A method and system for attribution analysis of overdue indicator changes

Abstract

The present invention provides a method and system for attribution analysis of overdue index changes, belonging to the field of financial information technology. Specifically, it includes: using the distribution data of the variable index characteristics in different intervals within the abnormal period of the index and the deviation from other unit periods to determine the potential associated index characteristics in the variable index characteristics. Based on the change situation of the distribution data of the detailed index characteristics in different intervals and the change situation of the credit index in different unit periods with different potential associated index characteristics within the specified period, the correlation coefficients and associated index characteristics of different potential associated index characteristics are determined, and the change attribution analysis of the credit index is carried out based on the sorting results of the correlation coefficients of the associated index characteristics, improving the accuracy of the attribution analysis process.

Description

Technical Field

[0001] The present invention belongs to the field of financial information technology, and particularly relates to a method and system for attribution analysis of overdue indicator changes. Background Art

[0002] In the process of carrying out credit business, it is necessary to focus on multiple dimensions of credit indicators such as the delinquency rate, migration rate, recovery rate, non-performing rate, and overdue rate. Generally, if an indicator is abnormal and cannot be intervened and processed in time, it will inevitably cause losses to credit institutions. Therefore, how to quickly achieve attribution analysis of indicator changes and timely control has become an urgent technical problem to be solved.

[0003] In order to achieve attribution analysis of indicator changes of credit indicators, in the invention patent application CN202410258403.X "A Method and System for Abnormal Inspection and Attribution Early Warning of Credit Risk", the determination of concerned credit characteristics is carried out through the analysis of the correlation degree of indicator changes between risk indicators and credit characteristics, so that the optimization processing of risk control strategies can be carried out according to the correlation degree of indicator changes of concerned credit characteristics, reducing the risk of credit application processing. However, there are the following technical problems:

[0004] When credit indicators change, it is often caused by changes in multiple dimensions of credit characteristics. There is no obvious change in any single dimension of credit characteristics. Therefore, if attribution analysis cannot be carried out according to the ranking results of credit characteristics, it may lead to only focusing on a certain dimension and unable to accurately achieve attribution analysis of changes in credit characteristics.

[0005] In view of the above technical problems, the present invention provides a method and system for attribution analysis of overdue indicator changes. Summary of the Invention

[0006] To achieve the object of the present invention, the present invention adopts the following technical solutions:

[0007] According to one aspect of the present invention, a method for attribution analysis of overdue indicator changes is provided.

[0008] A method for attribution analysis of overdue indicator changes specifically includes:

[0009] S1 Using the change data of credit indicators of credit business within a specified time period to determine the abnormal time period of indicators, and combining the fluctuation conditions of credit characteristics between different unit time periods to determine the concerned credit indicators in the credit indicators;

[0010] S2 is based on the feature data under the detailed index features, divides the credit application users of the credit business into multiple intervals, and determines the variable index features in the detailed index features according to the change situation of the distribution data of the detailed index features in different intervals within different unit time periods during the specified time period;

[0011] S3 uses the deviation situation of the distribution data of the variable index features in different intervals within the index abnormal time period and other unit time periods to determine the potential associated index features in the variable index features;

[0012] S4 is based on the change situation of the distribution data of the detailed index features in different intervals within different unit time periods of the potential associated index features and the change situation of the credit index during the specified time period, determines the correlation coefficients and associated index features of different potential associated index features, and performs a change attribution analysis of the credit index based on the sorting result of the correlation coefficients of the associated index features.

[0013] A further technical solution is that the specified time period is determined according to the number of credit application users of the credit business, where the more the number of credit users of the credit business, the shorter the specified time period.

[0014] A further technical solution is that the credit index includes the collection rate, migration rate, recovery rate, non-performing rate, and overdue rate.

[0015] A further technical solution is that the index abnormal time period is a unit time period when the credit index is not within the preset index interval.

[0016] A further technical solution is that the method for determining the associated index features is as follows:

[0017] Based on the change situation of the distribution data of the detailed index features in different intervals within different unit time periods of the potential associated index features and the change situation of the credit index, determine the change correlation coefficients within different unit time periods;

[0018] Based on the change correlation coefficients within different unit time periods, determine the correlation coefficients of the potential associated index features, and based on the correlation coefficients, determine the associated index features in the potential associated index features.

