Multi-Model Application Method Based on Sub-Model Fusion and Multi-Level Routing

Abstract

The present invention provides a multi-model application method based on sub-model fusion and multi-level routing, belonging to the technical field of financial management. Specifically, it includes: determining the priority data sources in the available data source group according to the data deviation coefficient from the stable data sources in the available data source group and the matching situation of user data of different historical credit users, constructing a credit model based on the sub-models of different priority data sources, using the data deviation coefficient from the priority data sources to determine the routing levels of the stable data sources in the available data source group, and when the priority data source is abnormal, switching to the stable data source at the next routing level in the corresponding available data source group, thereby improving the stability of the credit model.

Description

Technical Field

[0001] The present invention belongs to the technical field of financial management, and particularly relates to a multi-model application method based on sub-model fusion and multi-level routing. Background Art

[0002] In order to evaluate the credit risk of users, credit institutions often combine multiple data sources to evaluate the credit risk. In the patent application for invention CN202410155274.1, "An External Data Source Query Decision Method and Device", the management of external data sources is carried out according to the historical approval data and yield rate of credit applications, that is, the credit risk of users is identified and processed through multiple data sources.

[0003] In the actual process of evaluating and processing credit risks, dozens of third-party data are often connected simultaneously. The entire cooperative modeling process, from the initial sampling, outsourcing, data return, modeling, interface docking, etc. to the final model launch, may take 3 - 6 months. However, often after the actual model is launched, due to reasons such as business cooperation, unstable data sources, and poor actual effects, a certain data source may perform poorly and unstably, requiring re-modeling, which affects the online strategy and is time-consuming and laborious for re-development.

[0004] In view of the above technical problems, specifically, the present application provides a multi-model application method based on sub-model fusion and multi-level routing. Summary of the Invention

[0005] To achieve the object of the present invention, the present invention adopts the following technical solutions:

[0006] Specifically, the present application provides a multi-model application method based on sub-model fusion and multi-level routing, which specifically includes:

[0007] S1 Construct sub-models for different data sources to determine the stable data sources of the data sources based on historical call abnormal data and call delay data;

[0008] S2 Based on the deviation situation of user data of different historical credit users, determine the data deviation coefficients between different stable data sources, use the data deviation coefficients to divide the stable data sources into different data source groups, and determine the available data source groups in the data source groups based on the constituent data of the stable data sources and the data deviation coefficients between them;

[0009] S3 When the association situation of the historical call abnormal data of the stable data sources in the available data source groups meets the requirements, determine the preferred data sources in the available data source groups according to the data deviation coefficients with the stable data sources in the available data source groups and the matching situation of user data of different historical credit users;

[0010] S4 constructs a credit model based on sub-models of different preferred data sources, determines the routing level of the stable data sources in the available data source group by using the data deviation coefficient from the preferred data source, and switches to the stable data source at the next routing level in the corresponding available data source group when the preferred data source is abnormal.

[0011] The beneficial effects of the present invention are as follows:

[0012] By using the data deviation coefficient from the preferred data source, the routing level of the stable data sources in the available data source group is determined, thus avoiding the technical problem that the large data deviation during the switching process of the stable data sources leads to low stability of the credit model, ensuring the stability of the credit data processing of the credit model, and at the same time improving the processing reliability of credit applications.

[0013] When the preferred data source is abnormal, switch to the stable data source at the next routing level in the corresponding available data source group, thus avoiding the technical problem that the construction processing difficulty and duration of the credit model are difficult to meet the requirements due to the frequent construction of the original credit model, improving the switching processing efficiency of the credit model, ensuring the processing efficiency of credit applications, and at the same time reducing the impact of abnormal calls of the data source of the credit model on the credit model.

[0014] A further technical solution lies in constructing sub-models for different data sources, specifically including:

[0015] Based on the user data of the data source, determine the input quantity of the sub-model of the data source;

[0016] Construct the sub-model based on the input quantity of the sub-model.

[0017] A further technical solution lies in that the historical call abnormal data includes the response abnormal times of the data source, and the response abnormal times include the response deviation times, no response times, and response timeout times.

[0018] A further technical solution lies in that the call delay data includes the call processing delays of the data source at different historical call times.

