, the degree of influence on the final prediction structure can be automatically adjusted. This method is more scientific and reasonable.

[0065] In order to improve the accuracy of the target capital flow, in an optional implementation manner, after the above step S401, the method further includes:

[0066] Step S601, obtaining a first accuracy and a second accuracy, wherein the first accuracy is the accuracy of an output result of the first model, and the second accuracy is the accuracy of an output result of the second model;

[0067] Step S602, when the first accuracy is greater than the second accuracy, increasing the first weight coefficient and decreasing the second weight coefficient;

[0068] Step S603: when the first accuracy is less than the second accuracy, reduce the first weight coefficient and increase the second weight coefficient.

[0069] In the above embodiment, the inter-bank funds flow management system can automatically adjust the weight coefficients λ1 and λ2 of the two models in the combination scheme according to the historical prediction results and historical actual values ​​of each model. If the historical prediction results of the first model are closer to the historical actual values, the value of λ1 (i.e., the above-mentioned first weight coefficient) is appropriately increased; if the historical prediction results of the second model are closer to the historical actual values, the value of λ2 (i.e., the above-mentioned second weight coefficient) is appropriately increased, but the sum of the first weight coefficient λ1 and the second weight coefficient λ2 is always kept at 1. Not only is the weight coefficient automatically adjusted, but the adjustment of the weight coefficient is also made more reasonable and accurate. Thereby, the system automatically adjusts the proportion of each model's prediction results in the final prediction results according to the accuracy of the model's prediction results, further improving the accuracy of the prediction.

[0070] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0071] The embodiment of the present application also provides a device for determining an inter-bank fund flow. It should be noted that the device for determining an inter-bank fund flow in the embodiment of the present application can be used to execute the method for determining an inter-bank fund flow provided in the embodiment of the present application. The device is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions thereof will not be repeated. As used below, the term "module" may be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0072] The following is an introduction to the device for determining the inter-bank fund flow provided in the embodiment of the present application.

[0073] Figure 6 is a structural block diagram of a device for determining inter-bank fund flow according to an embodiment of the present application. Figure 6 As shown, the device comprises:

[0074] The first acquisition unit 10 is used to acquire current inter-bank fund flow characteristic data, wherein the current inter-bank fund flow characteristic data is divided into time series characteristic data, residual series characteristic data and characteristic values.

[0075] Specifically, the current inter-bank transaction details data and macroeconomic data are collected in real time, and the inter-bank transaction details data and macroeconomic data are pre-processed to obtain the current inter-bank fund flow characteristic data.

[0076] The input-output unit 20 is used to input the above-mentioned current inter-bank capital flow feature data into the corresponding inter-bank capital flow determination model to obtain the corresponding model output result. The above-mentioned inter-bank capital flow determination model includes a first inter-bank capital flow determination model, a second inter-bank capital flow determination model and a third inter-bank capital flow determination model. The above-mentioned first inter-bank capital flow determination model is a convergence model obtained by inputting the sample time series feature data into the ARIMA model for training, the above-mentioned second inter-bank capital flow determination model is a convergence model obtained by inputting the sample residual sequence feature data into the LSTM model for training, and the above-mentioned third inter-bank capital flow determination model is a convergence model obtained by inputting the sample feature value into the LGB model for training.

[0077] Specifically, after obtaining the current inter-bank funds flow characteristic data, the time series characteristic data is input into the first inter-bank funds flow determination model, the residual sequence characteristic data is input into the second inter-bank funds flow determination model and the characteristic value is input into the third inter-bank funds flow determination model to obtain the model output results of each model. The function and effect of this method is that it can comprehensively utilize the advantages of different models to improve the prediction accuracy and stability of inter-bank funds flow.

[0078] The combination unit 30 is used to perform weighted combination on the output results of the multiple models to obtain the target inter-bank capital flow.

[0079] Specifically, since each model has its own unique characteristics and scope of application, there may be correlations between different models. The weighted combination can make up for the shortcomings of a single model, improve the overall forecasting effect, and further improve the accuracy of the forecast. Therefore, the use of multi-model weighted combination method can effectively improve the forecast accuracy of inter-bank fund flows, which plays an important role in risk control and fund management.