[0019] A further technical solution is that the associated index features are potential associated index features with correlation coefficients greater than the preset correlation coefficient threshold.

[0020] Second aspect, the present invention provides a computer system, comprising: a memory and a processor connected by communication, and a computer program stored on the memory and capable of running on the processor, characterized in that: when the processor runs the computer program, it executes the above-mentioned attribution analysis method for overdue index changes.

[0021] Third aspect, the present invention provides an attribution analysis model for overdue index changes, adopting the above-mentioned attribution analysis method for overdue index changes, specifically including:

[0022] A focus target screening module, a variable feature screening module, a correlation feature screening module, and an attribution analysis processing module;

[0023] Among them, the focus target screening module is responsible for determining the abnormal index period based on the change data of the credit index of the credit business within a specified period, and determining the focus credit index in the credit index by combining the fluctuations of the credit characteristics between different unit periods;

[0024] The variable feature screening module is responsible for dividing the credit application users of the credit business into multiple intervals based on the feature data under the detailed index features, and determining the variable index features in the detailed index features according to the change of the distribution data of the detailed index features in different intervals within different unit periods in the specified period;

[0025] The correlation feature screening module is responsible for determining the potential correlation index features in the variable index features by using the deviation of the distribution data of the variable index features in different intervals within the index abnormal period from that in other unit periods;

[0026] The attribution analysis processing module is responsible for determining the correlation coefficients and correlation index features of different potential correlation index features based on the change of the distribution data of the detailed index features in different intervals within different unit periods of the potential correlation index features and the change of the credit index within the specified period, and performing attribution analysis of the change of the credit index based on the sorting result of the correlation coefficients of the correlation index features.

[0027] The beneficial effects of the present invention are as follows:

[0028] 1. By determining the focus credit index in the credit index based on the abnormal index period and the fluctuations of the credit characteristics between different unit periods, the screening of the focus credit index is realized from two perspectives: the fluctuation of the credit characteristics within the specified period and the abnormal index period of the credit characteristics within the specified period, which also lays a foundation for targeted attribution analysis of the credit index with higher volatility and higher abnormal index degree.

[0029] 2. The change attribution analysis of credit indicators is carried out based on the sorting results of the correlation coefficients of associated indicator features, avoiding the technical problem that the accuracy of the attribution analysis result is difficult to meet the requirements caused by solely considering a single associated indicator feature, achieving a comprehensive consideration of multiple associated indicator features from the perspective of the sorting results, ensuring the accuracy of the attribution analysis process, and laying a foundation for further generating differentiated intervention means to avoid losses of credit institutions.

[0030] Other features and advantages will be described in the following specification. 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.

[0031] 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. Description of the Drawings

[0032] 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.

[0033] Figure 1 is a flowchart of a method for change attribution analysis of overdue indicators;

[0034] Figure 2 is a flowchart of a method for determining the concerned credit indicators in credit indicators;

[0035] Figure 3 is a flowchart of a method for determining the variable indicator features in detailed indicator features;

[0036] Figure 4 is a flowchart of a method for determining associated indicator features;

[0037] Figure 5 is a framework diagram of a computer system. Detailed Embodiments

[0038] In order to enable those skilled in the art of this technology to better understand the technical solutions in this specification, the following will combine the drawings in the embodiments of this specification to clearly and completely describe the technical solutions 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 creative efforts shall fall within the scope of protection of this specification.

[0039] When credit indicators change, it is often caused by changes in credit characteristics in multiple dimensions. There is no obvious change in any single dimension of credit characteristics. Therefore, it is necessary to conduct attribution analysis based on the ranking results of credit characteristics to achieve the attribution analysis of the occurrence of credit characteristic changes.

[0040] Indicator anomaly period: The unit period when the credit indicator is not within the preset indicator range.

[0041] Attention credit indicator: Based on the fluctuation of credit characteristics between different unit periods, determine the change amount of credit characteristics between different adjacent unit periods, and use the unit period with a change amount greater than the preset change amount as the change unit period. Determine the total number of periods based on the sum of the number of indicator anomaly periods and change unit periods, and use the credit indicators within the preset quantity range of the total number of periods as the attention credit indicators in the credit indicators.