[0019] A further technical solution lies in that the method for determining the stable data source of the data source is:

[0020] Determine the abnormal response times of the data source based on the historical call abnormal data of the data source, and determine the proportion of the abnormal response times of the data source based on the proportion of the abnormal response times;

[0021] Based on the call delay data of the data source, determine the average call processing delay of the data source at different historical call times, and determine the delay response coefficient of the data source according to the ratio of the average call processing delay to the preset delay threshold;

[0022] Determine the response deviation factor of the data source through the average value of the delay response coefficient and the proportion of abnormal response times, and use the response deviation factor to determine whether the data source is a stable data source.

[0023] A further technical solution is that when the response deviation factor of the data source is greater than the preset deviation factor threshold, it is determined that the data source does not belong to a stable data source.

[0024] A further technical solution is that the routing levels of the stable data sources in the available data source group are determined from large to small according to the data deviation coefficients between the stable data sources and the preferred data sources in the available data source group.

[0025] A further technical solution is that the anomalies of the preferred data source include anomalies in calls or call delays that do not meet requirements.

[0026] Other features and advantages will be described in the following specification, and the objectives and other advantages of the present invention are achieved and obtained by the structures specifically pointed out in the specification and the drawings.

[0027] To make the above objectives, features, and advantages of the present invention more obvious and understandable, the following specific preferred embodiments are given, and in conjunction with the accompanying drawings, the detailed description is as follows. Brief Description of the Drawings

[0028] By referring to the drawings and describing its exemplary embodiments in detail, the above and other features and advantages of the present invention will become more obvious.

[0029] Figure 1 is a flowchart of a multi-model application method based on sub-model fusion and multi-level routing;

[0030] Figure 2 is a flowchart of a method for determining stable data sources of a data source;

[0031] Figure 3 is a flowchart of a method for determining the data deviation coefficient between stable data sources;

[0032] Figure 4 is a flowchart of a method for determining an available data source group in a data source group;

[0033] Figure 5 is a flowchart of a method for determining a preferred data source in an available data source group. Detailed implementation mode

[0034] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this specification. Obviously, the described embodiments are only a part of the embodiments of this specification, rather than all the embodiments. Based on the embodiments of this specification, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of this specification.

[0035] In the daily cooperative modeling process, we will connect dozens of third-party data at the same time, and select 3-6 of them for cooperation according to the modeling effect. The entire cooperative modeling process starts from sampling, outsourcing, data return, modeling, interface docking, etc. until the final model goes online, and the entire cycle may take 3-6 months. However, often after the actual model goes online, due to reasons such as business cooperation, unstable data sources, and poor actual effects, a certain data source has poor performance and instability, etc., and it is necessary to rebuild the model, resulting in the online strategy being affected, and redevelopment is time-consuming and laborious.

[0036] In this application, sub-models are developed based on each data source separately, and then the final model is obtained through various model fusion methods. When there is a problem with a single data source, other routes can be quickly switched, improving accuracy and adaptability. At the same time, the design of the model route can also save data costs and reduce losses caused by fluctuations in a single data source.

[0037] Based on the proportion of abnormal response times, determine the proportion of abnormal response times of the data source. According to the ratio of the average value of the call processing delay to the preset delay threshold, determine the delay response coefficient of the data source. Through the average value of the delay response times and the proportion of abnormal response times, determine the response deviation factor of the data source. The data source with a response deviation factor less than 0.1 is used as a stable data source.

[0038] The data deviation coefficient between stable data sources is determined according to the proportion of the number of deviations in user data of different stable data sources among historical credit users.

[0039] The available data source group is a data source group in which the number of stable data sources is greater than 5, and the average value of the data deviation coefficients between different stable data sources is less than 0.05.

[0040] When there is no date in the available data source group where the call delays of the stable data sources do not meet the requirements, it is determined that the association situation of the historical call abnormal data of the stable data sources in the available data source group meets the requirements.

[0041] The priority data source is the stable data source with the smallest average value of the data deviation coefficient from other stable data sources and the smallest average proportion of the number of deviation information items in the user data of historical credit users.

[0042] Embodiment 1

[0043] As Figure 1 shown, the present application provides a multi-model application method based on sub-model fusion and multi-level routing, specifically including:

[0044] S1 Construct sub-models for different data sources to determine the stable data sources of the data sources based on historical call abnormal data and call delay data.

[0045] S2 Based on the deviation conditions of the user data of different historical credit users, determine the data deviation coefficients between different stable data sources, divide the stable data sources into different data source groups by using the data deviation coefficients, and determine the available data source groups in the data source groups by using the constituent data of the stable data sources and the data deviation coefficients between them.