[0080] In addition, the whole process processing method is as follows Figure 3 As shown, the original sample, that is, the set of all historical data, is obtained, the original sample is preprocessed to extract the features with high importance to construct a training sample, and multiple preset training models are trained based on the training samples to obtain the above-mentioned inter-bank funds flow determination model, including the first inter-bank funds flow determination model, the second inter-bank funds flow determination model and the third inter-bank funds flow determination model, and the weighted number method is used to combine multiple trained prediction models to obtain the final prediction result, and finally the trained and combined prediction model can be applied to the actual inter-bank funds flow data for actual prediction.

[0081] In this embodiment, the first acquisition unit is used to obtain the current interbank fund flow feature data, and the current interbank fund flow feature data is divided into time series feature data, residual sequence feature data and feature value; the input and output unit is used to input the current interbank fund flow feature data into the corresponding interbank fund flow determination model to obtain the corresponding model output result, and the interbank fund flow determination model includes the first interbank fund flow determination model, the second interbank fund flow determination model and the third interbank fund flow determination model. The first interbank fund flow determination model is a convergence model obtained by inputting the sample time series feature data into the ARIMA model for training, the second interbank fund flow determination model is a convergence model obtained by inputting the sample residual sequence feature data into the LSTM model for training, and the third interbank fund flow determination model is a convergence model obtained by inputting the sample feature value into the LGB model for training; the combination unit is used to weightedly combine the output results of multiple models to obtain the target interbank fund flow. According to the fluctuation characteristics of interbank cash flow, this application combines traditional machine learning algorithms and deep neural network algorithms, gives full play to the advantages of each algorithm, and constructs multiple interbank fund flow determination models, thereby improving the prediction accuracy. Substitute the historical prediction results and historical actual values ​​of each inter-bank fund flow determination model into the univariate linear regression model training to determine the weight coefficient of each inter-bank fund flow determination model. Not only is the weight coefficient automatically adjusted, but the adjustment of the weight coefficient is also made more reasonable and accurate. Thereby, according to the accuracy of the prediction results of each inter-bank fund flow determination model, the proportion of the prediction results in the final prediction results is automatically adjusted, further improving the accuracy of the prediction. The present application solves the problem that the prior art does not fully consider the fluctuation characteristics of inter-bank fund flows and directly discards the residual data, resulting in low prediction accuracy of inter-bank fund flows.

[0082] In order to improve the prediction accuracy, in an optional embodiment, the device further includes:

[0083] A second acquisition unit is used to acquire multiple sets of historical inter-bank fund flow characteristic data before inputting the current inter-bank fund flow characteristic data into the corresponding inter-bank fund flow determination model to obtain the corresponding model output result, wherein the historical inter-bank fund flow characteristic data includes the sample time series characteristic data, the sample residual sequence characteristic data and the sample characteristic value, and the sample time series characteristic data includes a periodic sequence and a trend sequence;

[0084] A training unit, used to execute a training step, input the sample time series feature data, the sample residual sequence feature data and the sample feature value in each group of the historical inter-bank fund flow feature data into a corresponding predetermined training model for iterative training, and obtain a corresponding training output result, wherein the predetermined training model includes the ARIMA model, the LSTM model and the LGB model;

[0085] A testing unit, used to execute the testing step, testing all the above training output results by a testing method to obtain a testing result, wherein the above testing method at least includes residual white noise testing and root mean square error;

[0086] A repeating unit is used to, when the above-mentioned test results do not meet the model accuracy requirements, repeatedly execute the above-mentioned training steps and the above-mentioned test steps in sequence at least once until the above-mentioned test results meet the above-mentioned model accuracy requirements or the number of training times reaches a maximum number of iterations, thereby respectively obtaining the above-mentioned first inter-bank funds flow determination model, the above-mentioned second inter-bank funds flow determination model and the above-mentioned third inter-bank funds flow determination model.

[0087] In the above embodiment, Figure 4 A flow chart for training a model for determining interbank capital flow provided in an embodiment of the present application combines a variety of prediction models to give full play to the advantages of each model and make up for the shortcomings, thereby improving the accuracy of prediction. The ARIMA model is suitable for predicting interbank capital flows for the periodicity and trend of time series, and the LSTM model is suitable for predicting irregular time series, so the LSTM model is used to predict interbank capital flows for irregular residuals. The LGB model can calculate the feature importance of each eigenvalue through an algorithm, thereby automatically selecting eigenvalues ​​that contribute greatly to the prediction results, and is suitable for predicting data with many eigenvalues. Therefore, the LGB model is used to train macroeconomic data and predict interbank capital flows. Therefore, multiple groups of historical interbank capital flow feature data are first obtained, including sample time series feature data, sample residual sequence feature data, and sample eigenvalues, wherein the sample time series feature data includes a periodic sequence and a trend sequence. The sample time series feature data in each group of historical interbank capital flow feature data is input into the ARIMA model to be trained, the sample residual sequence feature data is input into the LSTM model to be trained, and the LGB model to be trained with the sample eigenvalue is iteratively trained. The training output results of the model are tested by residual white noise test, calculation of root mean square error, etc. If the test result does not meet the model accuracy requirements, the training and test steps are repeated until the test result meets the requirements or the number of iterations reaches the maximum number of iterations. Finally, the first inter-bank fund flow determination model, the second inter-bank fund flow determination model and the third inter-bank fund flow determination model are obtained.

