Golden deposit-free online leasing method and system

Abstract

The invention provides a guarantee-deposit-free online leasing method and system, and relates to the technical field of online leasing, and the method comprises the following steps: carrying out the training and learning of the personal data of historical leasing users through a neural network model; inputting personal data of a new leasing user into the trained and learned neural network model, and determining a first credit risk assessment prediction result of the new leasing user according to an output result, the personal data of the historical leasing user and the new leasing user comprises personal identification information data, personal basic data, personal credit data and personal family data; and matching a mobile phone model for leasing for a new leasing user. Compared with a traditional deposit-free leasing method, the deposit-free online leasing method has the advantages that the personal data of the new leasing user adopted by credit risk assessment prediction is more comprehensive, so that the accuracy of a credit risk assessment prediction result of the new leasing user is effectively improved.

Description

Technical Field

[0001] The present application relates to the field of online leasing technology, and in particular to a deposit-free online leasing method and system. Background Art

[0002] As the demand for online rental business continues to expand, the application of deposit-free rental technology is becoming more and more extensive. At the same time, with the rapid iteration of smart phone technology, new mobile phones are constantly being introduced, and users' demand for trying new mobile phones is growing, which promotes the development of online models of mobile phone deposit-free rental. For example, the invention patent application with publication number CN115587872A discloses an online rental method based on blockchain. The above-mentioned online rental method based on blockchain receives the rental demand information of the user's online rental, marks and determines the rental target; based on blockchain technology, loads the smart contract corresponding to the rental target; tracks the user's signing and payment process, and generates a signed contract; when the online rental expires, performs performance verification according to the signed contract, and performs deposit refund processing according to the performance verification result, completing the online mobile phone rental process.

[0003] In the process of online mobile phone rental, it is necessary to evaluate the credit of the rental user to reduce the risk of bad rental behavior and improve the stability of online mobile phone deposit-free rental. For example, the invention patent with publication number CN 117474640B discloses a deposit-free rental method, which obtains comprehensive data of deposit-free rental, constructs a comprehensive evaluation model for deposit-free rental based on the comprehensive data of deposit-free rental, obtains the comprehensive evaluation coefficient data of deposit-free rental calculated by the comprehensive evaluation model of deposit-free rental, and evaluates the credit of the rental user.

[0004] However, when evaluating the credit of the lessee in the above-mentioned deposit-free leasing method, the credit evaluation data of the lessee is obtained by collecting the relevant data of the credit evaluation of the lessee in the deposit-free leasing, extracting the relevant data, and obtaining the credit evaluation data group of the lessee. The collected relevant data only includes the credit score data of the lessee, the number of overdue payments of the lessee, and the authenticity data of the identity of the lessee. Therefore, the collected data is relatively single and one-sided. The customer credit risk status obtained based on this evaluation prediction has the problem of low accuracy, which increases the risk of bad leasing behavior. Summary of the invention

[0005] The purpose of this application is to provide a deposit-free online leasing method and system to solve the problem that the deposit-free leasing method in the related art has low accuracy in assessing and predicting the customer's credit risk status.

[0006] In the first aspect, the present application provides a deposit-free online rental method using the following technical solutions:

[0007] A deposit-free online leasing method comprises the following steps:

[0008] Collect personal data of historical rental users and rental defaults;

[0009] The personal data of historical rental users are trained and learned through the neural network model, and the link weights between the neurons of the neural network model are adjusted to determine the intrinsic connection between personal data and rental defaults;

[0010] Collecting personal data of new renters, inputting the personal data of new renters into the trained neural network model, and determining a first credit risk assessment prediction result of the new renter based on the output result, wherein the personal data of the historical renters and the new renters include personal identification information data, personal basic information data, personal credit information data, and personal family information data;

[0011] Receive leasing demands from new leasing users, match available mobile phone models for leasing to new leasing users, and generate leasing orders based on the choices of new leasing users.

