Multi-dimensional data-driven tenant credit portrait construction method, system and equipment

Abstract

The invention relates to the technical field of data processing and credit assessment, in particular to a multi-dimensional data-driven tenant credit portrait construction method, system and equipment, and the method comprises the steps: obtaining at least one piece of core dimension data and at least one piece of peripheral dimension data associated with tenant credit assessment; performing feature extraction on the at least one piece of core dimension data and the at least one piece of peripheral dimension data to obtain at least one core feature and at least one peripheral feature; performing feature cross combination on each core feature and at least one peripheral feature to obtain at least one target feature; and constructing a tenant credit portrait based on the at least one target feature. The technical problems that a traditional tenant credit assessment method is single in dimension, insufficient in data association mining, incapable of coping with credit dynamic changes, insufficient in data utilization and the like are mainly solved.

Description

Technical Field

[0001] The present application relates to the technical field of data processing and credit assessment, and specifically to a method, system and device for constructing a tenant credit profile driven by multi-dimensional data. Background Art

[0002] In today's rental market, tenant credit assessment is crucial for landlords and rental agencies, and its accuracy is directly related to the risk control and profit guarantee of the rental business. However, existing tenant credit assessment technology has many defects and shortcomings.

[0003] Traditional credit assessment methods often rely on data from a single dimension or a limited number of dimensions, such as focusing on the tenant's rental history, and only paying attention to simple information such as whether the rent is paid on time and whether there is any lease default. This single-dimensional assessment method ignores the complexity and multifaceted nature of the tenant's credit status and cannot fully and accurately reflect the tenant's true credit level.

[0004] On the one hand, existing technologies lack effective means of collecting and utilizing some stable factors related to tenants’ long-term credit characteristics, such as education, age and other core dimension information. Although these factors are relatively stable, they can reflect tenants’ credit foundation and potential risks to a certain extent. They have not been fully incorporated into the evaluation system, resulting in inaccurate judgments on the long-term trend of tenants’ credit.

[0005] On the other hand, existing methods also have obvious shortcomings in dealing with dynamic changes in tenants' credit status. Peripheral dimensional data such as recent consumer behavior and social network activities that can reflect changes in tenants' current credit status in real time are rarely systematically integrated into the credit assessment process due to the irregularity of their changes and the difficulty of data acquisition. Even some methods that attempt to incorporate these data simply list them, but fail to deeply explore the inherent relationship between them and the core dimensional data, and are unable to update the tenant's credit profile in a timely and accurate manner based on these dynamic data. When faced with sudden changes in tenants' credit, it is difficult for the lessor to detect and take corresponding measures in a timely manner, thus facing potential risks such as rent arrears, house damage, and even legal disputes.

[0006] In addition, existing credit assessment technologies lack systematicity and scientificity in data processing and feature extraction. There is no complete feature extraction method for all types of collected data, and it is impossible to convert raw data into core features and peripheral features that are truly valuable for credit assessment, resulting in the failure to fully tap the value of the data and greatly reducing the accuracy of the assessment results. Moreover, in the process of feature combination and credit profile construction, there is a lack of effective means to organically integrate features of different dimensions. Instead, each data point is viewed in isolation, and a comprehensive, dynamic and accurate tenant credit assessment model has not been built, making it difficult to meet the growing demand for refined management of tenant credit risks in modern leasing businesses. Summary of the invention

[0007] This application mainly solves the technical problems of traditional tenant credit assessment methods, such as single dimension, insufficient data association mining, inability to cope with dynamic changes in credit, and insufficient data utilization.

[0008] An embodiment of the present application provides a method for constructing a tenant credit profile driven by multidimensional data, and the method for constructing a tenant credit profile includes: Acquiring at least one core dimension data and at least one peripheral dimension data associated with the tenant credit assessment; Performing feature extraction on at least one core dimension data and at least one peripheral dimension data respectively to obtain at least one core feature and at least one peripheral feature; Cross-combining each core feature with at least one peripheral feature to obtain at least one target feature; Build a tenant credit profile based on at least one target feature.