[0042] Change indicator characteristics: Determine the proportion of the quantity in different intervals within the unit period based on the distribution data of the detailed indicator characteristics in different intervals within the unit period. Use the proportion of the quantity to determine the deviation amount of the proportion of the quantity in different intervals between adjacent unit periods, and use the average value of the deviation amounts of the proportion of the quantity in different intervals to determine the characteristic change coefficient between adjacent unit periods. Determine the comprehensive change coefficient based on the average value of the characteristic change coefficients of different unit periods, and use the detailed indicator characteristics with a comprehensive change coefficient greater than 0.6 as the change indicator characteristics.

[0043] Potential associated indicator characteristics: Determine the proportion of the quantity in different intervals within the unit period of the change indicator characteristics during the indicator anomaly period, and determine the interval with a deviation from the proportion of the quantity in other unit periods based on the deviation from other unit periods, and use it as the distribution deviation interval. Determine the deviation period proportion factor of different indicator anomaly periods through the proportion of the quantity in other unit periods with the distribution deviation interval, use the average value of the deviation period proportion factors of different indicator anomaly periods to determine the characteristic distribution deviation coefficient of the change indicator characteristics, and use the change indicator characteristics with a characteristic distribution deviation coefficient greater than 0.5 as the potential associated indicator characteristics.

[0044] Associated indicator characteristics: Based on the change situation of the distribution data of the detailed indicator characteristics in different intervals within different unit periods of the potential associated indicator characteristics and the change situation of the credit indicators, determine the change correlation coefficient within different unit periods. Based on the average value of the change correlation coefficients within different unit periods, determine the correlation coefficient of the potential associated indicator characteristics, and use the potential associated indicator characteristics with a correlation coefficient greater than 0.6 as the associated indicator characteristics.

[0045] Embodiment 1 To solve the above problems, according to one aspect of the present invention, as Figure 1As shown, a method for attribution analysis of overdue indicator changes is provided, specifically including:

[0046] S1 Determine the abnormal indicator period based on the change data of the credit indicators of the credit business within a specified period, and determine the concerned credit indicators in the credit indicators by combining the fluctuations of the credit characteristics between different unit periods;

[0047] S2 Based on the characteristic data under the detailed indicator characteristics, divide the credit application users of the credit business into multiple intervals, and determine the variable indicator characteristics in the detailed indicator characteristics according to the change of the distribution data of the detailed indicator characteristics in different intervals within different unit periods in the specified period;

[0048] S3 Use the deviation of the distribution data of the variable indicator characteristics in different intervals within the abnormal indicator period from that in other unit periods to determine the potential associated indicator characteristics in the variable indicator characteristics;

[0049] S4 Based on the change of the distribution data of the detailed indicator characteristics in different intervals and the change of the credit indicators of different potential associated indicator characteristics within different unit periods in the specified period, determine the correlation coefficients and associated indicator characteristics of different potential associated indicator characteristics, and conduct attribution analysis of the change of the credit indicators based on the sorting results of the correlation coefficients of the associated indicator characteristics.

[0050] Furthermore, the specified period is determined according to the number of credit application users of the credit business. The more the number of credit users of the credit business, the shorter the specified period.

[0051] Specifically, the credit indicators include the overdue reminder rate, migration rate, recovery rate, non-performing rate, and overdue rate.

[0052] It can be understood that the abnormal indicator period is the unit period when the credit indicator is not within the preset indicator interval.

[0053] Specifically, as Figure 2 shown, the method for determining the concerned credit indicators in the credit indicators is:

[0054] Obtain the abnormal indicator period of the credit indicator within the specified period, and use the proportion of the number of the abnormal indicator period within the specified period to determine the abnormal indicator coefficient of the credit indicator;

[0055] Based on the fluctuations of credit characteristics between different unit time periods, determine the change amount of credit characteristics between different adjacent unit time periods, and determine the changing unit time periods within the unit time period according to the change amount. Determine the index change coefficient of the credit index through the proportion of the number of the changing time periods in the specified time period;

[0056] Based on the index change coefficient and the index anomaly coefficient, determine the attention coefficient of the credit index, and determine the attention credit index in the credit index through the attention coefficient.

[0057] Further, the attention credit index is a credit index whose attention coefficient is greater than the preset coefficient threshold.