[0012] Optionally, the personal identification information data includes name, gender, ethnicity, place of origin and date of birth; the personal basic information data includes education, marriage, housing, means of transportation, work unit, working hours, nature of occupation, job level, title and time of obtaining the title; the personal credit information data includes application credit score, annual income, credit card, securities, bank deposits and insurance data; the personal family information data includes family size, age structure, employment status and social insurance participation data.

[0013] Optionally, a sample number threshold of historical rental users in the neural network model is set. When the sample number of historical rental users reaches the threshold, a group of historical rental user samples is eliminated before each training and learning of the personal data of historical rental users through the neural network model. The eliminated historical rental user samples are selected according to the following priority: personal data lacks key information > personal data has the lowest degree of completeness > historical rental users are the oldest.

[0014] Optionally, the receiving of rental demands of new rental users and matching available mobile phone models for rental for the new rental users includes the following steps: constructing a user portrait based on the purchase history, browsing behavior and evaluation feedback information of the new rental users, analyzing the rental preferences of the new rental users through cluster analysis and association rule methods based on the rental demands of the new rental users, and generating a list of mobile phone models matching the new rental users through collaborative filtering and content-based recommendation algorithms.

[0015] Optionally, the function expression of the neural network model is:

[0016]

[0017] The error function is

[0018] In the formula, net jt is the input of the jth neuron in the tth layer, y jt is the output of the jth neuron in the tth layer, x i is the output of the i-th neuron in the t-1th layer to the j-th neuron in the t-th layer, w ij is the connection weight between the i-th neuron in the t-1th layer and the j-th neuron in the t-th layer, y pm ′ is the target output, y pm is the actual output, θ jt is the output threshold of the jth neuron in the tth layer, MSE is the mean square error between the output value of the neural network model and the expected value of the sample, P is the number of training samples, and M is the output node;

[0019] The training and learning of personal data of historical rental users by a neural network model and adjusting the link weights between neurons of the neural network model include the following steps:

[0020] Normalize the personal data of historical rental users, convert the personal data of historical rental users into sample data of unified standard, input the data into the neural network model, and calculate the output of the neural network model;

[0021] Calculate the mean square error (MSE) between the output value and the sample expected value in the neural network model, then reversely calculate from the output layer to the input layer, adjust the weights w of each link in the direction of reducing the mean square error (MSE), and terminate the training when the mean square error (MSE) is less than the given allowable output threshold (θ).

[0022] Optionally, the normalization of the personal data of historical rental users to convert the personal data of historical rental users into sample data of unified standards includes the following steps: for non-numeric data, assigning different numerical values ​​in the range of [0, 1] to each personal data according to the degree of its influence on personal credit assessment in actual work; for numerical data, converting the personal data through the normal distribution S function to convert it into a numerical value distributed in (0, 1).

[0023] Optionally, after inputting the personal data of the new rental user into the trained neural network model and calculating the output result, the verification unit calculates the rental default rate of the new rental user, and determines the second credit risk assessment prediction result of the new rental user. If the second credit risk assessment prediction result is the same as the first credit risk assessment prediction result, the new rental user is matched with a mobile phone model available for rental based on the first credit risk assessment prediction result. Otherwise, the third credit risk assessment prediction result of the new rental user is calculated by the review unit, and the item between the first credit risk assessment prediction result and the second credit risk assessment prediction result that is the same as the third credit risk assessment prediction result is selected as the basis for matching the new rental user with a mobile phone model available for rental.

[0024] Optionally, the calculation of the lease default rate of the new lease user by the verification unit includes the following steps: substituting the personal data of the historical lease user into the Logit prediction model, selecting the optimal personal data weight according to the SPSS software, substituting the personal data of the new lease user into the Logit distribution function, and calculating the lease default rate of the new lease user;

[0025] The Logit prediction model is:

[0026] logitP=a0+a1y1+a2y2+...+a m y m ,

[0027] Where P is the rental default rate of historical rental users, y1, y2, ..., y m is the personal data of historical rental users, a0, a1, ..., a m The weight of personal data;

[0028] The Logit distribution function is:

[0029]

[0030] Where P is the rental default rate of new tenants, y1, y2, ..., y m is the personal data of the new rental user, a0, a1, ..., a m is the optimal individual data weight selected by SPSS software.

