Enterprise early warning model training method and device
By determining multiple optional parameter values based on the initial parameter interval set by experts in the enterprise early warning model, a parameter combination solution is solved, and the problem of insufficient interpretability and timeliness of the existing model is improved, and the accuracy and explanatory model are improved. and reduces the frequency of expert adjustments.
Patent Information
- Application Number
- CN202410952889.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-16
- Publication Date
- 2024-11-15
AI Technical Summary
The existing enterprise early warning model has problems such as low interpretability and poor timeliness in the financial field, especially the expert rule scoring model needs to be frequently adjusted to adapt to changes in the economic environment, which has caused a large amount of expert labor costs.
By obtaining the indicator dimension values of multiple sample enterprises in different time periods, determining multiple optional parameter values based on the initial parameter interval set by experts, forming different parameter combination schemes, training the enterprise early warning model, and making it in the initial parameters set by experts Improve accuracy and interpretability within the interval.
It improves the accuracy and interpretability of the enterprise early warning model, so that the parameter changes in the model within different periods are obvious, helping experts adjust the initial parameter value of the next period, and reducing the expert's labor costs.
Smart Images

Figure CN118966524A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of machine learning technology, and in particular to a training method and device for an enterprise early warning model. Background Art
[0002] With the development of machine learning and big data, machine learning has a powerful ability to mine hidden information in massive data. More and more companies are using machine learning to train enterprise early warning models. In the financial field, the interpretability of early warning results is crucial, but the model parameter values learned by machine learning algorithms are often not interpretable.
[0003] Among the existing enterprise early warning models, there are also models that use parameter values set by experts to score enterprises. The model parameter values of the expert rule scoring model are often highly interpretable, but have low timeliness and need to be dynamically adjusted according to different economic environments, which consumes expert manpower costs.
[0004] Therefore, how to better train enterprise early warning models needs further research. Summary of the invention
[0005] The present application provides a method and device for training an enterprise early warning model, which is used to make the parameters of each indicator dimension of the trained enterprise early warning model within the initial parameter range set by experts, thereby improving the accuracy and interpretability of the model.
[0006] In a first aspect, an embodiment of the present application provides a training method for an enterprise early warning model, the method comprising: obtaining first indicator values and first enterprise operating conditions of multiple first sample enterprises under each indicator dimension within a first time period; setting an initial parameter range for any parameter under any indicator dimension based on experts, and determining multiple first optional parameter values under the indicator dimension; combining multiple first optional parameter values under each indicator dimension to form N different first parameter combination schemes; any parameter under each indicator dimension under each first parameter combination scheme corresponds to an optional parameter value; for each sample enterprise, determining N first predicted operating conditions of the sample enterprise under the N different first parameter combination schemes according to the first indicator value of the sample enterprise under each indicator dimension; determining a parameter combination scheme that conforms to the first time period based on the N first predicted operating conditions of the multiple first sample enterprises and the first enterprise operating conditions of the multiple first sample enterprises; and determining an enterprise early warning model based on the parameter combination scheme for the first time period.
[0007] By adopting the above method, the enterprise early warning model is trained based on multiple first sample enterprises in the first period and the initial parameter range of any parameter under each indicator dimension set by experts. The first optional parameter value of the trained enterprise early warning model conforms to the first period and is within the initial parameter range set by experts, thereby improving the accuracy and interpretability of each parameter in the enterprise early warning model.
[0008] In a possible implementation, based on an initial parameter interval set by an expert for any parameter under any indicator dimension, a plurality of first optional parameter values under the indicator dimension are determined, including: based on an initial parameter value and an initial parameter interval set by an expert for any parameter under any indicator dimension, taking the initial parameter value as a reference, a plurality of first optional parameter values under the indicator dimension that are located in the initial parameter interval are determined.
[0009] In a possible implementation, taking the initial parameter value as a reference, determining a plurality of first optional parameter values located in the initial parameter interval under the indicator dimension, including: based on a parameter step set by an expert for the indicator dimension, taking the initial parameter value as a reference, and determining a plurality of first optional parameter values located in the initial parameter interval according to the parameter step.
[0010] By adopting the above method, based on the parameter step set by the expert for any parameter under each indicator dimension, multiple first optional parameter values in the initial parameter interval can be determined according to the initial parameter interval and the set parameter step. In this way, multiple first optional parameter values are evenly distributed in the initial parameter interval, so that the accuracy of each parameter in the trained enterprise early warning model is higher.
[0011] In a possible implementation, the initial parameter interval set by the expert for any parameter under any indicator dimension is determined by the initial parameter value and parameter threshold set by the expert for any parameter under any indicator dimension.
[0012] In a possible implementation, any parameter under any indicator dimension includes a weight parameter representing the weight of the indicator dimension, and / or a parameter boundary value representing the operating status of the indicator dimension under different indicator values.