[0088] In order to ensure the quality and integrity of the data, in an optional implementation manner, the second acquisition unit includes:

[0089] An acquisition module is used to acquire historical inter-bank fund flow data, wherein the historical inter-bank fund flow data includes inter-bank transaction details data and macroeconomic data within a historical period, wherein the inter-bank transaction details data at least includes transaction time, transaction amount and transaction direction, and the macroeconomic data at least includes macroeconomic factors, market environment factors, business cycle characteristics, internal bank plans, key regions and holiday characteristics;

[0090] A preprocessing module performs a preprocessing operation on the macroeconomic data in the historical inter-bank fund flow data to obtain the sample characteristic value, wherein the preprocessing operation at least includes abnormal value processing, missing value filling, normalization processing and outlier processing;

[0091] The decomposition module uses the STL decomposition method to extract the time series characteristics and residual series characteristics in the above-mentioned inter-bank transaction details data, and obtains the above-mentioned sample time series characteristic data and the above-mentioned residual series characteristic data respectively. The above-mentioned time series characteristics include periodic information and trend information.

[0092] In the above embodiment, Figure 5 This is a schematic diagram of the structure of the inter-bank funds flow management system. Figure 5 The data collection module collects historical inter-bank transaction details and historical macroeconomic data. The historical inter-bank transaction details include transaction time, transaction amount, transaction direction and other information. The historical macroeconomic data includes macroeconomic factors (such as GDP, inflation rate, unemployment rate, etc.), market environment factors (such as stock market index, bond market index, foreign exchange market index, etc.), business cycle characteristics (such as the beginning and end of the year, the beginning and end of the month, tax period, etc.), internal plans, key regions and holiday characteristics (such as short holidays, Spring Festival, etc.). The collected historical inter-bank transaction details and historical macroeconomic data are saved to the AI ​​database through the interface server. Figure 4 The data preprocessing module in the above-mentioned macroeconomic data in the above-mentioned historical interbank fund flow data performs preprocessing operations to obtain the above-mentioned sample characteristic values. The above-mentioned preprocessing operations at least include outlier processing, missing value filling, normalization processing and outlier processing to ensure the quality and integrity of the data and eliminate the impact of the dimension on the prediction results. Interbank fund flow fluctuations are seasonal, cyclical and trendy. It is necessary to predict future interbank fund flows by analyzing the laws contained in the historical interbank fund flow fluctuations. Therefore, the STL decomposition method is used to decompose and extract the time series characteristics and residual series characteristics existing in the above-mentioned interbank transaction details data to obtain the above-mentioned sample time series characteristic data and the above-mentioned residual series characteristic data.

[0093] In order to avoid the limitations and deviations that may exist in a single model, in an optional implementation, the above-mentioned combined module includes:

[0094] A first input submodule inputs the output result of the first model and the output result of the second model into the combination scheme to obtain a first inter-bank fund flow, wherein the output result of the first model is the output result of the first inter-bank fund flow determination model, and the output result of the second model is the output result of the second inter-bank fund flow determination model;

[0095] The second input submodule inputs the above-mentioned first inter-bank funds flow and the output result of the third model into the above-mentioned combination scheme to obtain the above-mentioned target inter-bank funds flow. The output result of the above-mentioned third model is the output result of the above-mentioned third inter-bank funds flow determination model.

[0096] In the above embodiment, when obtaining the final prediction result, the output results of each model are not simply averaged. The output results of the first model and the output results of the second model are used as input, and a combination scheme is used to obtain the first inter-bank funds flow. The output results of the first inter-bank funds flow and the third model are input into the combination scheme to obtain the output results of the determination model of the target inter-bank funds flow. By combining the output results of different models, the prediction results of multiple models can be comprehensively considered to improve the accuracy and stability of the inter-bank funds flow prediction. By combining the output results of different models, more comprehensive and integrated inter-bank funds flow prediction information can be obtained to help decision makers better formulate funds management strategies and risk control measures.