[0031] Optionally, the calculation of the third credit risk assessment prediction result of the new lease user by the review unit includes the following steps:

[0032] Assume that the credit risk assessment discriminant function is: Y = C1X1 + C2X2 + ... + C p X p ,

[0033] Where Y is the credit risk assessment indicator, X1, X2, ..., X P The personal data of the rental user, C1, C2, ..., C P The weight of personal data;

[0034] The historical rental user samples are divided into two groups according to the credit risk assessment of good and poor. The group with good credit risk assessment is group A, the sample size is n1 groups, and the average credit risk assessment index is The group with poor credit risk assessment is group B, with a sample size of n2 groups and an average credit risk assessment index of

[0035] Calculate the weighted average of the average credit risk assessment index Y c :

[0036]

[0037] The personal data of the new rental user X 0 1, X 0 2, ..., X 0 P Substitute the credit risk assessment discriminant function to obtain the credit risk assessment index of the new rental user:

[0038] Y0=C1X 0 1+C2X 0 2+...+C p X 0 p ;

[0039] when When Y0>Y c , then the credit risk assessment of the new lease user is excellent. If Y0 <Y c , then the credit risk assessment of the new lease user is poor;

[0040] when When Y0 <Y c , then the credit risk assessment of the new lease user is excellent. If Y0>Y c , then the credit risk assessment of the new lease user is poor.

[0041] In the second aspect, the deposit-free online rental system provided by this application adopts the following technical solution:

[0042] A deposit-free online rental system, comprising:

[0043] An information collection unit, used to collect personal data of historical rental users and rental default situations;

[0044] A first credit risk assessment prediction unit, used to train and learn personal data of historical rental users through a neural network model, collect personal data of new rental users, and calculate a first credit risk assessment prediction result of the new rental user;

[0045] The rental matching unit is used to receive the rental demand of the new rental user, match the mobile phone models available for rental for the new rental user, and generate a rental order according to the selection of the new rental user.

[0046] In summary, this application at least includes the following beneficial technical effects:

[0047] 1. The deposit-free online rental method of the present application uses a neural network model to train and learn the personal data of historical rental users, and inputs the personal data of new rental users into the trained neural network model, and determines the first credit risk assessment prediction result of the new rental user based on the output result, so as to match the mobile phone models available for rental for the new rental user. Since the personal data of historical rental users and new rental users include personal identification information data, personal basic information data, personal credit information data and personal family information data, the personal data of new rental users used in the credit risk assessment prediction in the deposit-free online rental method of the present application is more comprehensive than that of the traditional deposit-free rental method. At the same time, the neural network model is used to deeply explore the intrinsic connection between the personal data of historical rental users and rental default situations, which provides a reliable basis for the credit risk assessment prediction of new rental users, thereby effectively improving the accuracy of the credit risk assessment prediction results of new rental users and reducing the risk of bad rental behavior.

[0048] 2. The first credit risk assessment prediction result is compared and verified by the verification unit. If the verification is correct, the first credit risk assessment prediction result can be output as the basis for matching the available mobile phone models for leasing to the new leasing user; if the verification is incorrect, the credit risk assessment prediction result is reviewed by the review unit, and the more reliable one of the first credit risk assessment prediction result or the second credit risk assessment prediction result is selected as the basis for matching the available mobile phone models for leasing to the new leasing user, thereby further improving the accuracy of the credit risk assessment prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a structural block diagram of the deposit-free online rental system in this application. DETAILED DESCRIPTION

[0050] The following is combined with Figure 1 , further details of this application are given.

[0051] The embodiment of the present application discloses a deposit-free online leasing method.

[0052] A deposit-free online leasing method comprises the following steps:

[0053] S1. Collect personal data of historical rental users and rental default situations.

[0054] S2. Train and learn the personal data of historical rental users through the neural network model, adjust the link weights between the neurons that make up the neural network model, and determine the intrinsic connection between personal data and rental defaults.

