CN116433020A

CN116433020A - Method and equipment for resource main body risk early warning

Info

Publication number: CN116433020A
Application number: CN202310360285.9A
Authority: CN
Inventors: 李红松; 孙常龙
Original assignee: Alibaba China Co Ltd
Current assignee: Alibaba China Co Ltd
Priority date: 2023-03-31
Filing date: 2023-03-31
Publication date: 2023-07-14

Abstract

The application provides a method and equipment for early warning risk of a resource main body. According to the method, time sequence characteristic data of a target resource main body are generated according to characteristic information of the target resource main body to be predicted at a plurality of historical time points; predicting the variation of a preset index of a target resource main body in a future period according to the time sequence characteristic data; when the change amount of the preset index of the target resource main body in the future period meets the early warning condition, the risk early warning processing is executed, the change amount of the preset index in the longer future period can be accurately predicted by utilizing the characteristics and time sequence information of a plurality of historical time points, the change amount of the preset index indicates the development state of the target resource main body in the future period, the change amount of the preset index of the target resource main body in the future period meets the early warning condition, the important risk event of the target resource main body after the future period is illustrated, and the risk event of the resource main body after the future period can be accurately predicted, and the early warning processing is carried out in advance.

Description

Method and equipment for resource main body risk early warning

Technical Field

The present disclosure relates to computer technologies, and in particular, to a method and apparatus for risk early warning of a resource main body.

Background

For various resource main bodies, the influence of a plurality of factors such as market environment, self development condition and the like exists the risk of occurrence of events such as default, greatly reduced evaluation level, greatly reduced data statistics index and the like of the resource main bodies, and huge losses can be brought to audience users of the resource main bodies. The risk of the risk event of the resource main body in a longer period in the future is analyzed and early-warned, so that the loss of the user can be effectively reduced or avoided.

In order to early warn the risk event of the resource main body, the conventional machine learning model predicts the current risk event of the resource main body according to simple characteristic data such as basic information, historical records, credibility evaluation level and the like of the resource main body, but the accuracy for predicting the future long-term risk of the resource main body is very low, even can not be predicted, and the prediction and early warning of the long-term risk of the resource main body can not be performed.

Disclosure of Invention

The application provides a method and equipment for early warning of risk of a resource main body, which are used for solving the problems that in the prior art, the accuracy of predicting the future long-term risk of the resource main body is very low and even unpredictable.

In one aspect, the present application provides a resource main body risk early warning method, applied to a server deployed in a cloud, including:

Receiving a risk early warning request sent by a terminal side device to a target resource main body;

acquiring characteristic information of the target resource main body at a plurality of historical time points;

generating time sequence characteristic data of the target resource main body according to the characteristic information of the historical time points;

predicting the change quantity of a preset index of the target resource main body in a future period according to the time sequence characteristic data;

and when the change amount of the preset index of the target resource main body in the future period meets the early warning condition, sending early warning information to the terminal side equipment.

In another aspect, the present application provides a method for risk early warning of a resource main body, including:

acquiring characteristic information of a target resource main body to be predicted at a plurality of historical time points;

and executing risk early warning processing when the change amount of the preset index of the target resource main body in a future period meets the early warning condition.

On the other hand, the application provides a resource main body risk early warning method, which is applied to a locally deployed server and comprises the following steps:

Responding to an early warning monitoring request of a target resource main body, and acquiring characteristic information of the target resource main body at a plurality of historical time points;

and sending out warning information when the change amount of the preset index of the target resource main body in the future period meets the early warning condition.

In another aspect, the present application provides a server, including: a processor, and a memory communicatively coupled to the processor; the memory stores computer-executable instructions; the processor executes computer-executable instructions stored in the memory to implement the method of any of the above aspects.

In another aspect, the present application provides a computer-readable storage medium having stored therein computer-executable instructions that, when executed by a processor, are configured to implement the method of any one of the above aspects.

According to the method and the device for risk early warning of the resource main body, the characteristic information of the target resource main body to be predicted at a plurality of historical time points is obtained, and time sequence characteristic data of the target resource main body are generated according to the characteristic information of the historical time points; predicting the variation of a preset index of a target resource main body in a future period according to the time sequence characteristic data; when the change amount of the preset index of the target resource main body in the future period meets the early warning condition, the risk early warning processing is executed, the change amount of the preset index in the longer future period can be accurately predicted by utilizing the characteristics and time sequence information of a plurality of historical time points, the change amount of the preset index indicates the development state of the target resource main body in the future period, the change amount of the preset index of the target resource main body in the future period meets the early warning condition, the important risk event of the target resource main body after the future period is illustrated, and the risk event of the resource main body after the future period can be accurately predicted, and the early warning processing is carried out in advance.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the application and together with the description, serve to explain the principles of the application.

FIG. 1 is a schematic diagram of an exemplary resource-agent risk early warning system architecture applicable to the present application;

FIG. 2 is a flowchart of a method for resource principal risk early warning according to an exemplary embodiment of the present application;

FIG. 3 is a block diagram of a predictive model provided in an exemplary embodiment of the present application;

FIG. 4 is a flow chart of predictive model training provided in an exemplary embodiment of the present application;

fig. 5 is a flowchart of a risk early warning method based on a target resource main body evaluation level according to an exemplary embodiment of the present application;

FIG. 6 is an interactive flow chart of resource principal risk early warning provided in an exemplary embodiment of the present application;

FIG. 7 is a flowchart of a method for resource principal risk early warning according to another exemplary embodiment of the present application;

fig. 8 is a schematic structural diagram of a resource main body risk early warning device according to an exemplary embodiment of the present application;

fig. 9 is a schematic structural diagram of a cloud server according to an embodiment of the present application.

Specific embodiments thereof have been shown by way of example in the drawings and will herein be described in more detail. These drawings and the written description are not intended to limit the scope of the inventive concepts in any way, but to illustrate the concepts of the present application to those skilled in the art by reference to specific embodiments.

Detailed Description

Reference will now be made in detail to exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, the same numbers in different drawings refer to the same or similar elements, unless otherwise indicated. The implementations described in the following exemplary examples are not representative of all implementations consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with some aspects of the present application as detailed in the accompanying claims.

The terms referred to in this application are explained first:

transformer: is a deep learning model employing a self-attention mechanism that can be assigned different weights depending on the importance of the various portions of the input data. The model is mainly used in the fields of natural language processing (Natural Language Processing, NLP) and Computer Vision (CV).

Debt subject: the resource entity that issues the company bond is also referred to as bond issuing entity.

Evaluation grade: the result of evaluating the credibility or reliability of the resource main body is such as credit rating, qualification rating and the like.