[0046] S3 When the association conditions of the historical call abnormal data of the stable data sources in the available data source groups meet the requirements, determine the priority data sources in the available data source groups according to the data deviation coefficients with the stable data sources in the available data source groups and the matching conditions in the user data of different historical credit users.

[0047] S4 Construct a credit model based on the sub-models of different priority data sources, determine the routing levels of the stable data sources of the available data source groups by using the data deviation coefficients with the priority data sources, and switch to the stable data sources at the next routing level in the corresponding available data source group when the priority data source is abnormal.

[0048] Further, constructing sub-models for different data sources specifically includes:

[0049] Based on the user data of the data source, determine the input quantity of the sub-model of the data source.

[0050] Construct the sub-model based on the input quantity of the sub-model.

[0051] Specifically, the historical call abnormal data includes the response abnormal times of the data source, where the response abnormal times include response deviation times, no response times, and response timeout times.

[0052] It should be noted that the call delay data includes the call processing delay of the data source at different historical call times.

[0053] It can be understood that asFigure 2 As shown, the method for determining the stable data source of the data source is as follows:

[0054] Determine the number of abnormal responses of the data source based on the historical call abnormal data of the data source, and determine the proportion of the number of abnormal responses of the data source based on the proportion of the number of abnormal responses;

[0055] Based on the call delay data of the data source, determine the average value of the call processing delay of the data source at different historical call times, and determine the delay response coefficient of the data source according to the ratio of the average value of the call processing delay to the preset delay threshold;

[0056] Determine the response deviation factor of the data source through the average value of the delay response coefficient and the proportion of the number of abnormal responses, and use the response deviation factor to determine whether the data source is a stable data source.

[0057] Furthermore, when the response deviation factor of the data source is greater than the preset deviation factor threshold, it is determined that the data source does not belong to the stable data source.

[0058] In another possible embodiment, the method for determining the stable data source of the data source is as follows:

[0059] Determine the number of abnormal responses of the data source based on the historical call abnormal data of the data source, and determine the proportion of the number of abnormal responses of the data source based on the proportion of the number of abnormal responses;

[0060] Based on the call delay data of the data source, determine the average value of the call processing delay of the data source at different historical call times;

[0061] Determine the response deviation factor of the data source through the product of the average value of the call processing delay and the proportion of the number of abnormal responses, and use the response deviation factor to determine whether the data source is a stable data source.

[0062] Specifically, as Figure 3 shown, the method for determining the data deviation coefficient between stable data sources is as follows:

[0063] Determine the number of information items with deviations in the user data of different stable data sources for different historical credit users based on the deviation situation of the user data of different historical credit users;

[0064] Use the number of information items with deviations to determine the deviated credit users among the historical credit users;

[0065] Determine the data deviation coefficient between different stable data sources according to the proportion of the number of deviated credit users in the historical credit users.

[0066] Further, the deviated credit user is a historical credit user whose number of information items with deviations does not meet the requirements.

[0067] It can be understood that dividing the stable data sources into different data source groups specifically includes:

[0068] Dividing the stable data sources with the data deviation coefficient between each other within the preset deviation coefficient range into the same data source group.

[0069] Specifically, as Figure 4 shown, the method for determining the available data source group in the data source group is:

[0070] Using the composition data of the stable data sources in the data source group, determine the number of stable data sources in the data source group;

[0071] Taking the average value of the data deviation coefficients between the stable data sources in the data source group, determine the average deviation coefficient of the data source group;

[0072] According to the number of the stable data sources and the average deviation coefficient, determine the group availability factor of the data source group, and based on the group availability factor, determine whether the data source group is an available data source group.

[0073] Further, the group availability factor of the data source group is the ratio of the number of stable data sources to the average deviation coefficient.

[0074] It should be noted that when the group availability factor is greater than the preset available factor threshold, it is determined that the data source group is an available data source group.

[0075] In another possible embodiment, the method for determining the available data source group in the data source group is:

[0076] Using the composition data of the stable data sources in the data source group, determine the number of stable data sources in the data source group;

[0077] Taking the data deviation coefficients between the stable data sources in the data source group, determine the stable data sources whose data deviation coefficients with other stable data sources are not within the preset range, and use them as internal deviation data sources;

[0078] According to the ratio of the number of the stable data sources to the number of the internal deviation data sources, determine the group availability coefficient of the data source group, and based on the group availability coefficient, determine whether the data source group is an available data source group.