[0013] In a possible implementation, after determining the enterprise early warning model based on the parameter combination scheme of the first time period, it also includes: obtaining the second indicator values and second enterprise operating conditions of multiple second sample enterprises under each indicator dimension in the second time period; using the first optional parameter value of each parameter under each indicator dimension in the parameter combination scheme of the first time period as the initial parameter value of each parameter under each indicator dimension; based on any parameter under any indicator dimension, determine multiple second optional parameter values under the indicator dimension according to the initial parameter value of the parameter and the initial parameter range of the parameter; combine the multiple second optional parameter values under each indicator dimension to form N different second parameter combination schemes; for each sample enterprise, determine N second predicted operating conditions of the sample enterprise under the N different second parameter combination schemes according to the second indicator value of the sample enterprise under each indicator dimension; determine the parameter combination scheme that meets the second time period based on the N second predicted operating conditions of the multiple second sample enterprises and the second enterprise operating conditions of the multiple second sample enterprises; and update the enterprise early warning model based on the parameter combination scheme of the second time period.
[0014] By adopting the above method, the first optional parameter value in the parameter combination scheme of the first period is used as the initial parameter value of each parameter of the enterprise early warning model of the second period, so that the parameters of the enterprise early warning model that meets the second period after training are more accurate.
[0015] In a possible implementation, the method further includes: displaying parameter values of each parameter of the enterprise early warning model under each indicator dimension in different time periods.
[0016] By adopting the above method, by displaying the changing trend graph of each parameter of the enterprise early warning model in different time periods, industry experts can understand the changing trend of each parameter, which helps experts to set the initial parameter value for the enterprise early warning model in the next time period as a reference.
[0017] In a second aspect, an embodiment of the present application provides a training device for an enterprise early warning model, the device comprising: an acquisition module, used to obtain first indicator values and first enterprise operating conditions of multiple first sample enterprises under each indicator dimension within a first time period; a processing module, used to set an initial parameter interval for any parameter under any indicator dimension based on experts, and determine multiple first optional parameter values under the indicator dimension; and used to combine multiple first optional parameter values under each indicator dimension to form N different first parameter combination schemes; any parameter under each indicator dimension under each first parameter combination scheme corresponds to an optional parameter value; the processing module is also used to, for each sample enterprise, determine N first predicted operating conditions of the sample enterprise under the N different first parameter combination schemes according to the first indicator value of the sample enterprise under each indicator dimension; a determination module, used to determine the parameter combination scheme that conforms to the first time period based on the N first predicted operating conditions of the multiple first sample enterprises and the first enterprise operating conditions of the multiple first sample enterprises; and used to determine the enterprise early warning model based on the parameter combination scheme for the first time period.
[0018] In one possible implementation, the processing module is specifically used to determine, based on the initial parameter value and initial parameter interval set by an expert for any parameter under any indicator dimension, a plurality of first optional parameter values located in the initial parameter interval under the indicator dimension, with the initial parameter value as a reference.
[0019] In a possible implementation, the processing module is further specifically used to determine, based on the parameter step set by the expert for the indicator dimension and taking the initial parameter value as a reference, a plurality of first optional parameter values located in the initial parameter interval according to the parameter step.
[0020] In a possible implementation, the initial parameter interval set by the expert for any parameter under any indicator dimension is determined by the initial parameter value and parameter threshold set by the expert for any parameter under any indicator dimension.
[0021] In a possible implementation, any parameter under any indicator dimension includes a weight parameter representing the weight of the indicator dimension, and / or a parameter boundary value representing the operating status of the indicator dimension under different indicator values.
[0022] In a possible implementation, the acquisition module is also used to obtain the second indicator values and second enterprise operating conditions of multiple second sample enterprises under each indicator dimension in the second time period; the processing module is also used to use the first optional parameter value of each parameter under each indicator dimension in the parameter combination scheme of the first time period as the initial parameter value of each parameter under each indicator dimension; and based on any parameter under any indicator dimension, determine multiple second optional parameter values under the indicator dimension according to the initial parameter value of the parameter and the initial parameter range of the parameter; combine the multiple second optional parameter values under each indicator dimension to form N different second parameter combination schemes; for each sample enterprise, determine the N second predicted operating conditions of the sample enterprise under the N different second parameter combination schemes according to the second indicator value of the sample enterprise under each indicator dimension; the determination module is also used to determine the parameter combination scheme that meets the second time period based on the N second predicted operating conditions of the multiple second sample enterprises and the second enterprise operating conditions of the multiple second sample enterprises; update the enterprise early warning model based on the parameter combination scheme for the second time period.
[0023] In a possible implementation, the device further includes a display module for displaying parameter values of each parameter of the enterprise early warning model under each indicator dimension in different time periods.
[0024] In a third aspect, an embodiment of the present application further provides a training device for an enterprise early warning model, the device comprising a memory and a processor, the memory being used to store computer programs or instructions; the processor being used to call the computer programs or instructions stored in the memory to execute a method as in any possible implementation of the first aspect.
[0025] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, in which instructions are stored. When a computer reads and executes the instructions, the computer executes a method in any possible implementation of the first aspect.