[0009] In the above-mentioned multi-dimensional data driven tenant credit profile construction method, as a preferred solution, before obtaining at least one core dimension data and at least one peripheral dimension data associated with the tenant credit assessment, it also includes: Determine a core dimension list and a peripheral dimension list associated with the tenant credit assessment; the core dimension list lists at least one core dimension, and the peripheral dimension list lists at least one peripheral dimension; The core dimensions are factors related to user credit that do not change or change regularly; The peripheral dimensions are factors related to user credit and that change irregularly.

[0010] In the above-mentioned multi-dimensional data driven tenant credit profile construction method, as a preferred solution, the obtaining of at least one core dimension data and at least one peripheral dimension data associated with the tenant credit assessment includes: Based on one or more core dimensions in the core dimension list, obtaining one or more core dimension data associated with the tenant credit assessment; Based on one or more peripheral dimensions in the peripheral dimension list, one or more peripheral dimension data associated with the tenant credit assessment is obtained.

[0011] In the above-mentioned multi-dimensional data-driven tenant credit profile construction method, as a preferred solution, the cross-combination of each core feature with at least one peripheral feature to obtain at least one target feature includes: Obtaining at least one peripheral feature associated with each core feature; Determining an association relationship between each core feature and at least one associated peripheral feature; At least one target feature is calculated through the association relationship.

[0012] In the above multi-dimensional data driven tenant credit profile construction method, as a preferred solution, the core features are obtained by the following formula : ; The peripheral features are obtained by the following formula : ; in: is the jth core feature, is the i-th core dimension data, and They are all coefficients when obtaining the jth core feature through the i-th core dimension data; The jth peripheral feature, is the i-th peripheral dimension data, and They are all coefficients when obtaining the jth peripheral feature through the i-th peripheral dimension data.

[0013] In the above-mentioned multi-dimensional data-driven tenant credit profile construction method, as a preferred solution, when the correlation relationship between a core feature and the associated peripheral features is a linear relationship, the target feature is obtained by the following formula: ; The target feature set is , .

[0014] In the above-mentioned multi-dimensional data-driven tenant credit profile construction method, as a preferred solution, when the correlation relationship between a core feature and the associated peripheral features is a nonlinear relationship, the target feature is obtained by the following formula: ; The target feature set is , , , and is the combination coefficient.

[0015] In the above-mentioned multi-dimensional data driven tenant credit profile construction method, as a preferred solution, the tenant credit profile is represented by the following formula: ; in, are model parameters, is the intercept term.

[0016] The present application also provides a multi-dimensional data-driven tenant credit profile construction system, the tenant credit profile construction system comprising: A data collection module, the data collection module is used to obtain at least one core dimension data and at least one peripheral dimension data associated with the tenant credit assessment; A feature extraction module, wherein the feature extraction module is used to extract features from at least one core dimension data and at least one peripheral dimension data respectively, to obtain at least one core feature and at least one peripheral feature; A feature cross-combination module, wherein the feature cross-combination module is used to perform feature cross-combination on each core feature and at least one peripheral feature to obtain at least one target feature; A credit profile building module is used to build a tenant credit profile based on at least one target feature.

[0017] The present application also provides a multi-dimensional data-driven tenant credit profile construction device, the device comprising: a memory, a processor, and a tenant credit profile construction program stored on the memory and executable on the processor, the tenant credit profile construction program implementing the tenant credit profile construction method as described above when executed by the processor.