[0097] In order to improve the overall prediction effect of multiple models, in an optional implementation manner, the device further includes:

[0098] A determination unit is used to determine the above-mentioned combination scheme before inputting the output results of the first model and the output results of the second model into the combination scheme to obtain the first inter-bank funds flow. The above-mentioned combination scheme is a scheme for adding the first product and the second product to obtain the inter-bank funds flow. The above-mentioned first product is the product between the first weight coefficient and the first prediction result, the above-mentioned first prediction result is the output result of the first model, the above-mentioned first model is any one of the above-mentioned inter-bank funds flow determination models, the above-mentioned first weight coefficient is the weight coefficient corresponding to the above-mentioned first model, the above-mentioned second product is the product between the second weight coefficient and the second prediction result, the above-mentioned second prediction result is the output result of the second model, the above-mentioned second model is any one of the above-mentioned inter-bank funds flow determination models other than the above-mentioned first model, the above-mentioned second weight coefficient is the weight coefficient corresponding to the above-mentioned second model, and the sum of the above-mentioned first weight coefficient and the above-mentioned second weight coefficient is equal to 1.

[0099] In the above embodiment, before combining the output results of multiple trained inter-bank fund flow determination models by using the weighted number method, it is necessary to determine a combination scheme, that is, the combination scheme of the inter-bank fund flow prediction result (that is, the above target fund flow) is the inter-bank fund flow prediction result = λ1×the prediction result of the first model (that is, the model output result of the first model) + λ2×the prediction result of the second model (that is, the model output result of the second model). Among them, λ1 is the weight coefficient of the first model, λ2 is the weight coefficient of the second model, and the output results of multiple models are combined in pairs in sequence to obtain the final prediction result. Of course, the above combination scheme is an optional combination scheme provided by the present invention, and those skilled in the art can also adopt other more efficient combination schemes (for example, machine learning algorithms) or inefficient combination methods to achieve the purpose, for example, the target inter-bank fund flow = λ1×the prediction result of the first model (that is, the model output result of the first model) + λ2×the prediction result of the second model (that is, the model output result of the second model) + λ3×the prediction result of the third model (that is, the model output result of the third model), and the third model is the inter-bank fund flow determination model remaining except the above first model and the above second model. The use of a combination of solutions can reduce the risk of single model forecasts and avoid the possible limitations and biases of a single model, thereby improving the reliability of inter-bank cash flow forecasts.

[0100] In order to improve the accuracy of the weight coefficient, in an optional implementation manner, the device further includes:

[0101] A third input unit is used to input all training output results output by the first inter-bank fund flow determination model and the first sample inter-bank fund flow into a univariate linear regression model before determining the combination scheme, so as to obtain the weight coefficient corresponding to the first inter-bank fund flow determination model, wherein the first sample inter-bank fund flow is the sample inter-bank fund flow corresponding to the sample time series feature data;

[0102] A fourth input unit is used to input all the above training output results and the second sample inter-bank fund flow output by the above second inter-bank fund flow determination model into the above univariate linear regression model to obtain the above weight coefficient corresponding to the above second inter-bank fund flow determination model, and the above second sample inter-bank fund flow is the sample inter-bank fund flow corresponding to the above sample residual sequence feature data;

[0103] The fifth input unit is used to input all the above-mentioned training output results and the third sample inter-bank funds flow output by the above-mentioned third inter-bank funds flow determination model into the above-mentioned univariate linear regression model to obtain the above-mentioned weight coefficient corresponding to the above-mentioned third inter-bank funds flow determination model, and the above-mentioned third sample inter-bank funds flow is the sample inter-bank funds flow corresponding to the above-mentioned sample characteristic value.

[0104] In the above embodiment, the weight coefficients in the combination scheme are not directly set weight coefficients, but the historical prediction results of each model (i.e. the above training output results) and the historical actual values ​​(i.e. the above sample inter-bank fund flows) are substituted into the univariate linear regression model training to determine the weight coefficients of each model. Because the contribution of each model to accuracy varies and changes over time, if it is simply averaged, the accuracy of the prediction results will be reduced. Therefore, the historical prediction results and historical actual values ​​of each model are substituted into the univariate linear regression model training. In this way, the weight coefficients are obtained. According to the accuracy of the prediction results of each model, the degree of influence on the final prediction structure can be automatically adjusted. This method is more scientific and reasonable. Of course, the technicians can also make appropriate adjustments after the model outputs the weight coefficients.