[0055] The neural network model of the present application can control the sample size of the historical rental user sample set for training and learning, so that the sample size of the historical rental user sample set is maintained at an optimal level, avoiding the problem that the neural network model cannot well grasp and learn the inherent laws and relationships contained in the sample set during the learning process due to too few samples, and at the same time avoiding the problem that the learning time is slowed down and the learning data extraction and induction ability is reduced due to too many samples. It is specifically achieved in the following way: a sample size threshold of historical rental users in a neural network model is set. When the sample size of historical rental users reaches the threshold, before each training and learning of the personal data of historical rental users through the neural network model, a group of historical rental user samples are eliminated. The eliminated historical rental user samples are selected according to the following priority: personal data lacks key information > personal data information is the least complete > historical rental users are the oldest.

[0056] In an optional embodiment, the function expression of the neural network model is:

[0057]

[0058] The error function is

[0059] In the formula, net jt is the input of the jth neuron in the tth layer, y jt is the output of the jth neuron in the tth layer, x i is the output of the i-th neuron in the t-1th layer to the j-th neuron in the t-th layer, w ij is the connection weight between the i-th neuron in the t-1th layer and the j-th neuron in the t-th layer, y pm ′ is the target output, y pm is the actual output, θ jt is the output threshold of the jth neuron in the tth layer, MSE is the mean square error between the output value of the neural network model and the expected value of the sample, P is the number of training samples, and M is the output node.

[0060] The training and learning of personal data of historical rental users by a neural network model and adjusting the link weights between neurons of the neural network model include the following steps:

[0061] The personal data of historical rental users is normalized, and the personal data of historical rental users is converted into sample data of unified standard, and input into the neural network model to calculate the output of the neural network model. The personal data of historical rental users is normalized, and the personal data of historical rental users is converted into sample data of unified standard, which includes the following steps: for non-numeric data, different numeric values ​​in the range of [0, 1] are assigned to each personal data according to the degree of influence on personal credit assessment in actual work; for numeric data, the personal data is converted into a numeric value distributed in (0, 1) by using the normal distribution S function.

[0062] Calculate the mean square error (MSE) between the output value and the sample expected value in the neural network model, then reversely calculate from the output layer to the input layer, adjust the weights w of each link in the direction of reducing the mean square error (MSE), and terminate the training when the mean square error (MSE) is less than the given allowable output threshold (θ).

[0063] S3. Collect personal data of new rental users, the personal data of historical rental users and new rental users include but are not limited to personal identification information data, personal basic information data, personal credit information data and personal family information data, among which personal identification information data include but are not limited to name, gender, ethnicity, place of origin and date of birth data; personal basic information data include but are not limited to education, marriage, housing, transportation, work unit, working hours, nature of occupation, job level, title and time of obtaining the title; personal credit information data include but are not limited to application credit score, annual income, credit card, securities, bank deposits and insurance data; personal family information data include but are not limited to family size, age structure, employment status and social insurance participation data.

[0064] The personal data of the new lease user is input into the trained neural network model, and the first credit risk assessment prediction result of the new lease user is determined based on the output result.

[0065] S4. After inputting the personal data of the new rental user into the trained neural network model and calculating the output result, the verification unit calculates the rental default rate of the new rental user, and determines the second credit risk assessment prediction result of the new rental user. If the second credit risk assessment prediction result is the same as the first credit risk assessment prediction result, the new rental user is matched with a mobile phone model available for rental based on the first credit risk assessment prediction result. Otherwise, the third credit risk assessment prediction result of the new rental user is calculated through the review unit, and the one of the first credit risk assessment prediction result and the second credit risk assessment prediction result that is the same as the third credit risk assessment prediction result is selected as the basis for matching the new rental user with a mobile phone model available for rental.

[0066] The first credit risk assessment prediction result is compared and verified by the verification unit. If the verification is correct, the first credit risk assessment prediction result can be output as a basis for matching available mobile phone models for rental to new rental users; if the verification is incorrect, the credit risk assessment prediction result is reviewed by the review unit, and the more reliable one of the first credit risk assessment prediction result or the second credit risk assessment prediction result is selected as the basis for matching available mobile phone models for rental to new rental users, thereby greatly improving the accuracy of the credit risk assessment prediction.