[0026] In a fifth aspect, an embodiment of the present application provides a computer program product, wherein instructions are stored in the computer program product. When a computer reads and executes the instructions, the computer executes the method in any possible implementation of the first aspect above. BRIEF DESCRIPTION OF THE DRAWINGS
[0027] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0028] Figure 1 A flowchart corresponding to a training method for an enterprise early warning model provided in an embodiment of the present application;
[0029] Figure 2 A schematic diagram of the relationship between the score corresponding to the predicted operating status and the operating status level provided in the embodiment of the present application;
[0030] Figure 3 A schematic diagram of the internal modules of a training device 3000 for an enterprise early warning model provided in an embodiment of the present application;
[0031] Figure 4 A structural schematic diagram of a training device 4000 for an enterprise early warning model provided in an embodiment of the present application. DETAILED DESCRIPTION
[0032] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.
[0033] The terms "first" and "second" in the specification and claims of this application and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the term "comprising" and any variation thereof are intended to cover non-exclusive protection. For example, a process, method, system, product or device comprising a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units inherent to these processes, methods, products or devices. "Multiple" in this application may mean at least two, for example, two, three or more, which is not limited in the embodiments of the present invention.
[0034] The following is a description of exemplary embodiments of the present application in conjunction with the accompanying drawings, including various details of the embodiments of the present application to facilitate understanding, which should be considered as merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope of the present application. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted in the following description. It should be noted that in the embodiments of the present application, certain software, components, models and other existing solutions in the industry may be mentioned, which should be considered as exemplary, and their purpose is only to illustrate the feasibility of the implementation of the technical solution of the present application, but it does not mean that the applicant has or will necessarily use the solution.
[0035] In the technical solution of this application, the acquisition, transmission, storage, and use of data are in compliance with the requirements of relevant national laws and regulations.
[0036] The embodiment of the present application provides a method for training an enterprise early warning model, which is used to make the parameters of each indicator dimension of the trained enterprise early warning model within the initial parameter range set by experts, thereby improving the accuracy and interpretability of the model.
[0037] Figure 1 A flowchart corresponding to a training method for an enterprise early warning model provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the process includes the following steps:
[0038] Step 101, obtaining first indicator values and first enterprise operating conditions of multiple first sample enterprises under each indicator dimension within a first time period.
[0039] Specifically, multiple first sample enterprises can be enterprises in the same region, such as multiple sample enterprises in Heze City, Shandong Province; multiple sample enterprises can also be enterprises in the same region and the same field, such as multiple sample enterprises in the new energy vehicle industry in Heze City, Shandong Province. It should be noted that this application trains an enterprise early warning model based on the first indicator values of multiple sample enterprises and the operating conditions of the first enterprises of multiple sample enterprises, and then predicts the operating conditions of a certain enterprise based on the trained enterprise early warning model. Therefore, multiple first sample enterprises and the enterprises whose operating conditions the enterprise early warning model actually needs to predict need to have the same attributes, and multiple first sample enterprises also need to have the same attributes. The same attributes can be regions and industries. Multiple first sample enterprises can also have other identical attributes, and this application does not impose specific restrictions.
[0040] The first time period can be one month, three months, one year, or other time lengths. It can be understood that the first time period is the first time period adjacent to the acquisition of multiple first sample enterprise information, that is, if the time for obtaining the first indicator values of multiple first sample enterprises and the first enterprise operating conditions of multiple first sample enterprises is April 1, 2024, and the length of the first time period is one quarter, then the first time period is from January 1, 2024 to March 31, 2024.
[0041] Each indicator dimension may be a plurality of indicator dimensions set by experts, and the operating status of a certain enterprise may be predicted according to each indicator dimension, for example, each indicator dimension may be an enterprise non-performing rate, an enterprise employee number, and an enterprise operating income. Each indicator dimension may be other indicators, and this application does not make specific limitations.
[0042] The first indicator value under each indicator dimension is the corresponding value of each indicator dimension in the first time period. For example, the bad debt rate of Enterprise A in the first time period is 10%, the number of employees of Enterprise A in the first time period is 18, and the operating income of Enterprise A in the first time period is 800,000.
[0043] The first enterprise operating status refers to the operating status of the enterprise in the first period of time. The operating status can be used to indicate the risk level of the enterprise. If the operating status of the enterprise is divided into four levels, the operating status can be divided into healthy, slight risk, comparative risk, and severe risk. The operating status of the enterprise can also be divided into other levels, which is not limited in this application.
[0044] Step 102, based on the expert setting the initial parameter range for any parameter under any indicator dimension, determine multiple first optional parameter values under the indicator dimension; combine the multiple first optional parameter values under each indicator dimension to form N different first parameter combination schemes; any parameter under each indicator dimension under each first parameter combination scheme corresponds to an optional parameter value.