[0105] In order to improve the accuracy of the target capital flow, in an optional implementation manner, the device further includes:

[0106] A third acquisition unit is used to acquire a first accuracy and a second accuracy after determining the combination scheme, wherein the first accuracy is the accuracy of the output result of the first model, and the second accuracy is the accuracy of the output result of the second model;

[0107] A first adjustment unit, configured to increase the first weight coefficient and decrease the second weight coefficient when the first accuracy is greater than the second accuracy;

[0108] The second adjustment unit is used to reduce the first weight coefficient and increase the second weight coefficient when the first accuracy is less than the second accuracy.

[0109] In the above embodiment, the inter-bank funds flow management system can automatically adjust the weight coefficients λ1 and λ2 of the two models in the combination scheme according to the historical prediction results and historical actual values ​​of each model. If the historical prediction results of the first model are closer to the historical actual values, the value of λ1 (i.e., the above-mentioned first weight coefficient) is appropriately increased; if the historical prediction results of the second model are closer to the historical actual values, the value of λ2 (i.e., the above-mentioned second weight coefficient) is appropriately increased, but the sum of the first weight coefficient λ1 and the second weight coefficient λ2 is always kept at 1. Not only is the weight coefficient automatically adjusted, but the adjustment of the weight coefficient is also made more reasonable and accurate. Thereby, the system automatically adjusts the proportion of each model's prediction results in the final prediction results according to the accuracy of the model's prediction results, further improving the accuracy of the prediction.

[0110] The above-mentioned device for determining inter-bank fund flow includes a processor and a memory, and the above-mentioned first acquisition unit, input-output unit and combination unit are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize corresponding functions. The above-mentioned modules are all located in the same processor; or, the above-mentioned modules are located in different processors in any combination.

[0111] The processor includes a kernel, which calls the corresponding program unit from the memory. One or more kernels can be set, and the kernel parameters can be adjusted to solve the problem that the existing technology does not fully consider the fluctuation characteristics of inter-bank fund flows and directly discards the residual data, resulting in low prediction accuracy of inter-bank fund flows.

[0112] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0113] An embodiment of the present invention provides a computer-readable storage medium, which includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method for determining the inter-bank fund flow.

[0114] Specifically, the methods for determining inter-bank fund flows include:

[0115] Step S201, obtaining current inter-bank fund flow characteristic data, the current inter-bank fund flow characteristic data is divided into time series characteristic data, residual sequence characteristic data and characteristic value;

[0116] Step S202, input the above-mentioned current inter-bank fund flow feature data into the corresponding inter-bank fund flow determination model to obtain the corresponding model output result, the above-mentioned inter-bank fund flow determination model includes a first inter-bank fund flow determination model, a second inter-bank fund flow determination model and a third inter-bank fund flow determination model, the above-mentioned first inter-bank fund flow determination model is a convergence model obtained by inputting the sample time series feature data into the ARIMA model for training, the above-mentioned second inter-bank fund flow determination model is a convergence model obtained by inputting the sample residual sequence feature data into the LSTM model for training, and the above-mentioned third inter-bank fund flow determination model is a convergence model obtained by inputting the sample feature value into the LGB model for training;

[0117] Step S203, performing weighted combination on the output results of multiple models to obtain the target inter-bank capital flow.

[0118] An embodiment of the present invention provides a processor, which is used to run a program, wherein the method for determining the inter-bank fund flow is executed when the program is run.

[0119] Specifically, the methods for determining inter-bank fund flows include:

[0120] Step S201, obtaining current inter-bank fund flow characteristic data, the current inter-bank fund flow characteristic data is divided into time series characteristic data, residual sequence characteristic data and characteristic value;

[0121] Step S202, input the above-mentioned current inter-bank fund flow feature data into the corresponding inter-bank fund flow determination model to obtain the corresponding model output result, the above-mentioned inter-bank fund flow determination model includes a first inter-bank fund flow determination model, a second inter-bank fund flow determination model and a third inter-bank fund flow determination model, the above-mentioned first inter-bank fund flow determination model is a convergence model obtained by inputting the sample time series feature data into the ARIMA model for training, the above-mentioned second inter-bank fund flow determination model is a convergence model obtained by inputting the sample residual sequence feature data into the LSTM model for training, and the above-mentioned third inter-bank fund flow determination model is a convergence model obtained by inputting the sample feature value into the LGB model for training;

[0122] Step S203, performing weighted combination on the output results of multiple models to obtain the target inter-bank capital flow.