[0018] According to the multi-dimensional data-driven tenant credit profile construction method, system and device of the above-mentioned embodiment, by determining the core dimension list and the peripheral dimension list, and obtaining at least one core dimension data and at least one peripheral dimension data based on the core dimension list, the comprehensiveness and pertinence of data collection are achieved, and the problem that the traditional evaluation method has a single dimension and cannot fully reflect the tenant's credit status is solved, making the evaluation basis more sufficient; then, in the process of extracting features from the data separately and cross-combining each core feature with at least one peripheral feature to obtain at least one target feature, the intrinsic relationship between the data of each dimension is deeply explored, which changes the previous situation of simply listing and treating the data in isolation, effectively improves the utilization value of the data and the accuracy of credit evaluation, and can more accurately capture the potential risks and changing trends of tenant credit, so as to better respond to the dynamic changes in tenant credit status, provide scientific basis for leasing decisions in a timely manner, reduce various risks caused by tenant credit problems in the leasing business, and ensure the stable operation and income of the leasing business. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 Flow chart of the tenant credit profile construction method provided in the embodiment of the present application (I); Figure 2 Flow chart of the tenant credit profile construction method provided in the embodiment of the present application (II); Figure 3 Flow chart of the tenant credit profile construction method provided in the embodiment of the present application (III); Figure 4 This is a functional block diagram of the tenant credit profile building system provided in an embodiment of the present application.

[0020] In the figure: 401, data collection module; 402, feature extraction module; 403, feature cross-combination module; 404, credit profile construction module. DETAILED DESCRIPTION

[0021] The present application is further described in detail below by specific embodiments in conjunction with the accompanying drawings. Wherein similar elements in different embodiments adopt associated similar element numbers. In the following embodiments, many detailed descriptions are intended to enable the present application to be better understood. However, those skilled in the art can easily recognize that some of the features can be omitted in different situations, or can be replaced by other elements, materials, and methods. In some cases, some operations related to the present application are not shown or described in the specification, in order to avoid the core part of the present application being overwhelmed by too much description, and for those skilled in the art, it is not necessary to describe these related operations in detail, and they can fully understand the related operations based on the description in the specification and the general technical knowledge in the art.

[0022] In addition, the features, operations or characteristics described in the specification can be combined in any appropriate manner to form various implementations. At the same time, the steps or actions in the method description can also be interchanged or adjusted in a manner that is obvious to those skilled in the art. Therefore, the various sequences in the specification and the drawings are only for the purpose of clearly describing a certain embodiment and are not meant to be a required sequence, unless otherwise specified that a certain sequence must be followed.

[0023] The serial numbers assigned to the components herein, such as "first", "second", etc., are only used to distinguish the objects described and do not have any order or technical meaning. The "connection" and "coupling" mentioned in this application, unless otherwise specified, include direct and indirect connections (couplings).

[0024] Please refer to Figure 1 In order to solve the technical problems of traditional tenant credit assessment methods such as single dimension, insufficient data association mining, inability to cope with dynamic changes in credit, and insufficient data utilization, a multi-dimensional data-driven tenant credit profile construction method is provided in an embodiment of the present application. The tenant credit profile construction method includes the following steps: Step S101: Acquire at least one core dimension data and at least one peripheral dimension data associated with tenant credit assessment.

[0025] In some embodiments, for obtaining core dimension data based on a core dimension list, when the core dimension includes educational information, the tenant's educational level, graduation school, graduation time and other data are extracted from the educational certification database recognized by the Ministry of Education through a secure data interface to ensure the authenticity and accuracy of the data, and the data is updated every two years to obtain possible changes in educational qualifications; when the core dimension includes occupational category, the current occupation is determined based on the industrial and commercial registration information and the tenant's qualification certification data in the professional field, and the occupational change information is updated by regularly cross-checking with industry association data, with an update cycle of one year; for the core dimension of age, data is synchronized in real time from the household registration management system of the public security department to ensure the timeliness and accuracy of the data, providing stable and reliable basic feature information for tenant credit assessment.

[0026] In some embodiments, when obtaining peripheral dimension data based on the peripheral dimension list, if the peripheral dimension involves the tenant's recent consumption behavior, by cooperating with financial data platforms such as banks and payment institutions and using the API interface provided by them, the tenant's consumption details in the past month are obtained every week, including consumption amount, consumption location, consumption type (such as daily necessities, luxury goods, entertainment consumption, etc.) and consumption time distribution, and credit card usage (overdraft limit, repayment record, whether there is installment payment, etc.); when the peripheral dimension is the tenant's recent social activity information, legal web crawler technology is used, and on the premise of complying with the usage rules and privacy policy of the social media platform, the tenant's login frequency, type of content posted (text, pictures, videos, etc.), and interaction with others (likes, comments, forwarding number and objects) on mainstream social platforms are collected every two weeks, and natural language processing technology is used to perform sentiment analysis on the content they post to determine their emotional tendency (positive, negative or neutral) and whether it involves discussions on economic credit-related topics (such as debt, investment, borrowing, etc.), so as to use this as dynamic peripheral dimension data for credit assessment.