[0058] In addition, it should be noted that the attention coefficient is determined according to the average value of the index change coefficient and the index anomaly coefficient.

[0059] Optionally, the method for determining the attention credit index in the credit index is as follows:

[0060] Obtain the index anomaly time periods of the credit index within the specified time period;

[0061] Based on the fluctuations of credit characteristics between different unit time periods, determine the change amount of credit characteristics between different adjacent unit time periods, and determine the changing unit time periods within the unit time period according to the change amount;

[0062] Determine the total number of time periods based on the sum of the number of the index anomaly time periods and the changing unit time periods, and use the total number of time periods to determine the attention credit index in the credit index.

[0063] Further, when the total number of time periods is not within the preset number range, determine the credit index as the attention credit index.

[0064] Optionally, the method for determining the attention credit index in the credit index includes steps S11 - S13, specifically:

[0065] S11 Based on the fluctuations of credit characteristics between different unit time periods, determine the change amount of credit characteristics between different adjacent unit time periods, and determine the changing unit time periods within the unit time period according to the change amount. Determine the index change coefficient of the credit index through the proportion of the number of the changing time periods in the specified time period and the change amount of credit characteristics between the changing time periods and the adjacent unit time periods;

[0066] S12 Obtain the abnormal time periods of the credit indicator within the specified time period, and determine the abnormal coefficient of the credit indicator by using the proportion of the number of abnormal time periods of the indicator within the specified time period and the characteristic quantity of the credit indicator in the abnormal time periods;

[0067] S13 Determine the attention coefficient of the credit indicator based on the indicator change coefficient and the abnormal coefficient of the indicator, and determine the attention credit indicator in the credit indicator through the attention coefficient.

[0068] Optionally, the above step S11 includes steps S111 - S113, specifically:

[0069] S111 Based on the fluctuation of the credit characteristics between different unit time periods, determine the change amount of the credit characteristics between different adjacent unit time periods. If it is determined that there is no changing unit time period within the unit time period according to the change amount, then determine that the credit indicator does not belong to the attention credit indicator. When there is a changing unit time period, proceed to step S112;

[0070] S112 When the proportion of the number of the changing time periods within the specified time period is greater than the preset proportion, determine that the credit indicator belongs to the attention credit indicator. When the proportion of the number of the changing time periods within the specified time period is not greater than the preset proportion, transfer to step S113;

[0071] S113 Determine the indicator change coefficient of the credit indicator through the proportion of the number of the changing time periods within the specified time period and the change amount of the credit characteristics between the changing time periods and the adjacent unit time periods. When the indicator change coefficient of the credit indicator is not within the preset interval, then transfer to step S12. When the indicator change coefficient of the credit indicator is within the preset interval, use the indicator change coefficient of the credit indicator to determine whether the credit indicator belongs to the attention credit indicator.

[0072] It should be noted that using the indicator change coefficient of the credit indicator to determine whether the credit indicator belongs to the attention credit indicator is specifically:

[0073] When the indicator change coefficient of the credit indicator is greater than the preset change coefficient threshold, then determine that the credit indicator is an attention credit indicator. Otherwise, directly determine that the credit indicator does not belong to the attention credit indicator.

[0074] Optionally, the above step S12 includes steps S121 - S123, specifically:

[0075] S121 Obtain the abnormal index time period of the credit index within the specified time period. When either the number or the proportion of the abnormal index time periods does not meet the requirements, determine the credit index as a concerned credit index. When both the number and the proportion of the abnormal index time periods meet the requirements, proceed to step S122;

[0076] S122 When there is an abnormal index time period in which the characteristic quantity of the credit characteristic is greater than the preset characteristic quantity threshold, proceed to step S123. When there is no abnormal index time period in which the characteristic quantity of the credit characteristic is greater than the preset characteristic quantity threshold, determine the credit index as a concerned credit index;

[0077] S123 Determine the index abnormal coefficient of the credit index by using the proportion of the number of the abnormal index time periods within the specified time period and the characteristic quantity of the credit index of the abnormal index time period. When the index abnormal coefficient of the credit index does not meet the requirements, determine the credit index as a concerned credit index. When the index abnormal coefficient of the credit index meets the requirements, proceed to step S13.