[0045] Exemplarily, the expert also needs to set an initial parameter interval for any parameter under each indicator dimension, and determine multiple first optional parameter values under each indicator dimension. Continuing with the above example, the enterprise early warning model has three indicator dimensions, namely, bad debt rate, number of employees and operating income. There are parameters under the three indicator dimensions. If there is one parameter under each of the three indicator dimensions, assuming that the parameter under the bad debt rate is parameter A, the parameter under the number of employees is parameter B, and the parameter under the operating income is parameter C, then set the initial parameter intervals for parameter A, parameter B and parameter C respectively. The initial parameter interval corresponding to parameter A can be [0.7, 0.9], the initial parameter interval corresponding to parameter B can be [0.5, 0.7], and the initial parameter interval corresponding to parameter C can be [0.5, 0.8].
[0046] Since the initial parameter interval corresponding to parameter A is [0.7, 0.9], the multiple first optional parameter values corresponding to parameter A of the defect rate can be any value between [0.7, 0.9], including 0.7 and 0.9; similarly, the multiple first optional parameter values corresponding to parameter B of the number of employees can be any value between [0.5, 0.7]; the multiple first optional parameter values corresponding to parameter C of operating income can be any value between [0.5, 0.8]. The number of multiple first optional parameter values under each indicator dimension can be different, that is, the number of first optional parameter values corresponding to parameter A of the defect rate and the number of first optional parameter values corresponding to parameters B of the number of employees and parameter C of operating income can be different.
[0047] If the first optional parameter value corresponding to parameter A that determines the defective rate is 3 values, namely 0.7, 0.8 and 0.9; the first optional parameter value corresponding to parameter B of the number of employees is 4 values, namely 0.5, 0.55, 0.6 and 0.7; the first optional parameter value corresponding to parameter C of the operating income is 4 values, namely 0.5, 0.6, 0.7 and 0.8, then multiple first optional parameter values under each indicator dimension are combined to form 3*4*4 different first parameter combination schemes, and the parameter under each indicator dimension under the 48 first parameter combination schemes corresponds to an optional parameter value.
[0048] It can be understood that there may be multiple parameters under each indicator dimension, and each of the multiple parameters has an initial parameter interval. According to the initial parameter interval of the multiple parameters under each indicator dimension, the optional parameter value of each parameter is selected, thereby forming N different first parameter combination schemes. For example, if the defect rate has parameters A and B; the number of employees has parameters C and D; and the operating income has parameters E and F, according to the initial parameter interval of each parameter, multiple first optional parameter values can be selected for each parameter. If the first optional parameter value corresponding to parameter A is 3 values, the first optional parameter value corresponding to parameter B is 4 values; the first optional parameter value corresponding to parameter C is 2 values, the first optional parameter value corresponding to parameter D is 5 values; the first optional parameter value corresponding to parameter E is 3 values, and the first optional parameter value corresponding to parameter F is 4 values; then the multiple first optional parameter values under each indicator dimension are combined to form 3*4*2*5*3*4 different first parameter combination schemes, and any parameter under each indicator dimension under the 1440 first parameter combination schemes corresponds to an optional parameter value.
[0049] Optionally, based on an initial parameter interval set by an expert for any parameter under any indicator dimension, multiple first optional parameter values under each indicator dimension are determined, including: based on the initial parameter value and initial parameter interval set by an expert for any parameter under any indicator dimension, taking the initial parameter value as a reference, multiple first optional parameter values located in the initial parameter interval under the indicator dimension are determined.
[0050] That is to say, the initial parameter interval can be set by setting the initial parameter value, taking the initial parameter value as a benchmark, determining the initial parameter interval, and then determining multiple first optional parameter values in the initial parameter interval under each indicator dimension. That is to say, taking parameter A under the defect rate as an example, the initial parameter interval of parameter A is set to [0.7, 0.9]. The initial parameter value can be set for parameter A first, such as 0.8, and 0.8 is used as a benchmark to determine the initial parameter interval to be [0.7, 0.9].
[0051] Optionally, the initial parameter interval set by the expert for any parameter under any indicator dimension is determined by the initial parameter value and parameter threshold set by the expert for any parameter under any indicator dimension.
[0052] That is to say, the initial parameter interval can be determined by the initial parameter value and parameter threshold set by the expert for any parameter under any indicator dimension. For example, to set the initial parameter interval of parameter A under the defect rate, the initial parameter value and parameter threshold can be set for parameter A first. If the initial parameter value of parameter A is set to 0.8, and the threshold of parameter A is set to +0.1, -0.1, then the initial parameter interval of parameter A is [0.7, 0.9]. If the threshold of parameter A is set to +0.1, -0.2, then the initial parameter interval of parameter A is [0.6, 0.9].
[0053] After determining the initial parameter interval of the parameter, multiple first optional parameters corresponding to the parameter can be determined, and the multiple first optional parameters are all within the initial parameter interval. There are many ways to select multiple first optional parameters within the initial parameter interval, such as selecting two values of the critical points of the initial parameter interval, or randomly selecting multiple parameter values within the initial parameter interval. Optionally, based on the initial parameter value, multiple first optional parameter values located in the initial parameter interval under the indicator dimension are determined. The multiple first optional parameter values located in the initial parameter interval can be determined based on the parameter step set by the expert for any parameter under the indicator dimension, based on the initial parameter value, and according to the parameter step.