[0079] Further, when the group availability coefficient of the data source group is within a preset availability coefficient range, the data source group is determined as an available data source group.

[0080] Specifically, determining that the correlation of the historical call abnormal data of the stable data source meets the requirements specifically includes:

[0081] Based on the historical call abnormal data of the stable data sources in the available data source group, determine the distribution data of the abnormal response times of different stable data sources in different time periods;

[0082] Based on the distribution data of the abnormal response times of different stable data sources in different time periods, determine the abnormal response times in different time periods, and use the abnormal response times to determine the response abnormal time periods of different stable data sources;

[0083] According to the correlation of the response abnormal time periods of different stable data sources, determine the number of stable data sources belonging to the response abnormal time periods in different time periods, and use the number of stable data sources belonging to the response abnormal time periods in different time periods to determine whether the correlation of the historical call abnormal data of the stable data source meets the requirements.

[0084] Further, when the abnormal response times in the time period are greater than a preset abnormal response times threshold, the time period is determined as a response abnormal time period.

[0085] A further technical solution is that when the proportion of the number of stable data sources belonging to the response abnormal time periods in different time periods in the number of the available data source group does not meet the requirements, it is determined that the correlation of the historical call abnormal data of the stable data source does not meet the requirements.

[0086] It can be understood that when the correlation of the historical call abnormal data of the stable data source does not meet the requirements, there is no need to determine the priority data source of the available data source group.

[0087] In another possible embodiment, determining that the correlation of the historical call abnormal data of the stable data source meets the requirements specifically includes:

[0088] Based on the historical call abnormal data of the stable data sources in the available data source group, determine the distribution data of the abnormal response times of different stable data sources in different time periods;

[0089] Based on the distribution data of the abnormal response times of different stable data sources in different time periods, determine the abnormal response times in different time periods;

[0090] Determine the sum of the number of abnormal responses of different stable data sources in different time periods, and use it as the sum of abnormal times. Then, use the sum of abnormal times in different time periods to determine whether the correlation of the historical call abnormal data of the stable data source meets the requirements.

[0091] Furthermore, when the number of time periods in which the sum of abnormal times is within the preset abnormal time range is less than the preset value of the number of time periods, it is determined that the correlation of the historical call abnormal data of the stable data source does not meet the requirements.

[0092] In another possible embodiment, determining that the correlation of the historical call abnormal data of the stable data source meets the requirements specifically includes:

[0093] Based on the historical call abnormal data of the stable data sources in the available data source group, determine the distribution data of the number of abnormal responses of different stable data sources in different time periods;

[0094] Based on the distribution data of the number of abnormal responses of different stable data sources in different time periods, determine the number of abnormal responses in different time periods, determine the sum of the number of abnormal responses of different stable data sources in different time periods, and use it as the sum of abnormal times. When the sum of abnormal times in different time periods does not meet the requirements, it is determined that the correlation of the historical call abnormal data of the stable data source does not meet the requirements;

[0095] When there are time periods in which the sum of abnormal times meets the requirements:

[0096] When it is determined by using the number of abnormal responses that there are no time periods with response abnormalities for different stable data sources, it is determined that the correlation of the historical call abnormal data of the stable data source meets the requirements;

[0097] When it is determined by using the number of abnormal responses that there are stable data sources with time periods of response abnormalities:

[0098] Obtain the number of stable data sources with time periods of response abnormalities. When the proportion of the number of stable data sources with time periods of response abnormalities is less than the preset proportion of the number of stable data sources, it is determined that the correlation of the historical call abnormal data of the stable data source does not meet the requirements;

[0099] When the proportion of the number of stable data sources with time periods of response abnormalities is not less than the preset proportion of the number of stable data sources:

[0100] According to the correlation of response abnormal periods of different stable data sources, determine the number of stable data sources that belong to the response abnormal period in different periods. When the number of periods in which the proportion of the number of stable data sources belonging to the response abnormal period does not meet the requirements is greater than the preset number of periods, it is determined that the correlation of the historical call abnormal data of the stable data source does not meet the requirements;

[0101] Determine that the number of periods in which the proportion of the number of stable data sources belonging to the response abnormal period does not meet the requirements is not greater than the preset number of periods:

[0102] Based on the number of abnormal responses of different stable data sources in different periods, determine the response abnormal coefficient of different periods, and determine whether the correlation of the historical call abnormal data of the stable data source meets the requirements based on the response abnormal coefficient of different periods.