In order to early warn the risk event of the resource main body, the conventional machine learning model predicts the current risk event of the resource main body according to simple characteristic data such as basic information, historical records, reliability evaluation level and the like of the resource main body. However, the accuracy for predicting the future long-term risk of the resource main body is very low, even unpredictable, and the prediction and early warning of the long-term risk of the resource main body cannot be performed.

In a longer period of the future, the method is influenced by factors such as future development plans, serious events (such as litigation, etc.) or environmental conditions, and huge barriers are caused to predicting evaluation levels or data statistics indexes of resource subjects after a longer period (such as 6 months) of the future due to the lack of a large amount of future period information. Considering the occurrence of major risk events of a resource subject, it is often not abrupt, but a process in which risks accumulate over a long period of time. During this long-term accumulation, some dangerous signals may occur in advance, such as long-term unhealthy data statistics, or a small down-regulation of a rating for a period of time before, and some dangerous signals may be obtained by using historical information. In addition, by analyzing the omnidirectional information of the resource main body for long-term persistence, the potential risk of the resource main body can be found. Whether dangerous signals or other information, it is possible to form specific patterns on the time series, which can be found by learning from historical data and used to predict risk events after a longer period (e.g., 6 months) in the future.

The application provides a resource main body risk early warning method, which comprises the steps of obtaining characteristic information of a target resource main body to be predicted at a plurality of historical time points, and generating time sequence characteristic data of the target resource main body according to the characteristic information of the historical time points; predicting the variation of a preset index of a target resource main body in a future period according to the time sequence characteristic data; when the change amount of the preset index of the target resource main body in the future period meets the early warning condition, the risk early warning processing is executed, the change amount of the preset index in the longer future period can be accurately predicted by utilizing the characteristics and time sequence information of a plurality of historical time points, the change amount of the preset index indicates the development state of the target resource main body in the future period, the change amount of the preset index of the target resource main body in the future period meets the early warning condition, the important risk event of the target resource main body after the future period is illustrated, and the risk event of the resource main body after the future period can be accurately predicted, and the early warning processing is carried out in advance.

The method provided by the application can be particularly applied to early warning of the following risk events, but is not limited to: the evaluation level of the resource main body is greatly reduced in the future period, the key data statistical index is greatly reduced in the future period, and the like.

Fig. 1 is a schematic diagram of an exemplary resource main risk early warning system architecture applicable to the present application, and as shown in fig. 1, the system architecture may specifically include a server, an end-side device, and a data production device.

The server may be a server cluster disposed in the cloud, or a server deployed locally. Communication links capable of being communicated are arranged between the server and each end side device, and communication connection between the server and each end side device can be achieved. The server stores a pre-trained prediction model, and predicts the variation of the preset index of the resource main body in a future period through the prediction model.

The terminal device may specifically be a hardware device with a network communication function, an operation function and an information display function, which includes, but is not limited to, a terminal device used by a user, such as a smart phone, a tablet computer, a desktop computer, and the like, and may also be an internet of things device, a server of other service platforms, and the like.

The data production facility may gather various types of historical data (public data or data available via the authority of the resource principal) for the resource principal and provide the historical data for the resource principal to the server. The server acquires historical data of a target resource main body to be predicted from the data production equipment, and generates characteristic information of the target resource main body at a plurality of historical time points based on the historical data. In an alternative embodiment, the end-side device may act as a data production device, obtain the historical data of the resource principal, and send the historical data of the resource principal to the server.

The user can be a holder of the resource main body product, a person who wants to purchase or know, a selling agency and the like, and interacts with the server through the used terminal side equipment to realize the resource main body risk early warning function. Specifically, the user may send a request/instruction to the server through the used end-side device, and send the relevant information of the target resource body to be predicted to the server. The server determines a target resource main body based on the request/instruction, acquires characteristic information of the target resource main body at a plurality of historical time points, and generates time sequence characteristic data of the target resource main body according to the characteristic information of the plurality of historical time points; according to the time sequence characteristic data, predicting the change quantity of a preset index of a target resource main body in a future period by using a prediction model; and executing risk early warning processing when the change amount of the preset index of the target resource main body in the future period meets the early warning condition. For example, the server may send early warning information to the end-side device, and may also send the target resource body to the end-side device with a change amount of the preset index in a future period. The terminal side equipment outputs early warning information to prompt a user that the target resource main body has a risk event that the preset index greatly fluctuates after a future preset period, and can also output the change quantity of the preset index of the target resource main body in the future period.

A typical application scenario is risk early warning based on an evaluation level of a resource main body, where a user can issue a holder of a product, a person who wants to purchase or know, a sales agency, and the like. When the user wants to know whether the resource main body has a long-term risk event, the user can send a request/instruction to the server through the used end-side equipment, and send related information of the resource main body to be predicted to the server. The server determines a target resource main body based on the request/instruction, acquires characteristic information of the target resource main body at a plurality of historical time points, and generates time sequence characteristic data of the target resource main body according to the characteristic information of the plurality of historical time points; according to the time sequence characteristic data, predicting the change quantity of the evaluation level of the target resource main body in a future period by using a prediction model; and executing risk early warning processing when the change amount of the evaluation level of the target resource main body in the future period is less than or equal to the preset level change amount (negative value). For example, the server may send first pre-warning information to the end-side device, the first pre-warning information indicating that the target resource body is at risk of a significant downturn of the assessment level within the future period. The server may also transmit the amount of change in the evaluation level of the target resource body in the future period to the end-side device. The terminal side equipment outputs the first early warning information and outputs the change quantity of the evaluation level of the target resource main body in a future period.

Another typical application scenario is risk early warning of a data statistics index of a resource main body, a server can predict the change amount of a preset data statistics index of a target company in a future period, and when the change amount of the preset data statistics index of the target resource main body in the future period is smaller than or equal to a change amount threshold (negative value) of the preset data statistics index, second risk early warning information is output, and the second risk early warning information is used for indicating the risk that the operating condition of the target resource main body is greatly reduced in the future period.

The following describes the technical solutions of the present application and how the technical solutions of the present application solve the above technical problems in detail with specific embodiments. The following embodiments may be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments. Embodiments of the present application will be described below with reference to the accompanying drawings.

Fig. 2 is a flowchart of a resource main body risk early warning method according to an exemplary embodiment of the present application. The execution resource main body of the embodiment is a server in the resource main body risk early warning system architecture. As shown in fig. 2, the method specifically comprises the following steps:

Step S201, obtaining feature information of a target resource main body to be predicted at a plurality of historical time points.

The feature information of any historical time point comprises information of multiple dimensions, including but not limited to basic information of a target resource main body, evaluation level information, various information in financial report data and various resource indexes. Converting non-numerical information in each dimension information into numerical information, and arranging the numerical information of a plurality of dimensions of the same historical time point according to a certain sequence to form characteristic information of the historical time point.