[0027] Step S102: extract features from at least one core dimension data and at least one peripheral dimension data respectively to obtain at least one core feature and at least one peripheral feature.

[0028] In some embodiments, when extracting core features from core dimension data, for the educational level information in the core dimension data, the quantitative score corresponding to the educational level (for example, 4 points for an undergraduate degree, 6 points for a master's degree, etc.), the reputation ranking score of the graduating school (which can be quantified by referring to the authoritative school ranking list), and the number of years between the graduation time and the current time are extracted as core features; for the core dimension data of occupational categories, the stability score of the occupation (determined based on factors such as the average turnover rate in the industry and career development prospects), the mean and standard deviation of the occupational income level are extracted as core features.

[0029] In some embodiments, when extracting peripheral features from peripheral dimension data, for recent consumption behavior in the peripheral dimension data, the fluctuation coefficient of the consumption amount (by calculating the ratio of the standard deviation to the mean), the proportion of luxury consumption, and the credit card overdraft rate (the ratio of the overdraft limit to the credit limit) are extracted as peripheral features; for social activity information, the sentiment tendency score (the ratio of the number of positive sentiment words minus the number of negative sentiment words to the total number of words) and the frequency of discussion of economic credit topics (the ratio of the number of posts involving economic credit-related topics to the total number of posts) are extracted as peripheral features.

[0030] Step S103: cross-combine each core feature with at least one peripheral feature to obtain at least one target feature.

[0031] Step S104: construct a tenant credit profile based on at least one target feature.

[0032] In some embodiments, the tenant credit profile may be constructed by: First, organize the target feature data set obtained by feature cross-combination and divide it into three subsets: training set (70%), validation set (15%) and test set (15%). Select appropriate machine learning algorithms (such as random forest, gradient boosting decision tree, neural network, etc.) to build a tenant credit assessment model, and use the training set to train the model. During the training process, the model's hyperparameters (such as the number of trees, maximum depth, and minimum number of leaf node samples in the random forest; learning rate, number of hidden layer nodes, activation function, etc. in the neural network) are repeatedly adjusted and optimized through cross-validation technology and hyperparameter optimization methods such as grid search and random search to improve the performance and generalization ability of the model. The performance of the model in the training process is evaluated on the validation set, using indicators such as accuracy, recall rate, F1 value, AUC, etc. as evaluation criteria. When the performance indicators of the model no longer improve in 5 consecutive iterations, stop training and save the model version with the best performance. The test set data is input into the trained model for independent evaluation to ensure that the model has good prediction accuracy and stability on unseen data (such as the test set accuracy reaches more than 80% and the AUC value reaches more than 0.85).

[0033] Then, the tenant's target feature data is input into the trained model to calculate the tenant's credit score (ranging from 0 to 100 points), and the credit level is divided according to the score (e.g. 80 to 100 points for high credit level, 40 to 79 points for medium credit level, and 0 to 39 points for low credit level).

[0034] Finally, a detailed tenant credit profile report is generated by combining the tenant's original feature data (including core dimension and peripheral dimension data) and the contribution of each feature in the model (obtained through feature importance analysis). The report covers the tenant's basic information (name, ID number, contact information, etc.), detailed analysis of each dimension data (advantages and risk points of core and peripheral dimensions, such as educational advantages, recent consumption risks, etc.), credit score and level details, potential risk point prompts (such as low education and large consumption fluctuations may have credit risks, frequent debt-related negative comments on social networks need attention, etc.) and targeted credit improvement suggestions (such as improving education, rationally planning consumption, and regulating social comments, etc.). The credit profile is presented to leasing business personnel in an intuitive and visual way (such as bar charts showing the scores of each dimension, line charts showing the trend of credit score changes, heat maps identifying risk areas, etc.), helping them to quickly and comprehensively understand the tenant's credit status, so as to make scientific and reasonable decisions in leasing decisions (such as whether to rent a house, determine the amount of rent deposit, set the lease term, etc.), and it is also convenient for credit risk analysis and management of tenant groups.