[0054] For example, if the initial parameter value of parameter A under defect rate is 0.8, and the initial parameter interval of parameter A is set to [0.7, 0.9], multiple first optional parameter values in the initial parameter interval can be determined based on the parameter step size set by the expert for parameter A under the indicator dimension. If the parameter step size of parameter A is 0.05, the multiple first optional parameter values are 0.7, 0.75, 0.8, 0.85, and 0.9.
[0055] Optionally, any parameter under any indicator dimension in the embodiment of the present application may include a weight parameter characterizing the weight of the indicator dimension, and / or a parameter boundary value characterizing the operating status of the indicator dimension under different indicator values.
[0056] Continuing with the above example, the enterprise early warning model corresponds to three indicator dimensions, namely, the bad rate, the number of employees and the operating income. The parameters of the bad rate, the number of employees and the operating income may include weight parameters that characterize the weights and parameter demarcation values that characterize the operating conditions of the indicator dimensions under different indicator values. The weight parameters of each indicator dimension can be set by an expert. The higher the weight, the more important the indicator dimension is for judging the operating conditions of the enterprise. For example, the weight of the bad rate can be set to 70%, the weight of the number of employees can be set to 70%, and the weight of the operating income can be set to 80%. The weight parameters under each indicator dimension can also be set to other values, which are not specifically limited in this application.
[0057] In addition, when the bad rate of an enterprise is less than or equal to 10%, it can be set to 20 points; when the bad rate is greater than 10% and less than or equal to 20%, it can be set to 40 points; when the bad rate is greater than 20%, it can be set to 70 points. Similarly, the number of employees in the enterprise can also be set to different scores when the number of employees is different. For example, when the number of employees is less than or equal to 20, it can be set to 70 points; when the number of employees is greater than 20 and less than or equal to 100, it can be set to 40 points; when the number of employees is more than 100, it can be set to 10 points. Operating income is similar to the bad rate and the number of employees. Different scores are set according to different ranges of operating income, which will not be elaborated here. It can be understood that the scores corresponding to each indicator dimension in the parameter demarcation value interval can represent the operating status of the enterprise.
[0058] In general, taking the bad rate as an example, the parameters under the bad rate can include weights. If the initial parameter interval of the weight under the bad rate is [0.7, 0.9], the initial parameter value is 0.8, and the parameter step is 0.05, then the weight under the bad rate can have 5 optional parameter values. The parameters under the bad rate can also include parameter demarcation values. If there are two demarcation values for the bad rate, the initial parameter interval of demarcation value 1 is [5%, 15%], the initial parameter value is 5%, and the parameter step is 5%, then the bad rate demarcation value 1 can have 3 optional parameter values, the initial parameter interval of demarcation value 2 is [20%, 25%], the initial parameter value is 20%, and the parameter step is 5%, then demarcation value 2 can have 2 optional parameter values, then the first optional parameter value under the bad rate is 5*3*2, 30 combinations. Similarly, calculate the number of combinations of the first optional parameter value under the number of employees and the number of combinations of the first optional parameter value under the operating income. If the first optional parameter of the number of employees has 10 combinations and the first optional parameter of the number of business operators has 40 combinations, then multiple first optional parameter values under each indicator dimension are combined to form 30*10*40 different first parameter combination schemes.
[0059] Step 103 , for each sample enterprise, according to the first indicator value of the sample enterprise under each indicator dimension, determine N first predicted operating conditions of the sample enterprise under N different first parameter combination schemes.
[0060] For example, taking the first sample enterprise among multiple first sample enterprises as an example, according to the first indicator value of the first sample enterprise under each indicator dimension, N first predicted operating conditions of the first sample enterprise under N different first parameter combination schemes can be determined.
[0061] The first indicator value of the first sample enterprise under each indicator dimension can be obtained from the Internet. This application does not specifically limit the method of obtaining the first indicator value of the sample enterprise. Assume that the bad rate of the first sample enterprise in the first period is 8%, the number of employees is 15, and the operating income is 300,000. If one of the first parameter combinations of N different first parameter combinations is: the weight of the bad rate is 0.75, the first demarcation value of the bad rate is 10%, and the second demarcation value is 20%; the weight of the number of employees is 0.7, the first demarcation value of the number of employees is 15, and the second demarcation value of the number of employees is 100; the weight of the operating income is 0.85, the first demarcation value of the operating income is 500,000, and the second demarcation value is 5 million. And the score corresponding to the interval where the bad rate is 8% is 20 points, the score corresponding to the interval where the number of employees is 15 is 70 points, and the score corresponding to the interval where the operating income is 300,000 is 80 points, combined with the weights corresponding to each indicator dimension, the first predicted operating status of the first sample enterprise under this parameter combination can be calculated. The first predicted operating status is calculated based on the weight corresponding to each indicator dimension * the score corresponding to the first indicator value under each indicator dimension. For example, the first predicted operating status of the first sample enterprise under the first parameter combination is 0.75*20+0.7*70+0.85*80=132. The relationship between the score corresponding to the predicted operating status and the operating status level provided in the embodiment of the present application is shown in FIG. Figure 2 As shown. According to the score corresponding to the first predicted operating status, the level of the first predicted operating status can be determined. Similarly, according to N different first parameter combination schemes, N first predicted operating statuses can be obtained.