[0067] In an optional embodiment, the calculation of the lease default rate of the new lease user by the verification unit includes the following steps: substituting the personal data of the historical lease user into the Logit prediction model, selecting the optimal personal data weight according to the SPSS software, substituting the personal data of the new lease user into the Logit distribution function, and calculating the lease default rate of the new lease user;

[0068] The Logit prediction model is:

[0069] logitP=a0+a1y1+a2y2+...+a m y m ,

[0070] Where P is the rental default rate of historical rental users, y1, y2, ..., y m is the personal data of historical rental users, a0, a1, ..., a m The weight of personal data;

[0071] The Logit distribution function is:

[0072]

[0073] Where P is the rental default rate of new tenants, y1, y2, ..., y m is the personal data of the new rental user, a0, a1, ..., a mis the optimal individual data weight selected by SPSS software.

[0074] In an optional embodiment, the calculation of the third credit risk assessment prediction result of the new lease user by the review unit includes the following steps:

[0075] Assume that the credit risk assessment discriminant function is: Y = C1X1 + C2X2 + ... + C p X p ,

[0076] Where Y is the credit risk assessment indicator, X1, X2, ..., X P The personal data of the rental user, C1, C2, ..., C P The weight of personal data;

[0077] The historical rental user samples are divided into two groups according to the credit risk assessment of good and poor. The group with good credit risk assessment is group A, the sample size is n1 groups, and the average credit risk assessment index is The group with poor credit risk assessment is group B, with a sample size of n2 groups and an average credit risk assessment index of

[0078] Calculate the weighted average of the average credit risk assessment index Y c :

[0079]

[0080] The personal data of the new rental user X 0 1, X 0 2, ..., X 0 P Substitute the credit risk assessment discriminant function to obtain the credit risk assessment index of the new rental user:

[0081] Y0=C1X 0 1+C2X 0 2+...+C p X 0 p ;

[0082] when When Y0>Y c , then the credit risk assessment of the new lease user is excellent. If Y0 <Y c , then the credit risk assessment of the new lease user is poor;

[0083] when When Y0 <Y c , then the credit risk assessment of the new lease user is excellent. If Y0>Y c , then the credit risk assessment of the new lease user is poor.

[0084] S5. Receive the rental demand of the new rental user, and match the mobile phone models available for rental for the new rental user in combination with the first credit risk assessment prediction result or the second credit risk assessment prediction result selected in step S4, and generate a rental order according to the selection of the new rental user. The receiving of the rental demand of the new rental user and matching the mobile phone models available for rental for the new rental user include the following steps: constructing a user profile according to the purchase history, browsing behavior and evaluation feedback information of the new rental user, analyzing the rental preferences of the new rental user according to the rental demand of the new rental user by cluster analysis and association rule methods, and generating a list of mobile phone models matching the new rental user by collaborative filtering and content-based recommendation algorithms.

[0085] The embodiment of the present application also discloses a deposit-free online rental system.

[0086] Reference Figure 1 , a deposit-free online rental system, comprising:

[0087] The information collection unit is used to collect personal data of historical rental users and rental default situations.

[0088] The first credit risk assessment prediction unit is used to train and learn the personal data of historical rental users through a neural network model, collect the personal data of new rental users, and calculate the first credit risk assessment prediction result of the new rental user.

[0089] The rental matching unit is used to receive the rental demand of the new rental user, match the mobile phone models available for rental for the new rental user, and generate a rental order according to the selection of the new rental user.

[0090] The verification unit is used to calculate the lease default rate of the new lease user, determine the second credit risk assessment prediction result of the new lease user, compare the second credit risk assessment prediction result with the first credit risk assessment prediction result, and output the first credit risk assessment prediction result to the lease matching unit as a basis for matching the new lease user with available lease mobile phone models.

[0091] The review unit is used to calculate the third credit risk assessment prediction result of the new rental user, compare the third credit risk assessment prediction result with the first credit risk assessment prediction result and the second credit risk assessment prediction result, and output the first credit risk assessment prediction result or the second credit risk assessment prediction result to the rental matching unit as a basis for matching the new rental user with available mobile phone models.