[0035] In the above-mentioned multi-dimensional data driven tenant credit profile construction method, as a preferred solution, before obtaining at least one core dimension data and at least one peripheral dimension data associated with the tenant credit assessment in step S101, the following steps are also included: Step S100: Determine a core dimension list and a peripheral dimension list associated with tenant credit assessment; the core dimension list lists at least one core dimension, and the peripheral dimension list lists at least one peripheral dimension.

[0036] The core dimensions are influencing factors that are related to user credit and do not change or change regularly.

[0037] The peripheral dimensions are factors related to user credit and that change irregularly.

[0038] It can be seen that in some embodiments, in order to comprehensively and accurately evaluate the credit status of tenants, it is crucial to conduct in-depth research on the constituent elements and influencing factors of tenant credit. Through the analysis of a large amount of historical rental data, the experience summary of industry experts, and the investigation of tenant behavior patterns, the core dimensions and peripheral dimensions are determined. The core dimensions cover long-term stable factors such as education, age, occupation, and marital status, which reflect the basic credit literacy and long-term economic and social background of tenants. For example, tenants with higher education may be relatively strong in credit awareness and financial planning ability; age is related to the economic stability and credit maturity of tenants; stable occupations and marital status often imply a more reliable source of income and life stability, thereby providing support for long-term credit performance. Peripheral dimensions include recent rental behavior (such as the timeliness of rent payment in the past three months, the use of housing facilities), recent consumer financial activities (credit card consumption records, loan applications and repayments in the past month), recent social network dynamics (social activity in the past two weeks, emotional tendencies in speech, and participation in economic credit topics), etc. These dynamic factors can capture short-term changes in tenant credit status in a timely manner, such as sudden excessive consumption, bad credit speech on social networks, or abnormal rental behavior, which may indicate changes in credit risk.

[0039] Please refer to Figure 2 In the above-mentioned multi-dimensional data driven tenant credit profile construction method, as a preferred solution, the step S101 of obtaining at least one core dimension data and at least one peripheral dimension data associated with the tenant credit assessment specifically includes the following steps: Step S201: Based on one or more core dimensions in the core dimension list, obtain one or more core dimension data associated with the tenant credit assessment.

[0040] In some embodiments, for example, for educational information, a secure data connection channel is established with an educational certification agency authorized by the Ministry of Education, and complete and accurate educational information of tenants is obtained in accordance with strict data encryption and transmission protocols, including detailed information such as the name of the school where the tenant graduated, educational level, and graduation time, and an automatic update mechanism is set every two years to ensure that the tenant's possible educational upgrade or change can be obtained in a timely manner. For another example, for age data, it is directly retrieved in real time from the household registration management system of the public security department, and the authoritative and accurate data resources of the public security system are used to ensure the timeliness and accuracy of age information, and avoid deviations caused by human input errors or outdated data. When obtaining occupational and marital status data, the information submitted by the tenant when applying for a lease, the enterprise registration and operation information of the industrial and commercial administration department (for verifying occupational information), and the marriage registration data of the civil affairs department are integrated. Through multi-source data cross-validation and regular review (once every six months), the authenticity and timeliness of the data are ensured, providing a solid and reliable basic data guarantee for long-term credit assessment.

[0041] Step S202: Based on one or more peripheral dimensions in the peripheral dimension list, obtain one or more peripheral dimension data associated with the tenant credit assessment.