[0062] Step 104, based on the N first predicted operating conditions of multiple first sample enterprises and the first enterprise operating conditions of multiple first sample enterprises, determine a parameter combination scheme that meets the first time period; determine an enterprise early warning model based on the parameter combination scheme of the first time period.
[0063] Specifically, taking the first sample enterprise among the multiple first sample enterprises as an example, after obtaining N first predicted operating conditions according to N different first parameter combination schemes, the N first predicted operating conditions are compared with the first enterprise operating conditions of the first sample enterprise, and the first parameter combination scheme that meets the first enterprise operating condition is screened out. For example, the first sample enterprise obtains N first predicted operating conditions according to N different first parameter combination schemes, assuming that N is 120, then 120 first predicted operating conditions are obtained. If 20 of the 120 first predicted operating conditions are healthy, and the first enterprise operating condition of the first sample enterprise is healthy, then 20 first parameter combination schemes can be determined based on the first sample enterprise.
[0064] According to the N first predicted operating conditions of each sample enterprise among the multiple first sample enterprises and the first enterprise operating conditions of each sample enterprise, the first parameter combination scheme determined by each sample enterprise can be determined. According to the intersection of the first parameter combination schemes determined by each sample enterprise, one or several first parameter combination schemes can be determined. If a first parameter combination scheme is determined, the enterprise early warning model that conforms to the first time period is determined according to the determined first parameter combination scheme. If several first parameter combination schemes are determined, a first parameter combination scheme is further determined according to the indicators of the enterprise early warning model corresponding to the several first parameter combination schemes. Specifically, the indicators may include indicators such as accuracy, recall rate, F1 score or loss function value. This application does not specifically limit the indicators for evaluating the enterprise early warning model.
[0065] Therefore, based on the first indicator values of multiple first sample enterprises under each indicator dimension in the first time period and the operating status of the first enterprises, the determined enterprise early warning model is an enterprise early warning model that conforms to the first time period. Based on this enterprise early warning model, the operating status of the enterprise in the first time period can be predicted.
[0066] As time changes, the parameters of the enterprise early warning model can also be updated. Specifically, after determining the enterprise early warning model based on the parameter combination scheme of the first time period, the second indicator values and the second enterprise operating conditions of multiple second sample enterprises under each indicator dimension in the second time period can also be obtained. The second time period is located after the first time period, and the second time period is also before the current time. The first optional parameter value of each parameter under each indicator dimension in the parameter combination scheme of the first time period is used as the initial parameter value of each parameter under each indicator dimension of the enterprise early warning model in the second time period. Taking the defective rate indicator dimension as an example, if after training in the first time period, it is determined that one of the parameter weight values of the defective rate in the first parameter combination is 0.7, then 0.7 is used as the initial value of the weight under the defective rate when training the enterprise early warning model in the second time period.
[0067] Based on any parameter under any indicator dimension, multiple second optional parameter values under the indicator dimension are determined according to the initial parameter value of the parameter and the initial parameter interval of the parameter, wherein the initial parameter interval can be the initial parameter interval set by the expert for the parameter in the first period, or can be determined according to the initial parameter value of the parameter in the second period and the parameter threshold set by the expert. Multiple second optional parameter values under each indicator dimension are combined to form N different second parameter combination schemes, and for each second sample enterprise, according to the second indicator value of the sample enterprise under each indicator dimension, N second predicted operating conditions of each sample enterprise under N different second parameter combination schemes are determined; based on the N second predicted operating conditions of multiple second sample enterprises and the second enterprise operating conditions of multiple second sample enterprises, a parameter combination scheme that meets the second period is determined, and the enterprise early warning model is updated based on the parameter combination scheme for the second period.
[0068] By adopting the above method, an enterprise early warning model that conforms to different time periods can be trained according to the indicator values of sample enterprises in different time periods under each indicator dimension and the enterprise operation status, and then any parameter under each dimension indicator of the enterprise early warning model trained for different time periods will also be updated. Optionally, the training device of the enterprise early warning model in the embodiment of the present application can also display the parameter values of each parameter under each indicator dimension in different time periods. By displaying the change trend chart of each parameter of the enterprise early warning model in different time periods, industry experts can understand the change trend of each parameter, which helps experts to set the initial parameter value of the enterprise early warning model for the next time period as a reference.
[0069] Figure 3 A schematic diagram of the internal modules of a training device 3000 for an enterprise early warning model provided in an embodiment of the present application is shown in FIG. Figure 3 As shown, the device may include: an acquisition module 301, a processing module 302, a determination module 303 and a display module 304, and optionally, a storage module, the storage module is used to store computer instructions or programs, and the processing module 302 can call the computer instructions or programs in the storage module.