[0103] Furthermore, when the number of periods in which the response abnormal coefficient is within the preset abnormal coefficient range is greater than the preset number of periods, it is determined that the correlation of the historical call abnormal data of the stable data source does not meet the requirements.

[0104] Specifically, the matching situation of the user data of the historical credit users includes the distribution data of the deviation information items and the vacancy information items of the user data of the stable data source in different historical credit users.

[0105] It should be noted that as Figure 5 shown, the method for determining the preferred data source in the available data source group is:

[0106] Determine the average value of the data deviation coefficients of the stable data source in the available data source group and other stable data sources to obtain the average value of the data deviation coefficients of the stable data source;

[0107] According to the matching situation of the user data of the stable data source in different historical credit users, determine the distribution data of the deviation information items and the vacancy information items in different historical credit users, and use the distribution data to determine the proportion of the number of historical credit users with deviation information items or vacancy information items, and use it as the proportion of information deviation users;

[0108] Use the average value of the data deviation coefficient mean and the proportion of information deviation user numbers to determine the adaptation deviation amount of the stable data source, and determine whether the stable data source is the preferred data source in the available data source group based on the adaptation deviation amount.

[0109] Furthermore, the preferred data source in the available data source group is the stable data source with the smallest adaptation deviation amount.

[0110] Optionally, the method for determining the preferred data source in the group of available data sources is as follows:

[0111] S41 Determine the distribution data of the deviation information items and missing information items among different historical credit users based on the matching situation of the stable data source with the user data of different historical credit users, and use the distribution data to determine the proportion of the number of historical credit users with deviation information items or missing information items. Combine the proportion of the number of deviation information items and missing information items among different historical credit users to determine the data matching deviation coefficient of the data source;

[0112] S42 Determine the data deviation amount of the stable data source based on the data deviation coefficient between the stable data source in the group of available data sources and other stable data sources;

[0113] S43 Use the average value of the data deviation amount and the data matching deviation coefficient to determine the adaptation deviation amount of the stable data source, and based on the adaptation deviation amount, determine whether the stable data source is the preferred data source in the group of available data sources.

[0114] Optionally, the above step S41 includes the following content:

[0115] S411 Determine the distribution data of the deviation information items and missing information items among different historical credit users based on the matching situation of the stable data source with the user data of different historical credit users, and use the distribution data to determine the proportion of the number of historical credit users with deviation information items or missing information items, which is used as the proportion of information deviation users. When the proportion of information deviation users of the stable data source does not meet the requirements, it is determined that the stable data source does not belong to the preferred data source in the group of available data sources. When the proportion of information deviation users of the stable data source meets the requirements, proceed to step S412;

[0116] S412 Determine the proportion of the number of deviation information items and missing information items among different historical credit users based on the distribution data of the deviation information items and missing information items among different historical credit users, and use it as the proportion of information deviation quantity. When there are historical credit users whose proportion of information deviation quantity does not meet the requirements, proceed to step S413. When there are no historical credit users whose proportion of information deviation quantity does not meet the requirements, proceed to step S414;

[0117] S413 When the proportion of the number of historical credit users whose proportion of information deviation quantity does not meet the requirements is greater than the proportion of deviation users, it is determined that the stable data source does not belong to the preferred data source in the group of available data sources. When the proportion of the number of historical credit users whose proportion of information deviation quantity does not meet the requirements is not greater than the proportion of deviation users, proceed to step S414;

[0118] S414 determines the proportion of the number of deviation information items or missing information items of the historical credit users by using the distribution data, and combines the proportion of the number of deviation information items and missing information items in different historical credit users to determine the data matching deviation coefficient of the data source. When the data matching deviation coefficient of the data source does not meet the requirements, it is determined that the stable data source does not belong to the preferred data sources in the available data source group. When the data matching deviation coefficient of the data source meets the requirements, it proceeds to step S42.