[0042] In some embodiments, the leasing management information system within the leasing enterprise is used to extract the tenant's leasing data for the past three months on a daily basis, such as the specific time and amount of rent payment, whether there is overdue payment and the number of overdue days, whether there is a record of damage and repair of housing facilities, and whether there is unauthorized sublease or renovation, etc. These data can directly reflect the tenant's credit performance and sense of responsibility in recent leasing behavior. Cooperate with financial institutions such as banks and consumer finance companies, and obtain the tenant's financial consumption details for the past month through the secure API interface provided by them, including the consumption location, consumption amount, repayment status (whether the credit card is repaid in full on time, whether there is a record of the minimum repayment amount, etc.), the number of applications, amount, term and repayment progress of various loans, etc., to accurately grasp the tenant's recent consumer finance activities and gain insight into their potential financial pressure and credit risks. Using legal and compliant web crawler technology combined with natural language processing algorithms, we capture tenants' social behavior data from mainstream social media platforms every two weeks, such as the number and type of content posted (text, pictures, videos, etc.), and the interaction of likes, comments, and reposts. We use sentiment analysis models to determine the emotional tendency of their speech (positive, negative, or neutral), and screen out discussions on topics related to economic credit (such as debt disputes, investment failures, credit loan recommendations, etc.), in order to evaluate tenants' credit image and potential risk behavior tendencies in social networks, and provide multi-dimensional dynamic information support for credit assessment.

[0043] Please refer to Figure 3 In the above-mentioned multi-dimensional data driven tenant credit profile construction method, as a preferred solution, the step S103 of cross-combining each core feature with at least one peripheral feature to obtain at least one target feature specifically includes the following steps: Step S301: Acquire at least one peripheral feature associated with each core feature.

[0044] In some embodiments, through deep mining and statistical analysis of historical tenant data, combined with the experience and knowledge of experts in the field, a knowledge base of the association between core features and peripheral features is established. For example, it is found that the core feature of education is associated with the proportion of luxury consumption in recent consumption behavior. Highly educated tenants are usually more rational in their consumption choices and have a lower proportion of luxury consumption; the core feature of career stability is associated with the emotional tendency score in social networks. Tenants with stable careers tend to be more positive and optimistic in social speech and have higher emotional tendency scores; the core feature of age has a certain negative correlation with the credit card overdraft rate. As the age increases, tenants may be more cautious in using credit cards, and the overdraft rate is relatively low. Based on these associations, for each core feature, at least one peripheral feature with potential association with it is screened from the extracted peripheral feature set to form a core feature-peripheral feature association pair, which provides a basic data combination for subsequent feature cross-combinations.

[0045] Step S302: Determine the association relationship between each core feature and at least one associated peripheral feature.

[0046] In some embodiments, association rule algorithms (such as Apriori algorithm) and correlation analysis methods (such as Pearson correlation coefficient and Spearman correlation coefficient) in data mining are used to quantitatively analyze and verify the correlation between core features and peripheral features. For example, the Pearson correlation coefficient between the education level score and the proportion of luxury consumption is calculated. If the result shows a negative correlation and the correlation is significant (such as a correlation coefficient less than -0.4), it indicates that there is a strong inverse correlation between the two, that is, the higher the education level, the lower the proportion of luxury consumption, reflecting that tenants with high education are more rational in their consumption behavior, which has a positive impact on credit assessment; for the career stability score and the social emotional tendency score, the Spearman correlation coefficient analysis found a significant positive correlation (such as a correlation coefficient greater than 0.3), indicating that tenants with stable careers show more positive emotional tendencies in social networks, which can be used as a positive reference factor in credit assessment. At the same time, combined with business logic and actual data distribution, the correlation relationship is further explained and verified. For example, the correlation mechanism between academic qualifications and consumer behavior is explained from the perspective of consumer psychology, and the impact of career stability on social emotions is explained from the perspective of sociology. This ensures that the determined correlation relationship is reasonable and reliable, and can truly reflect the internal logical relationship of the tenant's credit status, providing a solid theoretical basis and data support for subsequent feature cross-combination and credit assessment models.

[0047] Step S303: Calculate at least one target feature through the association relationship.