[0123] An embodiment of the present invention provides an inter-bank fund flow management system, which includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, at least the following steps are implemented:

[0124] Step S201, obtaining current inter-bank fund flow characteristic data, the current inter-bank fund flow characteristic data is divided into time series characteristic data, residual sequence characteristic data and characteristic value;

[0125] Step S202, input the above-mentioned current inter-bank fund flow feature data into the corresponding inter-bank fund flow determination model to obtain the corresponding model output result, the above-mentioned inter-bank fund flow determination model includes a first inter-bank fund flow determination model, a second inter-bank fund flow determination model and a third inter-bank fund flow determination model, the above-mentioned first inter-bank fund flow determination model is a convergence model obtained by inputting the sample time series feature data into the ARIMA model for training, the above-mentioned second inter-bank fund flow determination model is a convergence model obtained by inputting the sample residual sequence feature data into the LSTM model for training, and the above-mentioned third inter-bank fund flow determination model is a convergence model obtained by inputting the sample feature value into the LGB model for training;

[0126] Step S203, performing weighted combination on the output results of multiple models to obtain the target inter-bank capital flow.

[0127] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program for initializing at least the following method steps:

[0128] Step S201, obtaining current inter-bank fund flow characteristic data, the current inter-bank fund flow characteristic data is divided into time series characteristic data, residual sequence characteristic data and characteristic value;

[0129] Step S202, input the above-mentioned current inter-bank fund flow feature data into the corresponding inter-bank fund flow determination model to obtain the corresponding model output result, the above-mentioned inter-bank fund flow determination model includes a first inter-bank fund flow determination model, a second inter-bank fund flow determination model and a third inter-bank fund flow determination model, the above-mentioned first inter-bank fund flow determination model is a convergence model obtained by inputting the sample time series feature data into the ARIMA model for training, the above-mentioned second inter-bank fund flow determination model is a convergence model obtained by inputting the sample residual sequence feature data into the LSTM model for training, and the above-mentioned third inter-bank fund flow determination model is a convergence model obtained by inputting the sample feature value into the LGB model for training;

[0130] Step S203, performing weighted combination on the output results of multiple models to obtain the target inter-bank capital flow.

[0131] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0132] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0133] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0134] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0135] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0136] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0137] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0138] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0139] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0140] From the above description, it can be seen that the above embodiments of the present application achieve the following technical effects:

[0141] 1) The method for determining the interbank fund flow of the present application, first, obtain the current interbank fund flow characteristic data, the current interbank fund flow characteristic data is divided into time series characteristic data, residual sequence characteristic data and characteristic value; then, input the current interbank fund flow characteristic data into the corresponding interbank fund flow determination model to obtain the corresponding model output result, the interbank fund flow determination model includes the first interbank fund flow determination model, the second interbank fund flow determination model and the third interbank fund flow determination model, the first interbank fund flow determination model is a convergence model obtained by inputting the sample time series characteristic data into the ARIMA model for training, the second interbank fund flow determination model is a convergence model obtained by inputting the sample residual sequence characteristic data into the LSTM model for training, and the third interbank fund flow determination model is a convergence model obtained by inputting the sample characteristic value into the LGB model for training; finally, the output results of multiple models are weighted and combined to obtain the target interbank fund flow. According to the fluctuation characteristics of interbank cash flow, the present application combines the traditional machine learning algorithm and the deep neural network algorithm, gives full play to the advantages of each algorithm, and constructs multiple interbank fund flow determination models, thereby improving the prediction accuracy. Substitute the historical prediction results and historical actual values ​​of each inter-bank fund flow determination model into the univariate linear regression model training to determine the weight coefficient of each inter-bank fund flow determination model. Not only is the weight coefficient automatically adjusted, but the adjustment of the weight coefficient is also made more reasonable and accurate. Thereby, according to the accuracy of the prediction results of each inter-bank fund flow determination model, the proportion of the prediction results in the final prediction results is automatically adjusted, further improving the accuracy of the prediction. The present application solves the problem that the prior art does not fully consider the fluctuation characteristics of inter-bank fund flows and directly discards the residual data, resulting in low prediction accuracy of inter-bank fund flows.