[0119] Optionally, the above step S42 includes the following content:

[0120] S421 When the data matching deviation coefficient of the data source is within the preset matching deviation coefficient range, it proceeds to step S422. When the data matching deviation coefficient of the data source is not within the preset matching deviation coefficient range, it proceeds to step S424;

[0121] S422 Using the data deviation coefficient between the stable data source in the available data source group and other stable data sources, when it is determined that there are other stable data sources whose data deviation coefficients do not meet the requirements for the stable data source, it proceeds to step S423. When there are no other stable data sources whose data deviation coefficients do not meet the requirements for the stable data source, it proceeds to step S424;

[0122] S423 When the number of other stable data sources whose data deviation coefficients do not meet the requirements is greater than the stable data source quantity threshold, it is determined that the stable data source does not belong to the preferred data sources in the available data source group. When the number of other stable data sources whose data deviation coefficients do not meet the requirements is not greater than the stable data source quantity threshold, it proceeds to step S424;

[0123] S424 Using the data deviation coefficient between the stable data source in the available data source group and other stable data sources, determine the data deviation amount of the stable data source. When the data deviation amount of the stable data source does not meet the requirements, it proceeds to step S425. When the data deviation amount of the stable data source meets the requirements, it proceeds to step S43;

[0124] S425 When the data matching deviation coefficient of the data source is within the preset matching deviation coefficient range, it is determined that the stable data source does not belong to the preferred data sources in the available data source group. When the data matching deviation coefficient of the data source is not within the preset matching deviation coefficient range, it proceeds to step S43.

[0125] It should be noted that constructing a credit model based on sub - models of different preferred data sources specifically includes:

[0126] Combine the sub-models of different priority data sources to obtain a credit model.

[0127] Further, the routing levels of the stable data sources in the available data source group are determined from large to small according to the data deviation coefficients between the stable data sources and the priority data sources in the available data source group.

[0128] Specifically, the anomalies of the priority data sources include abnormal calls or call delays that do not meet the requirements.

[0129] Embodiment 2

[0130] In another possible embodiment, the method for determining the available data source group in the data source group is as follows:

[0131] Use the composition data of the stable data sources in the data source group to determine the number of stable data sources in the data source group. When the number of stable data sources is less than the preset number of data sources, it is determined that the data source group does not belong to the available data source group;

[0132] When the number of stable data sources is not less than the preset number of data sources:

[0133] Use the average value of the data deviation coefficients between the stable data sources in the data source group to determine the average deviation coefficient of the data source group. When the average deviation coefficient of the data source group does not meet the requirements, it is determined that the data source group does not belong to the available data source group;

[0134] When the average deviation coefficient of the data source group meets the requirements:

[0135] When the average deviation coefficient of the data source group is less than the preset coefficient threshold:

[0136] Then it is determined that the data source group belongs to the available data source group;

[0137] When the average deviation coefficient of the data source group is not less than the preset coefficient threshold:

[0138] Use the data deviation coefficients between the stable data sources in the data source group to determine the stable data sources whose data deviation coefficients from other stable data sources are not within the preset range, and use them as internal deviation data sources. When the proportion of the number of internal deviation data sources in the stable data sources does not meet the requirements, it is determined that the data source group does not belong to the available data source group;

[0139] When the proportion of the number of internal deviation data sources in the stable data sources meets the requirements:

[0140] Take the stable data sources other than the internal deviation data source as available data sources. When the number of available data sources in the data source group does not meet the requirements, it is determined that the data source group does not belong to the available data source group;

[0141] When the number of available data sources in the data source group meets the requirements:

[0142] Determine the group availability factor of the data source group according to the number of the stable data sources and the average deviation coefficient, and in combination with the proportion of the number of the internal deviation data sources in the stable data sources, and determine whether the data source group is an available data source group based on the group availability factor.

[0143] Embodiment 3

[0144] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the embodiments of the device, equipment, and non-volatile computer storage medium, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can refer to the partial description of the method embodiments.

[0145] The specific embodiments of this specification are described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be executed in a different order from that in the embodiments and still achieve the desired results. Additionally, the processes depicted in the figures do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0146] The above description is only for one or more embodiments of this specification and is not intended to limit this specification. For those skilled in the art, there can be various changes and modifications to one or more embodiments of this specification. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification shall be included within the scope of the claims of this specification.