Illustratively, at least one of the following characteristic information of the target resource body to be predicted at a plurality of historical time points is acquired: presetting indexes, time characteristics and resource value parameters. The resource value parameter may specifically include, but is not limited to, various information in the financial report data of the target resource main body, and various resource indexes of the target resource main body.

The feature information of the historical time points may contain different information dimensions when applied to different application scenarios, for example to different types of resource principals, or to the prediction of the amount of change of different preset indicators.

Optionally, when the method is applied to a risk early warning scene based on an evaluation level (preset index) of a resource main body, the time of the historical rating of the resource main body can be taken as a historical time point, and the characteristic information of the resource main body to be predicted when the historical rating is repeated can be obtained, which specifically can include at least one kind of the following characteristic information: evaluation level, rating time, financial reporting characteristics and resource index characteristics.

Optionally, when applied to a risk early warning scene based on a data statistics index (preset index) of a resource main body, a plurality of historical time points can be selected from a longer historical period (usually several years, such as 2 years, 3 years, etc.), and feature information of the resource main body to be predicted at the plurality of historical time points can be obtained, which specifically can include at least one type of feature information as follows: data statistics index, time point, financial report characteristic and resource index characteristic.

Step S202, generating time sequence characteristic data of the target resource main body according to the characteristic information of a plurality of historical time points.

After the characteristic information of a plurality of historical time points of the target resource main body is acquired, generating time sequence characteristic data according to the characteristic information of the historical time points, so that the data structure of the time sequence characteristic data is matched with the structure of the input data of the prediction model coding module.

For example, an encoding module and a decoding module of the transducer model may be used as the encoding module and the decoding module of the prediction model, respectively. Since the transducer model is designed for natural language processing, the transducer model includes an embedding layer that maps natural language information into embedded vectors. In the prediction model used in this embodiment, an embedding layer (embedding) in a transform model is removed, and the time series feature data is directly input into the encoding module by converting feature information of a plurality of history time points of the target resource main body into the time series feature data.

Optionally, in this step, the feature information of the plurality of historical time points may be arranged according to a time sequence to form the time sequence feature data.

Optionally, in this step, key features of multiple dimensions with high correlation with the preset index of the target resource main body may be screened from the feature information of multiple historical time points, and the key features of the historical time points are arranged according to time sequence to form the time sequence feature data.

In this embodiment, the time series characteristic data may be represented as a two-dimensional matrix including a characteristic dimension and a time dimension. The time dimension contains a plurality of different historical points in time. The feature dimension contains features of the target resource principal at specific historical points in time.

For example, taking a risk early warning scenario based on an evaluation level (preset index) of a resource main body as an example, assume that for any historical time point, the characteristics of the resource main body include: the method comprises the steps of evaluating the grade, grading time, various financial reporting characteristics, various resource index characteristics and the like, wherein the total number of the characteristics is 1084. The characteristics of the history at 12 ratings were acquired, 1084×12 pieces of characteristic data were obtained in total, and stored as 1084×12 pieces of time series characteristic data. The time series characteristic data contains 1083 (except time) evaluation levels and financial data characteristics of the resource main body at each of the past 12 history time points.

Based on the financial accounting data of the target resource entity, a plurality of (about 600) pieces of characteristic information can be extracted from the financial accounting data, including but not limited to the operation profit margin, the revenue per share and the fluid asset profit margin. Based on the resource metrics of the target resource entity, a plurality of (about 400) pieces of characteristic information can be extracted from the target resource entity, including but not limited to the total production value, inflation and deflation of the target resource entity, investment metrics, consumption metrics, financial metrics.

Step S203, according to the time sequence characteristic data, predicting the change amount of the preset index of the target resource main body in the future period.

In the step, time sequence characteristic data of the target resource main body are input into a trained prediction model, and the change amount of the preset index of the target resource main body in a future period is predicted through the prediction model.

The preset index is an index, for example, an evaluation level and a data statistics index, where the change amount can represent the possibility of occurrence of a risk event by the target resource main body.

Optionally, the time sequence feature data of the target resource main body can be directly input into an encoding module of the prediction model to be encoded, so as to obtain encoding features, and the encoding features are decoded by a decoding module, so that the variation of the preset index of the target resource main body in a future period is obtained. The predictive model may use a sequence-to-sequence (seq 2 seq) model such as a Recurrent Neural Network (RNN), convolutional Neural Network (CNN), etc.

Optionally, taking a prediction model based on a transform model as an example, in the step, according to a sequence (corresponding to a historical time point) in which each item of feature information in the time sequence feature data is located and a position in the sequence, generating a time sequence position feature; and (3) inputting the fusion characteristic obtained by fusing the time sequence position characteristic and the time sequence characteristic data into a coding module of the prediction model for coding to obtain a coding characteristic. Further, the coding features are decoded through a decoding module of the prediction model, and the variation of the preset index of the target resource main body in the future period is obtained. The method has the advantages that the self-attention mechanism of the transducer model coding module can be utilized to capture the rules or modes of the features in different time phases through the transducer model-based prediction model, so that the characterization related to risk early warning, particularly the characterization related to long-term risk, is generated, the change of the preset index of the target resource main body in a longer time period in the future can be accurately predicted, and a better early warning effect is obtained.

When the time sequence position feature is fused with the time sequence feature data, the time sequence position feature and the time sequence feature data can be summed to obtain a fused feature.

Fig. 3 is an architecture diagram of a prediction model according to an exemplary embodiment of the present application, and after summing time sequence feature data and time sequence position features, as shown in fig. 3, the time sequence feature data is used as input features of a coding module of the prediction model, and is subjected to coding processing of the coding module and decoding processing of the decoding module, so as to obtain a prediction result.

And step S204, executing risk early warning processing when the change amount of the preset index of the target resource main body in the future period meets the early warning condition.

In this embodiment, an early warning condition for performing risk early warning may be preconfigured, where the early warning condition indicates a boundary value of a variation of a preset index when a major risk event occurs in a resource main body. The pre-warning condition can be configured and adjusted according to the requirements of the actual application scene, and is not particularly limited herein.

After the change amount of the preset index of the target resource main body in the future period is predicted, judging whether the change amount of the preset index of the target resource main body in the future period meets the early warning condition or not. And executing risk early warning processing under the condition that the change amount of the preset index of the target resource main body in the future period meets the early warning condition.

Optionally, when the early warning conditions are set, a plurality of early warning conditions for carrying out different levels of risk levels can be set, and risk early warning processing modes corresponding to the early warning conditions of different risk levels are configured. The risk early warning processing modes performed under different levels of early warning conditions can be different. In this embodiment, the specific implementation manner of the set early warning condition and risk early warning process is not specifically limited.