[0070] Among them, the acquisition module 301 is used to obtain the first indicator values and first enterprise operating conditions of multiple first sample enterprises under each indicator dimension within the first time period; the processing module 302 is used to set the initial parameter range for any parameter under any indicator dimension based on experts, and determine multiple first optional parameter values under the indicator dimension; and is used to combine multiple first optional parameter values under each indicator dimension to form N different first parameter combination schemes; any parameter under each indicator dimension under each first parameter combination scheme corresponds to an optional parameter value; the processing module 302 is also used to, for each sample enterprise, determine the N first predicted operating conditions of the sample enterprise under the N different first parameter combination schemes according to the first indicator value of the sample enterprise under each indicator dimension; the determination module 303 is used to determine the parameter combination scheme that meets the first time period based on the N first predicted operating conditions of the multiple first sample enterprises and the first enterprise operating conditions of the multiple first sample enterprises; and is used to determine the enterprise early warning model based on the parameter combination scheme for the first time period.
[0071] In a possible implementation, the processing module 302 is specifically used to determine, based on the initial parameter value and initial parameter interval set by an expert for any parameter under any indicator dimension, a plurality of first optional parameter values located in the initial parameter interval under the indicator dimension with the initial parameter value as a reference.
[0072] In a possible implementation, the processing module 302 is further specifically used to determine, based on the parameter step set by the expert for the indicator dimension and taking the initial parameter value as a reference, a plurality of first optional parameter values located in the initial parameter interval according to the parameter step.
[0073] In a possible implementation, the initial parameter interval set by the expert for any parameter under any indicator dimension is determined by the initial parameter value and parameter threshold set by the expert for any parameter under any indicator dimension.
[0074] In a possible implementation, any parameter under any indicator dimension includes a weight parameter representing the weight of the indicator dimension, and / or a parameter boundary value representing the operating status of the indicator dimension under different indicator values.
[0075] In a possible implementation, the acquisition module 301 is further used to obtain the second indicator values and second enterprise operating conditions of multiple second sample enterprises under each indicator dimension in the second time period; the processing module 302 is further used to use the first optional parameter value of each parameter under each indicator dimension in the parameter combination scheme of the first time period as the initial parameter value of each parameter under each indicator dimension; and based on any parameter under any indicator dimension, determine multiple second optional parameter values under the indicator dimension according to the initial parameter value of the parameter and the initial parameter range of the parameter; and combine the multiple second optional parameter values under each indicator dimension to form N different second parameter combination schemes; and for each sample enterprise, determine the N second predicted operating conditions of the sample enterprise under the N different second parameter combination schemes according to the second indicator value of the sample enterprise under each indicator dimension; the determination module 303 is further used to determine the parameter combination scheme that meets the second time period based on the N second predicted operating conditions of the multiple second sample enterprises and the second enterprise operating conditions of the multiple second sample enterprises; and update the enterprise early warning model based on the parameter combination scheme for the second time period.
[0076] In a possible implementation, the device further includes a display module 304 for displaying parameter values of each parameter of the enterprise early warning model under each indicator dimension in different time periods.
[0077] Figure 4 A structural diagram of a training device 4000 for an enterprise early warning model provided in an embodiment of the present application. Figure 4 As shown, it includes at least one processor 401 and a memory 402 connected to the at least one processor 401. The specific connection medium between the processor 401 and the memory 402 is not limited in the embodiment of the present application. Figure 4 For example, the processor 401 and the memory 402 are connected via a bus. The bus can be divided into an address bus, a data bus, a control bus, etc.
[0078] In the embodiment of the present application, the memory 402 stores instructions that can be executed by at least one processor 401. The at least one processor 401 can implement the steps of the above-mentioned enterprise early warning model training method by executing the instructions stored in the memory 402.
[0079] The processor 401 is the control center of the computer device, and can use various interfaces and lines to connect various parts of the computer device, and perform resource settings by running or executing instructions stored in the memory 402 and calling data stored in the memory 402. Optionally, the processor 401 may include one or more processing units, and the processor 401 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 401. In some embodiments, the processor 401 and the memory 402 may be implemented on the same chip, and in some embodiments, they may also be implemented separately on independent chips.
[0080] Processor 401 can be a general-purpose processor, such as a central processing unit (CPU), a digital signal processor, an application-specific integrated circuit (ASIC), a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, and can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiments of the present application. A general-purpose processor can be a microprocessor or any conventional processor, etc. The steps of the method disclosed in the embodiments of the present application can be directly embodied as a hardware processor for execution, or can be executed by a combination of hardware and software modules in the processor.
[0081] The memory 402 is a non-volatile computer-readable storage medium that can be used to store non-volatile software programs, non-volatile computer executable programs and modules. The memory 402 may include at least one type of storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory, a random access memory (Random Access Memory, RAM), a static random access memory (Static Random Access Memory, SRAM), a programmable read-only memory (Programmable Read Only Memory, PROM), a read-only memory (Read Only Memory, ROM), an electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, EEPROM), a magnetic memory, a disk, an optical disk, etc. The memory 402 is any other medium that can be used to carry or store a desired program code in the form of an instruction or data structure and can be accessed by a computer, but is not limited thereto. The memory 402 in the embodiment of the present application can also be a circuit or any other device that can realize a storage function, for storing program instructions and / or data.