[0078] Further, divide the credit application users of the credit business into multiple intervals, specifically including:

[0079] Divide the credit application users into the corresponding intervals according to the intervals where the characteristic data of the credit application users under the detailed index characteristics are located.

[0080] It should be noted that, as Figure 3 shown, the method for determining the variable index characteristic in the detailed index characteristics is:

[0081] Determine the proportion of the number in different intervals within the unit time period based on the distribution data of the detailed index characteristics in different intervals within the unit time period;

[0082] Use the proportion of the number to determine the deviation interval between adjacent unit time periods, and determine the characteristic change coefficient between adjacent unit time periods according to the number of the deviation intervals;

[0083] Based on the characteristic change coefficient, determine the distribution change time periods in the unit time period, and use the number of the distribution change time periods to determine the variable index characteristic in the detailed index characteristics.

[0084] Further, the variable index characteristic is the detailed index characteristic in which the number of the distribution change time periods is greater than the preset time period number.

[0085] It can be understood that the detailed index characteristics include the work information, educational background information, credit application information, credit utilization information, and overdue information of the credit application users.

[0086] It should be further noted that the method for determining the variable index feature among the detailed index features is as follows:

[0087] Determine the quantity proportion in different intervals within the unit time period based on the distribution data of the detailed index features in different intervals within the unit time period;

[0088] Use the quantity proportion to determine the deviation amount of the quantity proportion in different intervals between adjacent unit time periods, and use the average value of the deviation amounts of the quantity proportion in different intervals to determine the feature change coefficient between adjacent unit time periods;

[0089] Determine the comprehensive change coefficient based on the average value of the feature change coefficients of different unit time periods, and use the comprehensive change coefficient to determine the variable index feature among the detailed index features.

[0090] Further, the variable index feature is the detailed index feature whose comprehensive change coefficient does not meet the requirements.

[0091] Optionally, the determination of the variable index feature among the detailed index features specifically includes the following steps S21 - S23:

[0092] S21 Determine the quantity proportion in different intervals within the unit time period based on the distribution data of the detailed index features in different intervals within the unit time period;

[0093] S22 Use the quantity proportion to determine the deviation amount of the quantity proportion in different intervals between adjacent unit time periods, and use the average value of the deviation amounts of the quantity proportion in different intervals and the quantity of the deviation intervals to determine the feature change coefficient between adjacent unit time periods;

[0094] S23 Determine the distribution change time period in the unit time period based on the feature change coefficients of different unit time periods, determine the comprehensive change coefficient based on the quantity of the distribution time period and the feature change coefficients of different distribution change time periods, and use the comprehensive change coefficient to determine the variable index feature among the detailed index features.

[0095] Optionally, steps S221 - S222 are included in the above step S22, specifically as follows:

[0096] S221 Use the quantity proportion to determine the deviation amount of the quantity proportion in different intervals between adjacent unit time periods. If it is determined that there is no deviation interval between different unit time periods and adjacent unit time periods using the deviation amount, then it is determined that the detailed index feature does not belong to the variable index feature. When there is a unit time period with a deviation interval between it and the adjacent unit time period, then enter step S222;

[0097] S222 Obtain the number of unit time periods with deviation intervals. When the number of unit time periods with deviation intervals is greater than the preset number of time periods, it is determined that the detailed indicator feature belongs to a variable indicator feature. When the number of unit time periods with deviation intervals is not greater than the preset number of time periods, proceed to step S223;

[0098] S223 Determine the characteristic change coefficient with adjacent unit time periods by using the average value of the deviation amount of the quantity proportion in different intervals and the number of deviation intervals. When the number of unit time periods with a characteristic change coefficient greater than the preset change coefficient threshold does not meet the requirements, it is determined that the detailed indicator feature belongs to a variable indicator feature. When the number of unit time periods with a characteristic change coefficient greater than the preset change coefficient threshold meets the requirements, proceed to step S23.