In this embodiment, by acquiring feature information of a target resource main body to be predicted at a plurality of historical time points, time sequence feature data of the target resource main body is generated according to the feature information of the plurality of historical time points; predicting the variation of a preset index of a target resource main body in a future period according to the time sequence characteristic data; when the change amount of the preset index of the target resource main body in the future period meets the early warning condition, the risk early warning processing is executed, the change amount of the preset index in the longer future period can be accurately predicted by utilizing the characteristics and time sequence information of a plurality of historical time points, the change amount of the preset index indicates the development state of the target resource main body in the future period, the change amount of the preset index of the target resource main body in the future period meets the early warning condition, the important risk event of the target resource main body after the future period is illustrated, and the risk event of the resource main body after the future period can be accurately predicted, and the early warning processing is carried out in advance.

In an alternative embodiment, in the scenario of risk early warning based on the evaluation level of the target resource main body, the preset index is the evaluation level, and in the step S203, the change amount of the evaluation level of the target resource main body in the future period is predicted according to the time sequence feature data. In step S204, when the change amount of the evaluation level of the target resource body in the future period is less than or equal to the preset level change amount, the first risk early warning information is output. The preset grade change amount is a negative number, and the first risk early warning information is used for indicating that the target resource main body has the risk of greatly adjusting the evaluation grade in a future period.

The target resource entity may be a debt entity or other business entity, etc. The preset level change amount can be set and adjusted according to the needs of the actual application scene, and is not particularly limited herein. For example, the preset level change amount may be-3, and when the change amount of the evaluation level of the resource main body in the future 6 months is less than or equal to-3, the evaluation level of the resource main body is greatly adjusted down, and the first risk early warning information is output.

Optionally, when the absolute value of the change amount of the evaluation level of the target resource body in the future period is greater than or equal to the second level change amount, third risk early warning information is output. The second level change amount is a larger positive number, and the third risk early warning information is used for indicating that the target resource main body has a situation that the evaluation level greatly fluctuates in a future period.

In an optional embodiment, in the scenario of risk early warning based on the data statistics index of the target resource main body, the preset index is the data statistics index, and in the step S203, the change amount of the preset data statistics index of the target resource main body in the future period is predicted according to the time sequence feature data. In step S204, when the variation of the preset data statistics index of the target resource main body in the future period is less than or equal to the variation threshold of the preset data statistics index, the second risk early warning information is output. The second risk early warning information is used for indicating the risk that the operating condition of the target resource main body is greatly reduced in a future period.

The target resource body may be a marketing company or other enterprise institutions. The variation threshold of the preset data statistics index can be set and adjusted according to the needs of the actual application scene, and different preset operation indexes can correspond to different variation thresholds, which are not particularly limited herein.

Optionally, when the absolute value of the variation of the preset data statistics index of the target resource main body in the future period is greater than or equal to the second variation threshold, fourth risk early warning information is output. The second variation threshold is a larger positive number, and the fourth risk early warning information is used for indicating that the target resource main body has a situation that the operation condition greatly fluctuates in a future period.

In the embodiment of the application, performing risk early warning based on different preset indexes is understood as different risk early warning scenes, performing risk early warning on different types of resource main bodies is also understood as different risk early warning scenes, and for different risk early warning scenes, a data set training prediction model of a specific risk early warning scene is required to be used. The original prediction models before training in different risk early warning scenes can be the same, but after training in different training sets, the prediction models with different model parameters are obtained, namely, the prediction models used in different risk early warning scenes have different model parameters.

In an alternative embodiment, a transform model-based prediction model is used to predict the amount of change in the target resource agent's preset index over a future period of time. Fig. 4 is a flowchart of the predictive model training provided in this embodiment. As shown in fig. 4, for a given resource main body type and a preset index in a specific risk early-warning scene, a prediction model used in the risk early-warning scene is obtained through training by the following steps:

step S401, a training set is constructed, wherein the training set comprises a plurality of pieces of training data, and the training data comprises actual variation information of preset indexes of a sample resource main body in a history period and characteristic information of a plurality of history time points of the sample resource main body before the history period.

In this embodiment, a plurality of resource principals of the type are selected as sample resource principals according to the type of the resource principal, and one or more pieces of training data of the sample resource principal are generated based on the historical data of the sample resource principal.

Specifically, the historical data can be acquired for one sample resource main body, the actual change amount of the preset index of the sample resource main body in one historical period is determined, and then the characteristic information of a plurality of historical time points is acquired in the historical data before the historical period. The actual variation of the preset index of the sample resource main body in a history period can be calculated according to the preset index of the sample resource main body in the beginning time of the history period and the preset index of the sample resource main body in the ending time of the history period.

Taking a risk early warning scenario based on the evaluation level of a resource main body as an example, for any one of the evaluation levels of any one of the resource main bodies, the actual variation information of the evaluation level of the resource main body in the history period of the preset period can be determined according to the result of the evaluation (i.e. the evaluation level) and the result of the last evaluation before the preset period (the period of the future period). Further, multiple historical ratings of the resource main body before the historical period of the preset duration are obtained as multiple historical time points, and feature information of the multiple historical time points is obtained. The characteristic information of the plurality of historical time points and the actual variation information of the evaluation level of the resource main body in the historical period of the preset duration form training data. Multiple pieces of training data for multiple resource principals may be obtained based on similar methods. The starting time of the preset time period (the time period of the future period) or the last rated time before the preset time period (the time period of the future period) is taken as a prediction node.

Optionally, for a sample resource main body with a large fluctuation of a preset index in a short period, that is, a sample resource main body with a preset index in a short period is quickly recovered after being greatly up/down regulated, so that the sample resource main body does not accord with the characteristic of long-term early warning, the sample resource main body can be screened according to the preset index value in the short period, and the sample resource main body with the fluctuation range of the preset index in the short period (the preset time range) being larger than or equal to the preset fluctuation value is removed. The preset time range and the preset fluctuation value can be set and adjusted according to actual application scenes and experience values, and are not particularly limited herein.

In this embodiment, the verification set and the test set may be constructed in a similar manner to that of the training set, where the data in the verification set and the test set should use newer historical data to fully verify and test the accuracy of the prediction model, and improve the accuracy of the trained prediction model. For example, the interval of the predicted nodes of the training data in the training set is 2017-2021.2 months, the verification set is 2021.3-2021.6 months, and the test set is 2021.7-2021.8. The predicted node refers to a start time of a preset duration (duration of a future period) or a time when the preset index was last determined before the preset duration (duration of the future period) when the training data is generated.

In this embodiment, in the same risk early warning scene, the training stage obtains the feature information of the historical time point of the sample resource main body, and the feature information of the historical time point of the target resource main body obtained during risk early warning in the above embodiment contains the feature information of the same number and type, which is not described herein again.

Step S402, generating time sequence characteristic data of the sample resource main body according to the characteristic information of the sample resource main body.

The implementation manner of generating the time sequence feature data of the target resource main body according to the feature information of the plurality of historical time points of the target resource main body in the step S202 is the same as that of the previous embodiment, and details of the foregoing description are omitted here.