[0048] In some embodiments, according to the correlation between the determined core features and the peripheral features, a suitable mathematical operation or function transformation is selected to construct the target feature. For example, for the correlation pair of the education level score and the proportion of luxury consumption, a multiplication operation is used to obtain the "education-luxury consumption stability index", that is, the education level score is multiplied by (1-luxury consumption proportion). The higher the index value, the more stable the tenant's education background and consumption behavior, and the lower the credit risk; for the career stability score and the social emotional tendency score, an addition operation is used to obtain the "career-social stability comprehensive index", which directly reflects the comprehensive stability of the tenant in the two dimensions of career and social, and has important reference value for credit assessment; for the correlation between age and credit card overdraft rate, an inverse proportional function relationship (such as target feature = age / (1+credit card overdraft rate)) is constructed to obtain the "age-credit overdraft risk index", and the larger the index value, the lower the credit risk of the tenant in terms of age and credit card use. Through these diverse calculation methods, the core features and peripheral features are organically integrated to generate a series of target features that can more comprehensively and deeply reflect the tenant's credit status. These target features not only contain the information of single-dimensional data, but also fully explore the inherent correlation and synergy between data of different dimensions, greatly enriching the data content and accuracy of credit assessment, and providing key data support for building accurate and effective tenant credit portraits.

[0049] In the above multi-dimensional data driven tenant credit profile construction method, as a preferred solution, the core features are obtained by the following formula ;

[0050] The peripheral features are obtained by the following formula : ; in: is the jth core feature, is the i-th core dimension data, and They are all coefficients when obtaining the jth core feature through the i-th core dimension data; The jth peripheral feature, is the i-th peripheral dimension data, and They are all coefficients when obtaining the jth peripheral feature through the i-th peripheral dimension data.

[0051] In the above-mentioned multi-dimensional data-driven tenant credit profile construction method, as a preferred solution, when the correlation relationship between a core feature and the associated peripheral features is a linear relationship, the target feature is obtained by the following formula: ; The target feature set is , .

[0052] In the above-mentioned multi-dimensional data-driven tenant credit profile construction method, as a preferred solution, when the correlation relationship between a core feature and the associated peripheral features is a nonlinear relationship, the target feature is obtained by the following formula: ; The target feature set is , , , and is the combination coefficient.

[0053] In the above-mentioned multi-dimensional data driven tenant credit profile construction method, as a preferred solution, the tenant credit profile is represented by the following formula: ; in, are model parameters, is the intercept term.

[0054] Please refer to Figure 4 The present application also provides a multi-dimensional data-driven tenant credit profile construction system, the tenant credit profile construction system comprising: A data collection module 401, wherein the data collection module 401 is used to obtain at least one core dimension data and at least one peripheral dimension data associated with the tenant credit assessment; A feature extraction module 402, wherein the feature extraction module 402 is used to extract features from at least one core dimension data and at least one peripheral dimension data, respectively, to obtain at least one core feature and at least one peripheral feature; A feature cross-combination module 403 is used to perform feature cross-combination on each core feature and at least one peripheral feature to obtain at least one target feature; The credit profile building module 404 is used to build a tenant credit profile based on at least one target feature.

[0055] The present application also provides a multi-dimensional data-driven tenant credit profile construction device, the device comprising: a memory, a processor, and a tenant credit profile construction program stored on the memory and executable on the processor, the tenant credit profile construction program implementing the tenant credit profile construction method as described above when executed by the processor.

[0056] Those skilled in the art will appreciate that all or part of the functions of the various methods in the above-mentioned embodiments can be implemented by hardware or by computer programs. When all or part of the functions in the above-mentioned embodiments are implemented by computer programs, the program can be stored in a computer-readable storage medium, and the storage medium can include: read-only memory, random access memory, disk, optical disk, hard disk, etc., and the program is executed by a computer to implement the above-mentioned functions. For example, the program is stored in the memory of the device, and when the program in the memory is executed by the processor, all or part of the above-mentioned functions can be implemented. In addition, when all or part of the functions in the above-mentioned embodiments are implemented by computer programs, the program can also be stored in a storage medium such as a server, another computer, disk, optical disk, flash disk or mobile hard disk, and can be downloaded or copied and saved in the memory of the local device, or the system of the local device is updated, and when the program in the memory is executed by the processor, all or part of the functions in the above-mentioned embodiments can be implemented.