[0142] 2) The inter-bank fund flow determination device of the present application comprises a first acquisition unit, which is used to acquire current inter-bank fund flow characteristic data, wherein the current inter-bank fund flow characteristic data is divided into time series characteristic data, residual sequence characteristic data and characteristic values; an input-output unit, which is used to input the current inter-bank fund flow characteristic data into a corresponding inter-bank fund flow determination model to obtain a corresponding model output result, wherein the inter-bank fund flow determination model comprises a first inter-bank fund flow determination model, a second inter-bank fund flow determination model and a third inter-bank fund flow determination model, wherein the first inter-bank fund flow determination model is a convergence model obtained by inputting sample time series characteristic data into an ARIMA model for training, the second inter-bank fund flow determination model is a convergence model obtained by inputting sample residual sequence characteristic data into an LSTM model for training, and the third inter-bank fund flow determination model is a convergence model obtained by inputting sample characteristic values ​​into an LGB model for training; and a combination unit, which is used to perform weighted combination of the output results of the above-mentioned multiple models to obtain a target inter-bank fund flow. This application combines traditional machine learning algorithms and deep neural network algorithms based on the characteristics of inter-bank cash flow fluctuations, takes advantage of each algorithm, and constructs multiple inter-bank fund flow determination models, thereby improving the accuracy of predictions. The historical prediction results and historical actual values ​​of each inter-bank fund flow determination model are substituted into the univariate linear regression model training to determine the weight coefficient of each inter-bank fund flow determination model. Not only is the weight coefficient automatically adjusted, but the adjustment of the weight coefficient is also made more reasonable and accurate. Thereby, according to the accuracy of the prediction results of each inter-bank fund flow determination model, its proportion in the final prediction result is automatically adjusted, further improving the accuracy of the prediction. This application solves the problem that the prior art does not fully consider the characteristics of inter-bank fund flow fluctuations and directly discards the residual data, resulting in low prediction accuracy of inter-bank fund flows.

[0143] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for determining inter-bank fund flow, characterized in that: include: Acquire current inter-bank fund flow characteristic data, wherein the current inter-bank fund flow characteristic data is divided into time series characteristic data, residual sequence characteristic data and characteristic value; Input the current inter-bank fund flow feature data into the corresponding inter-bank fund flow determination model to obtain the corresponding model output result, wherein the inter-bank fund flow determination model includes a first inter-bank fund flow determination model, a second inter-bank fund flow determination model and a third inter-bank fund flow determination model, wherein the first inter-bank fund flow determination model is a convergence model obtained by inputting the sample time series feature data into the ARIMA model for training, the second inter-bank fund flow determination model is a convergence model obtained by inputting the sample residual sequence feature data into the LSTM model for training, and the third inter-bank fund flow determination model is a convergence model obtained by inputting the sample feature value into the LGB model for training; The output results of multiple models are weighted and combined to obtain the target inter-bank funds flow.

2. The method according to claim 1, characterized in that Before inputting the current inter-bank fund flow characteristic data into the corresponding inter-bank fund flow determination model to obtain the corresponding model output result, the method further includes: Acquire multiple groups of historical inter-bank fund flow characteristic data, wherein the historical inter-bank fund flow characteristic data include the sample time series characteristic data, the sample residual sequence characteristic data and the sample characteristic value, and the sample time series characteristic data include a periodic sequence and a trend sequence; A training step of inputting the sample time series feature data, the sample residual sequence feature data and the sample feature value in each group of the historical inter-bank fund flow feature data into a corresponding predetermined training model for iterative training to obtain a corresponding training output result, wherein the predetermined training model includes the ARIMA model, the LSTM model and the LGB model; A testing step, testing all the training output results by a testing method to obtain a test result, wherein the testing method at least includes a residual white noise test and a root mean square error; When the verification result does not meet the model accuracy requirement, the training step and the verification step are repeated in sequence at least once until the verification result meets the model accuracy requirement or the number of training iterations reaches a maximum number of iterations, and the first inter-bank funds flow determination model, the second inter-bank funds flow determination model and the third inter-bank funds flow determination model are obtained respectively.