[0082] 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.
[0083] 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 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 produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0084] 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.
[0085] 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.
[0086] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.
Claims
1. A training method for an enterprise early warning model, It is characterized in that The method comprises: Obtain first indicator values and first enterprise operating conditions of multiple first sample enterprises under each indicator dimension within a first period of time; Based on the expert setting of an initial parameter interval for any parameter under any indicator dimension, multiple first optional parameter values under the indicator dimension are determined; multiple first optional parameter values under each indicator dimension are combined to form N different first parameter combination schemes; any parameter under each indicator dimension under each first parameter combination scheme corresponds to an optional parameter value; For each sample enterprise, according to the first indicator value of the sample enterprise under each indicator dimension, determine N first predicted operating conditions of the sample enterprise under the N different first parameter combination schemes; Based on the N first predicted operating conditions of the multiple first sample enterprises and the first enterprise operating conditions of the multiple first sample enterprises, a parameter combination scheme that conforms to the first time period is determined; and an enterprise early warning model is determined based on the parameter combination scheme for the first time period.
2. The method according to claim 1, It is characterized in that Based on the expert setting an initial parameter interval for any parameter under any indicator dimension, a plurality of first optional parameter values under the indicator dimension are determined, including: Based on the initial parameter value and initial parameter interval set by the expert for any parameter under any indicator dimension, and taking the initial parameter value as a reference, a plurality of first optional parameter values located in the initial parameter interval under the indicator dimension are determined.
3. The method according to claim 2, It is characterized in that Based on the initial parameter value, a plurality of first optional parameter values located in the initial parameter interval under the indicator dimension are determined, including: Based on the parameter step set by the expert for the indicator dimension and taking the initial parameter value as a reference, a plurality of first optional parameter values located in the initial parameter interval are determined according to the parameter step.
4. The method according to claim 2, It is characterized in that The initial parameter range set by the expert for any parameter under any indicator dimension is determined by the initial parameter value and parameter threshold set by the expert for any parameter under any indicator dimension.
5. The method according to any one of claims 1 to 4, It is characterized in that Any parameter under any indicator dimension includes a weight parameter representing the weight of the indicator dimension, and / or a parameter boundary value representing the operating status of the indicator dimension under different indicator values.
6. The method according to claim 5, It is characterized in that After determining the enterprise early warning model based on the parameter combination scheme of the first period, the method further includes: Obtain second indicator values and second enterprise operating conditions of multiple second sample enterprises under each indicator dimension in a second period; Using the first optional parameter value of each parameter under each indicator dimension in the parameter combination scheme of the first time period as the initial parameter value of each parameter under each indicator dimension; Based on any parameter under any indicator dimension, multiple second optional parameter values under the indicator dimension are determined according to the initial parameter value of the parameter and the initial parameter range of the parameter; multiple second optional parameter values under each indicator dimension are combined to form N different second parameter combination schemes; For each sample enterprise, according to the second indicator value of the sample enterprise under each indicator dimension, determine N second predicted operating conditions of the sample enterprise under the N different second parameter combination schemes; Based on the N second predicted operating conditions of the multiple second sample enterprises and the second enterprise operating conditions of the multiple second sample enterprises, determine a parameter combination scheme that meets the second time period; and update the enterprise early warning model based on the parameter combination scheme of the second time period.
7. The method according to claim 6, It is characterized in that The method further comprises: Display the parameter values of each parameter of the enterprise early warning model under each indicator dimension in different time periods.
8. A training device for an enterprise early warning model, It is characterized in that include: An acquisition module, used to acquire first indicator values and first enterprise operation status of multiple first sample enterprises under each indicator dimension within a first period of time; A processing module, configured to set an initial parameter interval for any parameter under any indicator dimension based on an expert, determine a plurality of first optional parameter values under the indicator dimension; combine the plurality of first optional parameter values under each indicator dimension to form N different first parameter combination schemes; and any parameter under each indicator dimension under each first parameter combination scheme corresponds to an optional parameter value; The processing module is further used to determine, for each sample enterprise, N first predicted operating conditions of the sample enterprise under the N different first parameter combination schemes according to the first indicator value of the sample enterprise under each indicator dimension; A determination module, configured to determine a parameter combination scheme that meets the first time period based on the N first predicted operating conditions of the plurality of first sample enterprises and the first enterprise operating conditions of the plurality of first sample enterprises; An enterprise early warning model is determined based on the parameter combination scheme of the first time period.
9. A training device for enterprise early warning model, It is characterized in that include: Memory, used to store computer programs or instructions; A processor, configured to call a computer program or instruction stored in the memory to execute the method as described in any one of claims 1 to 7.
10. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores a computer program, wherein the computer program includes program instructions, and when the program instructions are executed by a computer, the computer is caused to perform the method according to any one of claims 1 to 7.
11. A computer program product, It is characterized in that The computer program product stores instructions, and when a computer reads and executes the instructions, the computer executes the method according to any one of claims 1 to 7.