[0057] The above specific examples are used to illustrate the present application, which is only used to help understand the present application and is not intended to limit the present application. For technicians in the technical field to which the present application belongs, they can also make some simple deductions, deformations or substitutions based on the ideas of the present application.

Claims

1. A multi-dimensional data-driven tenant credit profile construction method, characterized in that: The tenant credit profile construction method includes: Acquiring at least one core dimension data and at least one peripheral dimension data associated with the tenant credit assessment; Performing feature extraction on at least one core dimension data and at least one peripheral dimension data respectively to obtain at least one core feature and at least one peripheral feature; Cross-combining each core feature with at least one peripheral feature to obtain at least one target feature; Build a tenant credit profile based on at least one target feature.

2. The multi-dimensional data driven tenant credit profile construction method according to claim 1, characterized in that: Before acquiring at least one core dimension data and at least one peripheral dimension data associated with the tenant credit assessment, the method further includes: Determine a core dimension list and a peripheral dimension list associated with the tenant credit assessment; the core dimension list lists at least one core dimension, and the peripheral dimension list lists at least one peripheral dimension; The core dimensions are factors related to user credit that do not change or change regularly; The peripheral dimensions are factors related to user credit and that change irregularly.

3. The multi-dimensional data driven tenant credit profile construction method according to claim 2, characterized in that: The obtaining of at least one core dimension data and at least one peripheral dimension data associated with the tenant credit assessment includes: Based on one or more core dimensions in the core dimension list, obtaining one or more core dimension data associated with the tenant credit assessment; Based on one or more peripheral dimensions in the peripheral dimension list, one or more peripheral dimension data associated with the tenant credit assessment is obtained.

4. The multi-dimensional data driven tenant credit profile construction method according to claim 1, characterized in that: The step of cross-combining each core feature with at least one peripheral feature to obtain at least one target feature includes: Obtaining at least one peripheral feature associated with each core feature; Determining an association relationship between each core feature and at least one associated peripheral feature; At least one target feature is calculated through the association relationship.

5. The multi-dimensional data driven tenant credit profile construction method according to claim 4, characterized in that: The core features are obtained by the following formula : ； The peripheral features are obtained by the following formula : ； in: is the jth core feature, is the i-th core dimension data, and They are all coefficients when obtaining the jth core feature through the i-th core dimension data; The jth peripheral feature, is the i-th peripheral dimension data, and They are all coefficients when obtaining the jth peripheral feature through the i-th peripheral dimension data.

6. The multi-dimensional data driven tenant credit profile construction method according to claim 5, characterized in that: When the correlation between a core feature and the associated peripheral features is a linear relationship, the target feature is obtained by the following formula: ； The target feature set is ， .

7. The multi-dimensional data driven tenant credit profile construction method according to claim 5, characterized in that: When the correlation between a core feature and the associated peripheral features is a nonlinear relationship, the target feature is obtained by the following formula: ； The target feature set is ， ， 、 and is the combination coefficient.

8. The multi-dimensional data driven tenant credit profile construction method according to claim 6 or 7, characterized in that: The tenant credit profile is expressed by the following formula: ； in, are model parameters, is the intercept term.

9. A multi-dimensional data-driven tenant credit profile construction system, characterized in that: The tenant credit profile building system includes: A data collection module, the data collection module is used to obtain at least one core dimension data and at least one peripheral dimension data associated with the tenant credit assessment; A feature extraction module, wherein the feature extraction module is used to extract features from at least one core dimension data and at least one peripheral dimension data respectively, to obtain at least one core feature and at least one peripheral feature; A feature cross-combination module, wherein the feature cross-combination module is used to perform feature cross-combination on each core feature and at least one peripheral feature to obtain at least one target feature; A credit profile building module is used to build a tenant credit profile based on at least one target feature.

10. A multi-dimensional data-driven tenant credit profile building device, characterized in that: The device includes: a memory, a processor, and a tenant credit profile building program stored in the memory and executable on the processor. When the tenant credit profile building program is executed by the processor, the tenant credit profile building method according to any one of claims 1 to 8 is implemented.

Images