[0099] Optionally, steps S231 - S233 are included in the above step S23, specifically:

[0100] S231 Determine the distribution change time periods in the unit time periods based on the characteristic change coefficients of different unit time periods. When the number of distribution change time periods does not meet the requirements, it is determined that the detailed indicator feature is a variable indicator feature. When the number of distribution change time periods meets the requirements, proceed to step S232;

[0101] S232 When the number of distribution change time periods is greater than the preset number of distribution change time periods, proceed to step S223. When the number of distribution change time periods is not greater than the preset number of distribution change time periods, proceed to step S224;

[0102] S233 Obtain the average value of the characteristic change coefficients of different distribution change time periods. When the average value of the characteristic change coefficients is greater than the preset characteristic change coefficient threshold, it is determined that the detailed indicator feature is a variable indicator feature. When the average value of the characteristic change coefficients is not greater than the preset characteristic change coefficient threshold, proceed to step S234;

[0103] S234 Determine the comprehensive change coefficient based on the number of distribution time periods and the characteristic change coefficients of different distribution change time periods, and use the comprehensive change coefficient to determine the variable indicator feature in the detailed indicator feature.

[0104] Furthermore, the method for determining the potential associated indicator feature in the variable indicator feature is:

[0105] Determine the quantity proportion in different intervals within the unit time period based on the distribution data of the detailed indicator feature in different intervals within the unit time period;

[0106] Determine the proportion of the quantity of the variable index feature in different intervals within a unit time period during the abnormal index time period, and determine the interval with a deviation from the proportion of the quantity in other unit time periods based on the deviation from other unit time periods, and use it as the distribution deviation interval;

[0107] Determine the deviation time period proportion factor of different abnormal index time periods based on the proportion of the quantity in other unit time periods with a distribution deviation interval;

[0108] Use the average value of the deviation time period proportion factors of different abnormal index time periods to determine the characteristic distribution deviation coefficient of the variable index feature, and determine whether the variable index feature is a potential associated index feature based on the characteristic distribution deviation coefficient.

[0109] Specifically, when the characteristic distribution deviation coefficient is greater than the preset distribution deviation coefficient threshold, it is determined that the variable index feature is a potential associated index feature.

[0110] It can be understood that when the deviation of the quantity proportion is greater than the preset proportion deviation threshold, the interval is determined as the distribution deviation interval.

[0111] In addition, it should be noted that the method for determining the potential associated index feature in the variable index feature includes steps S31 - S34, specifically:

[0112] S31 Determine the proportion of the quantity in different intervals within the unit time period based on the distribution data of the detailed index feature in different intervals within the unit time period;

[0113] S32 Determine the proportion of the quantity of the variable index feature in different intervals within a unit time period during the abnormal index time period, and determine the interval with a deviation from the proportion of the quantity in other unit time periods based on the deviation from other unit time periods, and use it as the distribution deviation interval;

[0114] S33 Determine the distribution deviation coefficient between the abnormal index time period and other unit time periods based on the quantity of the distribution deviation interval and the deviation of the proportion of the quantity in different distribution deviation intervals, and determine the comprehensive deviation coefficient of different abnormal index time periods based on the distribution deviation coefficients of different unit time periods;

[0115] S34 Use the average value of the comprehensive deviation coefficients of different abnormal index time periods to determine the characteristic distribution deviation coefficient of the variable index feature, and determine whether the variable index feature is a potential associated index feature based on the characteristic distribution deviation coefficient.

[0116] Optionally, steps S321 - S322 are included in the above step S32, specifically:

[0117] S321 Determine the proportion of the quantity of the variable index feature in different intervals within a unit time period during the index abnormal time period, and determine the interval with a deviation from the quantity proportion in other unit time periods according to the deviation situation from other unit time periods, and use it as the distribution deviation interval. When there are distribution deviation intervals in all index abnormal time periods, go to step S322. When there is an index abnormal time period without a distribution deviation interval, it is determined that the variable index feature does not belong to the potential associated index feature;

[0118] S322 When there are distribution deviation intervals between all index abnormal time periods and other unit time periods, go to step S33. When there is an index abnormal time period without a distribution deviation interval from other unit time periods, it is determined that the variable index feature does not belong to the potential associated index feature.