Claims

1. A multi-model application method based on sub-model fusion and multi-level routing, characterized in that Specifically include: Construct sub-models for different data sources to determine the stable data sources of the data sources based on historical call exception data and call latency data; Based on the deviation conditions of user data of different historical credit users, determine the data deviation coefficients between different stable data sources, use the data deviation coefficients to divide the stable data sources into different data source groups, and determine the available data source groups in the data source groups using the constituent data of the stable data sources and the data deviation coefficients between them; When the association conditions of the historical call exception data of the stable data sources in the available data source groups meet the requirements, determine the preferred data sources in the available data source groups according to the data deviation coefficients with the stable data sources in the available data source groups and the matching conditions of user data of different historical credit users; Construct a credit model based on sub-models of different preferred data sources, use the data deviation coefficients with the preferred data sources to determine the routing levels of the stable data sources in the available data source groups, and when the preferred data sources are abnormal, switch to the stable data sources of the next routing level in the corresponding available data source groups; Determine that the association conditions of the historical call exception data of the stable data sources meet the requirements, specifically including: Based on the historical call exception data of the stable data sources in the available data source groups, determine the distribution data of the abnormal response times of different stable data sources in different time periods; Based on the distribution data of the abnormal response times of different stable data sources in different time periods, determine the abnormal response times in different time periods, and use the abnormal response times to determine the response abnormal time periods of different stable data sources; According to the association conditions of the response abnormal time periods of different stable data sources, determine the number of stable data sources belonging to the response abnormal time periods in different time periods, and use the number of stable data sources belonging to the response abnormal time periods in different time periods to determine whether the association conditions of the historical call exception data of the stable data sources meet the requirements.

2. The multi-model application method based on sub-model fusion and multi-level routing according to claim 1, wherein Construct sub-models for different data sources, specifically including: Based on the user data of the data source, determine the input quantity of the sub-model of the data source; Construct the sub-model based on the input quantity of the sub-model.

3. The multi-model application method based on sub-model fusion and multi-level routing as described in claim 1, wherein The historical call exception data includes the abnormal response times of the data source, where the abnormal response times include response deviation times, no-response times, and response timeout times.

4. The multi-model application method based on sub-model fusion and multi-level routing according to claim 1, characterized in that The call latency data includes the call processing latency of the data source at different historical call times.

5. The multi-model application method based on sub-model fusion and multi-level routing as claimed in claim 1, wherein, The method for determining the stable data source of the data source is: Determine the abnormal response times of the data source based on the historical call exception data of the data source, and determine the proportion of the abnormal response times of the data source according to the proportion of the abnormal response times; Based on the call latency data of the data source, determine the average value of the call processing latency of the data source at different historical call times, and determine the latency response coefficient of the data source according to the ratio of the average value of the call processing latency to the preset latency threshold; Determine the response deviation factor of the data source through the average value of the delay response coefficient and the proportion of abnormal response times, and use the response deviation factor to determine whether the data source is a stable data source.

6. The multi-model application method based on sub-model fusion and multi-level routing according to claim 5, wherein When the response deviation factor of the data source is greater than the preset deviation factor threshold, it is determined that the data source does not belong to the stable data source.

7. The multi-model application method based on sub-model fusion and multi-level routing according to claim 1, wherein The method for determining the preferred data source in the available data source group is as follows: Determine the average value of the data deviation coefficients of the stable data sources in the available data source group by taking the average value of the data deviation coefficients of the stable data sources in the available data source group and other stable data sources; According to the matching situation of the stable data sources in the user data of different historical credit users, determine the distribution data of the deviation information items and the missing information items in different historical credit users, and use the distribution data to determine the proportion of the number of historical credit users with deviation information items or missing information items, and use it as the proportion of information deviation users; Determine the adaptation deviation amount of the stable data source by using the average value of the data deviation coefficient mean and the proportion of the number of information deviation users, and determine whether the stable data source is the preferred data source in the available data source group based on the adaptation deviation amount.

8. The multi-model application method based on sub-model fusion and multi-level routing according to claim 1, characterized in that The preferred data source in the available data source group is the stable data source with the smallest adaptation deviation amount.

9. The multi-model application method based on sub-model fusion and multi-level routing according to claim 1, wherein The routing level of the stable data sources in the available data source group is determined from large to small according to the data deviation coefficients of the stable data sources and the preferred data source in the available data source group.

10. The multi-model application method based on sub-model fusion and multi-level routing according to claim 1, characterized in that, Abnormalities in the preferred data source include abnormal calls or call delays that do not meet the requirements.

Images