3. The method according to claim 2, characterized in that Obtain multiple sets of historical inter-bank fund flow feature data, including: Obtaining historical inter-bank fund flow data, the historical inter-bank fund flow data including inter-bank transaction details data and macroeconomic data within a historical period, the inter-bank transaction details data including at least transaction time, transaction amount and transaction direction, the macroeconomic data including at least macroeconomic factors, market environment factors, business cycle characteristics, internal bank plans, key regions and holiday characteristics; Performing a preprocessing operation on the macroeconomic data in the historical inter-bank fund flow data to obtain the sample characteristic value, wherein the preprocessing operation at least includes abnormal value processing, missing value filling, normalization processing and outlier processing; The STL decomposition method is used to extract the time series features and residual series features in the inter-bank transaction details data, and the sample time series feature data and the residual series feature data are obtained respectively. The time series features include period information and trend information.

4. The method according to claim 1, characterized in that: The output results of multiple models are weighted and combined to obtain the target inter-bank fund flow, including: Inputting the output result of the first model and the output result of the second model into the combination scheme to obtain a first inter-bank capital flow, wherein the output result of the first model is the output result of the first inter-bank capital flow determination model, and the output result of the second model is the output result of the second inter-bank capital flow determination model; The first interbank funds flow and the output result of the third model are input into the combination scheme to obtain the target interbank funds flow, and the output result of the third model is the output result of the third interbank funds flow determination model.

5. The method according to claim 4, characterized in that Before inputting the output results of the first model and the second model into the combination scheme to obtain the first inter-bank capital flow, the method further includes: Determine the combination scheme, which is a scheme for obtaining an inter-bank fund flow by adding the first product and the second product, wherein the first product is the product of a first weight coefficient and a first prediction result, the first prediction result is the output result of a first model, the first model is any one of the inter-bank fund flow determination models, the first weight coefficient is the weight coefficient corresponding to the first model, the second product is the product of a second weight coefficient and a second prediction result, the second prediction result is the output result of a second model, the second model is any one of the inter-bank fund flow determination models other than the first model, the second weight coefficient is the weight coefficient corresponding to the second model, and the sum of the first weight coefficient and the second weight coefficient is equal to 1.

6. The method according to claim 5, characterized in that Before determining the combination scheme, the method further includes: Input all training output results output by the first inter-bank funds flow determination model and the first sample inter-bank funds flow into a univariate linear regression model to obtain the weight coefficient corresponding to the first inter-bank funds flow determination model, wherein the first sample inter-bank funds flow is the sample inter-bank funds flow corresponding to the sample time series feature data; Input all the training output results and the second sample inter-bank fund flow output by the second inter-bank fund flow determination model into the univariate linear regression model to obtain the weight coefficient corresponding to the second inter-bank fund flow determination model, where the second sample inter-bank fund flow is the sample inter-bank fund flow corresponding to the sample residual sequence feature data; All the training output results and the third sample inter-bank funds flow output by the third inter-bank funds flow determination model are input into the univariate linear regression model to obtain the weight coefficient corresponding to the third inter-bank funds flow determination model, and the third sample inter-bank funds flow is the sample inter-bank funds flow corresponding to the sample characteristic value.

7. The method according to claim 5, characterized in that After determining the combination scheme, the method further includes: Obtaining a first accuracy and a second accuracy, wherein the first accuracy is the accuracy of an output result of the first model, and the second accuracy is the accuracy of an output result of the second model; When the first accuracy is greater than the second accuracy, increasing the first weight coefficient and decreasing the second weight coefficient; When the first accuracy is less than the second accuracy, the first weight coefficient is decreased and the second weight coefficient is increased.

8. A device for determining inter-bank fund flow, characterized in that: The device comprises: A first acquisition unit is used to acquire current inter-bank fund flow characteristic data, wherein the current inter-bank fund flow characteristic data is divided into time series characteristic data, residual series characteristic data and characteristic value; An input-output unit, used for inputting the current inter-bank fund flow feature data into a corresponding inter-bank fund flow determination model to obtain a corresponding model output result, wherein the inter-bank fund flow determination model includes a first inter-bank fund flow determination model, a second inter-bank fund flow determination model and a third inter-bank fund flow determination model, wherein the first inter-bank fund flow determination model is a convergence model obtained by inputting sample time series feature data into an ARIMA model for training, the second inter-bank fund flow determination model is a convergence model obtained by inputting sample residual sequence feature data into an LSTM model for training, and the third inter-bank fund flow determination model is a convergence model obtained by inputting sample feature values ​​into an LGB model for training; The combination unit is used to perform weighted combination on the output results of multiple models to obtain the target inter-bank capital flow.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method according to any one of claims 1 to 7.

10. A computer program product comprising computer instructions, characterized in that: When the computer instructions are executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images