[0119] Optionally, the above step S33 includes steps S331 - S333, specifically:

[0120] S331 Determine the distribution deviation coefficient between the index abnormal time period and other unit time periods according to the quantity of the distribution deviation interval and the deviation situation of the proportion of the quantity of different distribution deviation intervals. When there is an index abnormal time period with a distribution deviation coefficient less than the preset coefficient threshold from other unit time periods, go to step S332. When the distribution deviation coefficients between all index abnormal time periods and other unit time periods are not less than the preset coefficient threshold, it is determined that the variable index feature belongs to the potential associated index feature;

[0121] S332 Obtain the quantity of other unit time periods with a distribution deviation coefficient less than the preset coefficient threshold in different index abnormal time periods. When there is an index abnormal time period with the quantity of other unit time periods with a distribution deviation coefficient less than the preset coefficient threshold not meeting the requirements, it is determined that the variable index feature does not belong to the potential associated index feature. When there is no index abnormal time period with the quantity of other unit time periods with a distribution deviation coefficient less than the preset coefficient threshold not meeting the requirements, go to step S333;

[0122] S333 Determine the comprehensive deviation coefficient of different index abnormal time periods according to the distribution deviation coefficients with different unit time periods. When there is an index abnormal time period with a comprehensive deviation coefficient less than the set deviation coefficient threshold, it is determined that the variable index feature does not belong to the potential associated index feature. When there is no index abnormal time period with a comprehensive deviation coefficient less than the set deviation coefficient threshold, go to step S34.

[0123] Specifically, as Figure 4 shown, the method for determining the associated index feature is:

[0124] Based on the changes in the distribution data of the detailed index features in different intervals within the unit time periods with different potential correlation index features and the changes in the credit index, determine the change correlation coefficients within different unit time periods;

[0125] Based on the change correlation coefficients within different unit time periods, determine the correlation coefficient of the potential correlation index feature, and based on the correlation coefficient, determine the correlation index feature in the potential correlation index feature.

[0126] Further, the correlation index feature is a potential correlation index feature with a correlation coefficient greater than the preset correlation coefficient threshold.

[0127] In addition, it should be noted that the analysis of the attribution of changes in credit indicators specifically includes:

[0128] Based on the sorting result of the correlation coefficients of the correlation index features, determine the order of the correlation index features;

[0129] Obtain the index abnormal time period of the credit indicator within the specified time period, and use the proportion of the number of the index abnormal time period within the specified time period to determine the index abnormal coefficient of the credit indicator;

[0130] Determine the number of attribution analysis of the credit indicator according to the index abnormal coefficient of the credit indicator, and perform the attribution analysis of the change of the credit indicator based on the number of attribution analysis and the order of the correlation index features.

[0131] Embodiment 2 Second aspect, as Figure 5 shown, the present invention provides a computer system, including: a memory and a processor connected by communication, and a computer program stored on the memory and capable of running on the processor, characterized in that: when the processor runs the computer program, it executes the above-mentioned method for attributing the change of overdue indicators.

[0132] Embodiment 3 Third aspect, the present invention provides an attribution analysis model for the change of overdue indicators, adopting the above-mentioned method for attributing the change of overdue indicators, specifically including:

[0133] A focus target screening module, a change feature screening module, a correlation feature screening module, and an attribution analysis processing module;

[0134] Among them, the focus target screening module is responsible for determining the index abnormal time period based on the change data of the credit indicator of the credit business within the specified time period, and combining the fluctuation situation of the credit characteristics between different unit time periods to determine the focus credit indicator in the credit indicator;

[0135] The variable feature screening module is responsible for dividing the credit application users of the credit business into multiple intervals based on the feature data under the detailed index features, and determining the variable index features in the detailed index features according to the change situation of the distribution data of the detailed index features in different intervals within different unit time periods during the specified time period;

[0136] The associated feature screening module is responsible for determining the potential associated index features in the variable index features by using the deviation situation between the distribution data of the variable index features in different intervals during the index abnormal time period and those in other unit time periods;

[0137] The attribution analysis processing module is responsible for determining the correlation coefficients and associated index features of different potential associated index features based on the change situation of the distribution data of the detailed index features in different intervals within different unit time periods of the potential associated index features and the change situation of the credit indexes during the specified time period, and performing the change attribution analysis of the credit indexes according to the sorting result of the correlation coefficients of the associated index features.

[0138] 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.

[0139] 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 than 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.