Step S403, inputting the time sequence characteristic data of the sample resource main body into an initial prediction model, and predicting the variation prediction information of the preset index of the sample resource main body in the corresponding history period.

In this step, the specific implementation manner of predicting the variation of the preset index of the target resource body in the future period according to the time sequence feature data in the step S203 is the same as that in the previous embodiment, and the specific reference is omitted here.

And step S404, calculating loss according to the variation prediction information and the actual variation information of the preset indexes of the sample resource main body in the corresponding historical period, and updating model parameters of the prediction model according to the loss to obtain a trained prediction model.

In an alternative embodiment, the cross entropy loss can be calculated according to the variation prediction information and the actual variation information of the preset index of the sample resource main body in the corresponding historical period, and the model parameters of the prediction model are updated according to the cross entropy loss, so that the trained prediction model is obtained.

In an alternative embodiment, the predictive model may be a classification model. The training set comprises preset variable quantity categories, and different categories correspond to different variable quantity intervals. When the preset index is a discrete value (such as an evaluation level), the variation interval corresponding to the category of the variation may only include one numerical value. For example, the categories of the amounts of change in the evaluation level correspond to the following amounts of change, respectively: -11, -10, -9, -8, -7, -6, -5, -4, -3, -2, -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 can be divided into 22 categories.

The actual variation information of the preset index of the sample resource main body in the training data in the training set in a history period can be a category corresponding to the actual variation of the preset index of the sample resource main body in the history period.

In a prediction result obtained through the prediction model, the variation prediction information of the preset index of the sample resource main body in the corresponding history period is the prediction category of the variation of the preset index of the sample resource main body in the corresponding history period.

The problem of unbalanced amounts of training data of different categories is often severe in view of the training set constructed from historical data. In an example of a risk early warning scenario based on an evaluation level of a resource main body, the number of training data included in each class in a training set generated based on existing historical data is shown in the following table 1, and it is obvious that the number of training data in each class is very unbalanced, and a problem of over fitting is easily caused when a prediction model is trained.

TABLE 1

Category(s)	Quantity of training data
		-11	5
-10	6
		-9	13
-8	50
		-7	30
-6	27
		-5	69
-4	122
		-3	161
-2	957
		-1	4003
0	116095
		1	2572
2	578
		3	50
4	37
		5	13
6	20
		7	2
8	18
		9	2
10	5

In order to solve the problem of overfitting caused by unbalance of a data set, when calculating loss according to variation prediction information and actual variation information of preset indexes of a sample resource main body in a corresponding historical period, determining weight coefficients of categories of variation according to the number of training data of different categories in a training set; and calculating the weighted cross entropy loss according to the prediction category of the variation of the preset index of the sample resource main body in the corresponding history period, the category of the actual variation and the weight coefficient. Further, model parameters of the prediction model are updated based on weighted cross entropy loss, so that the problem of over fitting can be effectively prevented, and the accuracy of the prediction model obtained through training is improved.

Illustratively, the weight coefficient of the category of the variation is determined according to the number of training data of different categories in the training set, and specifically, the weight coefficient of the different categories may be determined by adopting the following formula: w_i=sqrt (N/n_i), where w_i represents the i-th category, n_i represents the number of training data of the i-th category, N represents the total number of training data, and sqrt () is an open root number (open square) operation. By determining the weight coefficient of each category in the method, the problem of overfitting caused by unbalanced training data categories can be prevented, the overlarge weight of small categories (categories with less training data) can be prevented, and the performance of the prediction model can be effectively improved.

Alternatively, the weight coefficients of different categories may also be determined by normalizing N/n_i, so that the sum of the weight coefficients of different categories is 1. Where w_i represents the i-th category, n_i represents the number of training data of the i-th category, and N represents the total number of training data.

By the method of the embodiment, the prediction model applicable to various risk early warning scenes can be obtained through training. Based on the collected scene that the evaluation level of nearly 200 resource main bodies and the evaluation level before 6 months are subjected to 3-level or more down regulation, the accuracy of a prediction result obtained by using the prediction model can reach 82%, and the target exceeds the original 65%, so that effective risk early warning can be performed.

Fig. 5 is a flowchart of a risk early warning method based on a target resource main body evaluation level according to an exemplary embodiment of the present application. The method of the embodiment is applied to a scene of risk early warning based on the evaluation level of the target resource main body. As shown in fig. 5, the method specifically comprises the following steps:

step S501, obtaining feature information of a target resource main body to be predicted when the target resource main body is rated for multiple times.

In this embodiment, a time node of one history rating of the target resource main body is taken as one history time point, and feature information of the target resource main body during multiple history ratings is obtained, so that feature information of multiple history time points of the target resource main body is obtained.

Specifically, when the method is applied to a risk early warning scene based on an evaluation level (preset index) of a resource main body, the time of the historical rating of the resource main body can be taken as a historical time point, and the characteristic information of the resource main body to be predicted when the historical rating is repeated can be obtained, and specifically, the method can comprise at least one kind of characteristic information as follows: evaluation level, rating time, financial reporting characteristics and resource index characteristics.

Step S502, generating time sequence characteristic data of the target resource main body according to the characteristic information in the multiple historical grading.

After the characteristic information of the target resource main body in the multiple historical ratings is obtained, generating time sequence characteristic data according to the characteristic information in the multiple historical ratings, so that the data structure of the time sequence characteristic data is matched with the structure of the input data of the prediction model coding module.

Optionally, in this step, the characteristic information in the multiple historical ratings may be arranged according to a time sequence, so as to form time sequence characteristic data.

Optionally, in this step, key features of multiple dimensions with high correlation with the preset index of the target resource main body may be screened from the feature information during multiple historical ratings, and the key features during multiple historical ratings may be arranged according to time sequence to form time sequence feature data.

For example, taking a risk early warning scenario based on an evaluation level (preset index) of a resource main body as an example, assume that for any historical time point, the characteristics of the resource main body include: the method comprises the steps of evaluating level, evaluating time, various financial reporting characteristics, various resource index characteristics and the like, wherein the total number of the characteristics is 1084. The characteristics of the history at 12 ratings were acquired, 1084×12 pieces of characteristic data were obtained in total, and stored as 1084×12 pieces of time series characteristic data. The time series characteristic data contains 1083 (except time) evaluation levels and financial data characteristics of the resource main body at each of the past 12 history time points.

Step S503, according to the time sequence characteristic data, predicting the change amount of the evaluation level of the target resource main body in the future period.

In the step, time sequence characteristic data of the target resource main body are input into a trained prediction model, and the change amount of the evaluation level of the target resource main body in a future period is predicted through the prediction model.