[0140] The above is only one or more embodiments of this specification and is not used to limit this specification. For those skilled in the art, one or more embodiments of this specification can have various changes and modifications. 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. An attribution analysis method for changes in overdue indicators, characterized in that, Specifically, it includes: Determining the abnormal index time period based on the change data of the credit indexes of the credit business within a specified time period, and determining the concerned credit indexes in the credit indexes by combining the fluctuation of the credit characteristics between different unit time periods; Based on the characteristic data under the detailed index characteristics, dividing the credit application users of the credit business into multiple intervals, and determining the variable index characteristics in the detailed index characteristics according to the change of the distribution data of the detailed index characteristics in different intervals within different unit time periods in the specified time period; Using the deviation of the distribution data of the variable index characteristics in different intervals within the abnormal index time period from that in other unit time periods to determine the potential associated index characteristics in the variable index characteristics; Based on the change of the distribution data of the detailed index characteristics in different intervals and the change of the credit indexes in different unit time periods of the potential associated index characteristics within the specified time period, determining the correlation coefficients and associated index characteristics of different potential associated index characteristics, and performing attribution analysis of the change of the concerned credit indexes based on the sorting result of the correlation coefficients of the associated index characteristics; The credit indexes include the collection rate, migration rate, recovery rate, non-performing rate, and overdue rate; The method for determining the potential associated index characteristics in the variable index characteristics is as follows: Determining the quantity proportion of different intervals within the unit time period based on the distribution data of the detailed index characteristics in different intervals within the unit time period; Determining the quantity proportion of different intervals within the unit time period of the variable index characteristics within the abnormal index time period, and determining the intervals with deviation from the quantity proportion in other unit time periods based on the deviation from that in other unit time periods, and taking them as the distribution deviation intervals; Determining the deviation time period proportion factor of different abnormal index time periods through the quantity proportion of other unit time periods with distribution deviation intervals; Using the average value of the deviation time period proportion factors of different abnormal index time periods to determine the characteristic distribution deviation coefficient of the variable index characteristics, and determining whether the variable index characteristics are potential associated index characteristics according to the characteristic distribution deviation coefficient; 2. The attribution analysis method for overdue indicator changes according to claim 1, wherein, The specified time period is determined according to the number of credit application users of the credit business. The more the number of credit users of the credit business, the shorter the specified time period; 3. The attribution analysis method for overdue indicator changes according to claim 1, wherein The abnormal index time period is the unit time period when the credit index is not within the preset index interval; 4. The attribution analysis method for overdue indicator changes according to claim 1, characterized in that The method for determining the concerned credit indexes in the credit indexes is as follows: Obtaining the abnormal index time period of the credit index within the specified time period, and determining the index abnormal coefficient of the credit index by using the quantity proportion of the abnormal index time period within the specified time period; Based on the fluctuation of the credit characteristics between different unit time periods, determining the change amount of the credit characteristics between different adjacent unit time periods, and determining the variable unit time period within the unit time period according to the change amount. Determining the index change coefficient of the credit index by using the quantity proportion of the variable unit time period within the specified time period; Determine the attention coefficient of the credit indicators based on the index change coefficient and the index anomaly coefficient, and determine the attention credit indicators in the credit indicators through the attention coefficient.

5. The attribution analysis method for overdue index changes according to claim 4, characterized in that The attention credit indicator is a credit indicator whose attention coefficient is greater than a preset coefficient threshold.

6. The attribution analysis method for overdue indicator changes according to claim 1, characterized in that Divide the credit application users of the credit business into multiple intervals, specifically including: Divide the credit application users into corresponding intervals according to the intervals where the characteristic data of the credit application users under the detailed index characteristics are located.

7. The attribution analysis method for overdue indicator changes according to claim 1, wherein The method for determining the change index characteristics in the detailed index characteristics is as follows: Determine the proportion of the number in different intervals within the unit time period based on the distribution data of the detailed index characteristics in different intervals within the unit time period; Use the proportion of the number to determine the deviation interval between adjacent unit time periods, and determine the characteristic change coefficient between adjacent unit time periods according to the number of the deviation intervals; Based on the characteristic change coefficient, determine the distribution change time period within the unit time period, and use the number of the distribution change time periods to determine the change index characteristics in the detailed index characteristics.

8. A computer system, comprising: A memory and a processor connected by communication, and a computer program stored on the memory and capable of running on the processor, characterized in that: when the processor runs the computer program, it executes an attribution analysis method for overdue index changes according to any one of claims 1-7.

Images