Optionally, taking a prediction model based on a transform model as an example, in the step, according to a sequence (corresponding to a historical time point) in which each item of feature information in the time sequence feature data is located and a position in the sequence, generating a time sequence position feature; and (3) inputting the fusion characteristic obtained by fusing the time sequence position characteristic and the time sequence characteristic data into a coding module of the prediction model for coding to obtain a coding characteristic. Further, the coding features are decoded through a decoding module of the prediction model, and the change amount of the evaluation level of the target resource main body in the future period is obtained. Through the prediction model based on the transducer model, the correlation among the characteristics of different sequences (different historical time points) in the time sequence characteristic data of the target resource main body can be fully learned by utilizing the time sequence position characteristic, so that the change quantity of the evaluation level of the target resource main body in a longer time period in the future can be accurately predicted.

Step S504, when the change amount of the evaluation level of the target resource main body in the future period is equal to or more than the preset level change amount, outputting first risk early warning information, wherein the preset level change amount is a negative number, and the first risk early warning information is used for indicating that the target resource main body has the risk of greatly reducing the evaluation level in the future period.

In this embodiment, an early warning condition for performing risk early warning may be preconfigured, where the early warning condition indicates a boundary value of a variation of the evaluation level when a major risk event occurs in the resource main body. The pre-warning condition can be configured and adjusted according to the requirements of the actual application scene, and is not particularly limited herein.

After the change amount of the evaluation level of the target resource main body in the future period is predicted, judging whether the change amount of the evaluation level of the target resource main body in the future period meets the early warning condition. And executing risk early warning processing under the condition that the change amount of the evaluation level of the target resource main body in the future period meets the early warning condition.

In the embodiment, by acquiring the characteristic information of the target resource main body to be predicted when the target resource main body is rated for a plurality of times, generating time sequence characteristic data of the target resource main body according to the characteristic information when the target resource main body is rated for a plurality of times; according to the time sequence characteristic data, predicting the change quantity of the evaluation level of the target resource main body in a future period; when the change amount of the evaluation level of the target resource main body in the future period meets the early warning condition, risk early warning processing is executed, the change amount of the evaluation level of the target resource main body in the future longer period can be accurately predicted by utilizing the characteristic and time sequence information when the history is rated for many times, the change amount of the evaluation level indicates the development state of the target resource main body in the future period, the change amount of the evaluation level of the target resource main body in the future period meets the early warning condition, and the important risk event of the target resource main body after the future period is illustrated, so that the risk event of the resource main body after the future period can be accurately predicted, and early warning processing is carried out in advance.

Fig. 6 is an interactive flowchart of resource entity risk early warning according to an exemplary embodiment of the present application, where an executing entity of the present application is a locally deployed server. As shown in fig. 6, the method specifically comprises the following steps:

In step S601, the terminal device sends a risk early warning request to the server for the target resource body.

The risk early warning request for the target resource main body comprises identification information of the target resource main body, such as the name of the target resource main body, an organization code and the like.

Step S602, a server receives a risk early warning request for a target resource main body.

After receiving a risk early warning request for a target resource main body sent by end side equipment, a server acquires identification information of the target resource main body from the risk early warning request. Further, the server obtains feature information of the target resource main body at a plurality of historical time points according to the identification information of the target resource main body.

In step S603, the server acquires feature information of the target resource main body at a plurality of historical time points.

Step S604, the server generates time sequence characteristic data of the target resource main body according to the characteristic information of the plurality of historical time points.

Step S605, the server predicts the variation of the preset index of the target resource main body in the future period according to the time sequence characteristic data.

In this embodiment, the specific implementation manner of the steps S603 to S605 is the same as that of the steps S201 to S203, and details of the foregoing embodiment are referred to in the related content, and are not repeated here.

Step S606, the server sends early warning information to the terminal side equipment when the change amount of the preset index of the target resource main body in the future period meets the early warning condition.

After the change amount of the preset index of the target resource main body in the future period is predicted, judging whether the change amount of the preset index of the target resource main body in the future period meets the early warning condition or not. And sending early warning information to the terminal side equipment under the condition that the change amount of the preset index of the target resource main body in a future period meets the early warning condition.

Optionally, when the early warning conditions are set, a plurality of early warning conditions for carrying out different levels of risk levels can be set, and different levels of early warning information corresponding to the early warning conditions of different risk levels can be configured. The pre-warning processing corresponding to the pre-warning information of different levels can be different.

Step S607, the end device receiving server sends the early warning information.

And step 608, the terminal side equipment performs early warning processing according to the early warning information.

In this embodiment, taking the execution subject as the server deployed in the cloud as an example, an example of an interaction flow of resource subject risk early warning is provided.

Fig. 7 is a flowchart of a resource main body risk early warning method according to another exemplary embodiment of the present application, where an execution main body of the present application is a locally deployed server. As shown in fig. 7, the method specifically comprises the following steps:

step S701, in response to an early warning monitoring request for a target resource main body, acquiring feature information of the target resource main body at a plurality of historical time points.

In this embodiment, the user may send an early warning monitoring request for the target resource main body to the server through a client or an interface provided by the server, where the early warning monitoring request includes identification information of the target resource main body, for example, a name, an organization code, and the like of the target resource main body.

The server responds to the received early warning monitoring request of the target resource main body, and obtains the characteristic information of the target resource main body at a plurality of historical time points according to the identification information of the target resource main body.

The feature information acquired by the server in this step is the same as that in the aforementioned step S201, and will not be described here again.

Step S702, generating time sequence characteristic data of the target resource main body according to the characteristic information of a plurality of historical time points.

Step S703, predicting the variation of the preset index of the target resource body in the future period according to the time sequence feature data.

In this embodiment, the specific implementation manner of the steps S702-S703 is the same as that of the steps S202-S203, and detailed descriptions thereof are omitted herein.

Step S704, when the change amount of the preset index of the target resource main body in the future period meets the early warning condition, warning information is sent out.

After the change amount of the preset index of the target resource main body in the future period is predicted, judging whether the change amount of the preset index of the target resource main body in the future period meets the early warning condition or not. And under the condition that the change amount of the preset index of the target resource main body in the future period meets the early warning condition, the server sends out warning information.

Optionally, when the early warning conditions are set, a plurality of early warning conditions for carrying out different levels of risk levels can be set, and different levels of warning information corresponding to the early warning conditions of different risk levels can be configured.

In this embodiment, taking the execution subject as a locally deployed server as an example, an example of a process of risk early warning of a resource subject is provided.

Fig. 8 is a schematic structural diagram of a resource main body risk early warning device according to an exemplary embodiment of the present application. The resource main body risk early warning device provided by the embodiment of the application can execute the processing flow provided by the resource main body risk early warning method embodiment. As shown in fig. 8, the resource main body risk early warning device 80 includes: a feature acquisition module 81, a timing feature generation module 82, a prediction module 83, and an early warning module 84.

The feature obtaining module 81 is configured to obtain feature information of a target resource main body to be predicted at a plurality of historical time points.

The time sequence feature generation module 82 is configured to generate time sequence feature data of the target resource body according to feature information of a plurality of historical time points.

The prediction module 83 is configured to predict a variation of a preset index of the target resource body in a future period according to the time sequence feature data.

The early warning module 84 is configured to execute risk early warning processing when the variation of the preset index of the target resource body in the future period satisfies the early warning condition.

In an alternative embodiment, in implementing the obtaining of the feature information of the target resource body to be predicted at a plurality of historical time points, the feature obtaining module 81 is further configured to:

acquiring at least one piece of characteristic information of a target resource main body to be predicted at a plurality of historical time points, wherein the characteristic information comprises the following characteristics: presetting indexes, time characteristics and resource value parameters.

In an alternative embodiment, in implementing generating the time sequence feature data of the target resource body according to the feature information of the plurality of historical time points, the time sequence feature generating module 82 is further configured to:

and arranging the characteristic information of the plurality of historical time points according to the time sequence to form time sequence characteristic data.

In an alternative embodiment, when predicting the amount of change of the preset index of the target resource body in the future period according to the time sequence feature data, the prediction module 83 is further configured to:

generating time sequence position features according to the sequence of each item of feature information in the time sequence feature data and the position in the sequence; the time sequence position features and the time sequence feature data are fused to obtain fusion features, and the fusion features are input into a coding module of a prediction model for coding to obtain coding features; and decoding the coding features through a decoding module of the prediction model to obtain the variation of the preset index of the target resource main body in a future period.

In an alternative embodiment, the predetermined index is an evaluation level. In predicting the amount of change of the preset index of the target resource body in the future period according to the time sequence feature data, the prediction module 83 is further configured to:

and predicting the change amount of the evaluation level of the target resource main body in the future period according to the time sequence characteristic data.

Further, when implementing the risk early warning process when the variation of the preset index of the target resource main body in the future period meets the early warning condition, the early warning module 84 is further configured to:

and outputting first risk early warning information when the change amount of the evaluation level of the target resource main body in the future period is smaller than or equal to the change amount of the preset level, wherein the change amount of the preset level is negative.

In an alternative embodiment, the predetermined index is a data statistics index. In predicting the amount of change of the preset index of the target resource body in the future period according to the time sequence feature data, the prediction module 83 is further configured to:

and predicting the variation of the preset data statistics index of the target resource main body in a future period according to the time sequence characteristic data.

Outputting second risk early warning information when the variation of the preset data statistics index of the target resource main body in the future period is smaller than or equal to the variation threshold of the preset data statistics index, wherein the variation threshold of the preset data statistics index is negative, and the second risk early warning information is used for indicating the risk that the operating condition of the target resource main body is greatly reduced in the future period.

In an alternative embodiment, the resource principal risk early warning device 80 includes:

the model training module is used for training to obtain a prediction model by the following modes:

constructing a training set, wherein the training set comprises a plurality of pieces of training data, and the training data comprises actual variation information of preset indexes of a sample resource main body in a history period and characteristic information of a plurality of history time points of the sample resource main body before the history period; generating time sequence characteristic data of the sample resource main body according to the characteristic information of the sample resource main body; inputting time sequence characteristic data of the sample resource main body into an initial prediction model, and predicting variation prediction information of preset indexes of the sample resource main body in a corresponding historical period; calculating loss according to the variation prediction information and the actual variation information of the preset indexes of the sample resource main body in the corresponding historical period, and updating model parameters of the prediction model according to the loss to obtain a trained prediction model.

In an alternative embodiment, the training set includes preset variable categories, different categories correspond to different variable intervals, and the training data includes the category of the actual variable of the preset index of the sample resource body in a history period. The prediction model is a classification model, and the variation prediction information of the preset index of the sample resource main body in the corresponding history period is a prediction category of the variation.

When the loss is calculated according to the variation prediction information and the actual variation information of the preset index of the sample resource main body in the corresponding history period, the model training module is further used for:

determining the weight coefficient of the category of the variable quantity according to the quantity of the training data of different categories in the training set; and calculating the weighted cross entropy loss according to the prediction category of the variation of the preset index of the sample resource main body in the corresponding history period, the category of the actual variation and the weight coefficient.

The apparatus provided in this embodiment of the present application may be specifically configured to perform the method provided in any of the foregoing method embodiments, and specific functions and technical effects that may be achieved are not described herein.

Fig. 9 is a schematic structural diagram of a cloud server according to an embodiment of the present application. As shown in fig. 9, the cloud server includes: a memory 901 and a processor 902. Memory 901 for storing computer-executable instructions and may be configured to store various other data to support operations on a cloud server. The processor 902 is communicatively connected to the memory 901, and is configured to execute computer-executable instructions stored in the memory 901, so as to implement the technical solution provided in any one of the above method embodiments, and the specific functions and the technical effects that can be implemented are similar, and are not repeated herein.

Optionally, as shown in fig. 9, the cloud server further includes: firewall 903, load balancer 904, communications component 905, power component 906, and other components. Only some components are schematically shown in fig. 9, which does not mean that the cloud server only includes the components shown in fig. 9.

The embodiment of the application further provides a computer readable storage medium, in which computer executable instructions are stored, and when the computer executable instructions are executed by a processor, the computer executable instructions are used to implement the scheme provided by any one of the method embodiments, and specific functions and technical effects that can be implemented are not described herein.

The embodiment of the application also provides a computer program product, which comprises: the computer program is stored in a readable storage medium, and the computer program can be read from the readable storage medium by at least one processor of the cloud server, so that the at least one processor executes the computer program to enable the cloud server to execute the scheme provided by any one of the method embodiments, and specific functions and technical effects that can be achieved are not repeated herein. The embodiment of the application provides a chip, which comprises: the processing module and the communication interface can execute the technical scheme of the cloud server in the embodiment of the method. Optionally, the chip further includes a storage module (e.g. a memory), where the storage module is configured to store the instructions, and the processing module is configured to execute the instructions stored in the storage module, and execution of the instructions stored in the storage module causes the processing module to execute the technical solution provided in any one of the foregoing method embodiments.

The memory may be an object store (Object Storage Service, OSS).

The memory may be implemented by any type or combination of volatile or nonvolatile memory devices such as Static Random Access Memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic or optical disk.

The communication component is configured to facilitate wired or wireless communication between the device in which the communication component is located and other devices. The device where the communication component is located may access a wireless network based on a communication standard, such as a mobile hotspot (WiFi), a mobile communication network of a second generation mobile communication system (2G), a third generation mobile communication system (3G), a fourth generation mobile communication system (4G)/Long Term Evolution (LTE), a fifth generation mobile communication system (5G), or a combination thereof. In one exemplary embodiment, the communication component receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, the communication component further includes a Near Field Communication (NFC) module to facilitate short range communications. For example, the NFC module may be implemented based on Radio Frequency Identification (RFID) technology, infrared data association (IrDA) technology, ultra Wideband (UWB) technology, bluetooth (BT) technology, and other technologies.

The power supply component provides power for various components of equipment where the power supply component is located. The power components may include a power management system, one or more power sources, and other components associated with generating, managing, and distributing power for the devices in which the power components are located.

It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, magnetic disk storage, compact disk read-only memory (CD-ROM), optical storage, and the like) having computer-usable program code embodied therein.

The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

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

The memory may include volatile memory in a computer-readable medium, random Access Memory (RAM) and/or nonvolatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of computer-readable media.

Computer readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of storage media for a computer 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 disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. Computer-readable media, as defined herein, does not include transitory computer-readable media (transmission media), such as modulated data signals and carrier waves.

It should be noted that, the user information (including but not limited to user equipment information, user attribute information, etc.) and the data (including but not limited to data for analysis, stored data, presented data, etc.) referred to in the present application are information and data authorized by the user or fully authorized by each party, and the collection, use and processing of the related data need to comply with related laws and regulations and standards, and provide corresponding operation entries for the user to select authorization or rejection.

In addition, in some of the flows described in the above embodiments and the drawings, a plurality of operations appearing in a particular order are included, but it should be clearly understood that the operations may be performed out of order or performed in parallel in the order in which they appear herein, merely for distinguishing between the various operations, and the sequence number itself does not represent any order of execution. In addition, the flows may include more or fewer operations, and the operations may be performed sequentially or in parallel. It should be noted that, the descriptions of "first" and "second" herein are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, and are not limited to the "first" and the "second" being different types. The meaning of "a plurality of" is two or more, unless specifically defined otherwise.

It should be noted that, the user information (including but not limited to user equipment information, user personal information, etc.) and the data (including but not limited to data for analysis, stored data, presented data, etc.) referred to in the present application are information and data authorized by the user or fully authorized by each party, and the collection, use and processing of the related data need to comply with the related laws and regulations and standards, and provide corresponding operation entries for the user to select authorization or rejection.

Other embodiments of the present application will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the application following, in general, the principles of the application and including such departures from the present disclosure as come within known or customary practice within the art to which the application pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the application being indicated by the following claims.

It is to be understood that the present application is not limited to the precise arrangements and instrumentalities shown in the drawings, which have been described above, and that various modifications and changes may be effected without departing from the scope thereof. The scope of the application is limited only by the appended claims.

Claims

1. The resource main body risk early warning method is characterized by being applied to a server deployed at a cloud end and comprising the following steps of:

2. The resource main body risk early warning method is characterized by comprising the following steps of:

3. The method according to claim 2, wherein the obtaining feature information of the target resource body to be predicted at a plurality of history time points includes:

acquiring at least one piece of characteristic information of a target resource main body to be predicted at a plurality of historical time points, wherein the characteristic information comprises the following characteristics:

presetting indexes, time characteristics and resource value parameters.

4. The method of claim 2, wherein generating time series characteristic data of the target resource body according to the characteristic information of the plurality of historical time points comprises:

and arranging the characteristic information of the plurality of historical time points according to time sequence to form time sequence characteristic data.

5. The method according to claim 2, wherein predicting the amount of change in the target resource body preset index in the future period based on the time series characteristic data comprises:

generating time sequence position features according to the sequence of each item of feature information in the time sequence feature data and the position in the sequence;

the time sequence position features and the time sequence feature data are fused to obtain fusion features, and the fusion features are input into a coding module of a prediction model for coding to obtain coding features;

and decoding the coding features through a decoding module of the prediction model to obtain the variation of the preset index of the target resource main body in a future period.

6. The method according to any one of claims 2 to 5, wherein the preset index is an evaluation level, and the predicting, according to the time series characteristic data, the change amount of the preset index in the future period of time, includes:

And predicting the change amount of the evaluation level of the target resource main body in a future period according to the time sequence characteristic data.

7. The method according to claim 6, wherein the performing risk early warning processing when the amount of change in the preset index of the target resource body in the future period satisfies an early warning condition includes:

and outputting first risk early warning information when the change amount of the evaluation level of the target resource main body in the future period is smaller than or equal to the change amount of a preset level, wherein the change amount of the preset level is a negative number.

8. The method according to any one of claims 2 to 5, wherein the preset index is a data statistics index, and the predicting, according to the time sequence feature data, a variation of the preset index in a future period of time, by the target resource body includes:

9. The method according to claim 8, wherein the performing risk early warning processing when the target resource body satisfies an early warning condition in a variation of a preset index in a future period of time includes:

And outputting second risk early warning information when the variation of the preset data statistics index of the target resource main body in a future period is smaller than or equal to a variation threshold of the preset data statistics index, wherein the variation threshold of the preset data statistics index is a negative number resource main body.

10. The method of claim 5, wherein the predictive model is trained by:

constructing a training set, wherein the training set comprises a plurality of pieces of training data, and the training data comprises actual variation information of a preset index of a sample resource main body in a history period and characteristic information of a plurality of history time points of the sample resource main body before the history period;

generating time sequence characteristic data of the sample resource main body according to the characteristic information of the sample resource main body;

inputting the time sequence characteristic data of the sample resource main body into an initial prediction model, and predicting the variation prediction information of a preset index of the sample resource main body in a corresponding history period;

calculating loss according to the variation prediction information and the actual variation information of the preset indexes of the sample resource main body in the corresponding historical period, and updating model parameters of the prediction model according to the loss to obtain a trained prediction model.

11. The method of claim 10, wherein the training set comprises categories of pre-set variation, different categories corresponding to different variation intervals,

the training data comprises the category of the actual variation of the preset index of the sample resource main body in a historical period;

the prediction model is a classification model, the variation prediction information of the preset index of the sample resource main body in the corresponding history period is a prediction category of the variation,

calculating the loss according to the variation prediction information and the actual variation information of the preset index of the sample resource main body in the corresponding history period, including:

determining the weight coefficient of the category of the variable quantity according to the quantity of the training data of different categories in the training set;

and calculating weighted cross entropy loss according to the prediction category of the variation of the preset index of the sample resource main body in the corresponding history period, the category of the actual variation and the weight coefficient.

12. The resource main body risk early warning method is characterized by being applied to a locally deployed server and comprising the following steps of:

13. A server, comprising: a processor, and a memory communicatively coupled to the processor;

the memory stores computer-executable instructions;

the processor executes computer-executable instructions stored in the memory to implement the method of any one of claims 1-12.

14. A computer readable storage medium having stored therein computer executable instructions which when executed by a processor are adapted to carry out the method of any one of claims 1-12.