[0092] The embodiment of the present application also provides a corresponding device and a computer storage medium for implementing the solution provided in the embodiment of the present application. The device includes a memory and a processor, the memory is used to store instructions or codes, and the processor is used to execute instructions or codes so that the device executes the method described in the embodiment of the present application. The computer storage medium stores the code, and when the code is executed, the device running the code implements the method described in the embodiment of the present application.

[0093] It can be seen from the description of the above implementation methods that those skilled in the art can clearly understand that all or part of the steps in the above embodiment methods can be implemented by means of software plus a general hardware platform. Based on such an understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a storage medium, such as a read-only memory, a magnetic disk, and an optical disk.

[0094] The embodiments of this specific implementation are all preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. The same components are represented by the same figure marks. Therefore, any equivalent changes made based on the structure, shape, and principle of the present application should be included in the protection scope of the present application.

Claims

1. A deposit-free online leasing method, characterized in that: The following steps are involved: Collect personal data of historical rental users and rental defaults; The personal data of historical rental users are trained and learned through the neural network model, and the link weights between the neurons of the neural network model are adjusted to determine the intrinsic connection between personal data and rental defaults; Collecting personal data of new renters, inputting the personal data of new renters into the trained neural network model, and determining a first credit risk assessment prediction result of the new renter based on the output result, wherein the personal data of the historical renters and the new renters include personal identification information data, personal basic information data, personal credit information data, and personal family information data; Receive leasing demands from new leasing users, match available mobile phone models for them, and generate leasing orders based on their choices.

2. A deposit-free online leasing method according to claim 1, characterized in that: The personal identification information data includes name, gender, ethnicity, place of origin and date of birth; the personal basic information data includes education, marriage, housing, means of transportation, work unit, working hours, nature of occupation, job level, title and time of obtaining the title; the personal credit information data includes application credit score, annual income, credit card, securities, bank deposits and insurance data; the personal family information data includes family size, age structure, employment status and social insurance participation data.

3. A deposit-free online leasing method according to claim 1, characterized in that: A sample number threshold of historical rental users in the neural network model is set. When the sample number of historical rental users reaches the threshold, a group of historical rental user samples is eliminated before each training and learning of the personal data of historical rental users through the neural network model. The eliminated historical rental user samples are selected according to the following priority: personal data lacks key information > personal data information is the least complete > historical rental users are the oldest.

4. A deposit-free online leasing method according to claim 1, characterized in that: The receiving of the rental demand of the new rental user and matching the mobile phone models available for rental for the new rental user comprises the following steps: constructing a user portrait according to the purchase history, browsing behavior and evaluation feedback information of the new rental user, analyzing the rental preference of the new rental user according to the rental demand of the new rental user by cluster analysis and association rule methods, and generating a list of mobile phone models matching the new rental user by collaborative filtering and content-based recommendation algorithms.

5. A deposit-free online leasing method according to claim 1, characterized in that: The functional expression of the neural network model is: The error function is In the formula, net jt is the input of the jth neuron in the tth layer, y jt is the output of the jth neuron in the tth layer, x i is the output of the i-th neuron in the t-1th layer to the j-th neuron in the t-th layer, w ij is the connection weight between the i-th neuron in the t-1th layer and the j-th neuron in the t-th layer, y pm ′ is the target output, y pm is the actual output, θ jt is the output threshold of the jth neuron in the tth layer, MSE is the mean square error between the output value of the neural network model and the expected value of the sample, P is the number of training samples, and M is the output node; The training and learning of the personal data of historical rental users by the neural network model and adjusting the link weights between the neurons of the neural network model include the following steps: Normalize the personal data of historical rental users, convert the personal data of historical rental users into sample data of unified standard, input the data into the neural network model, and calculate the output of the neural network model; Calculate the mean square error (MSE) between the output value and the sample expected value in the neural network model, then reversely calculate from the output layer to the input layer, adjust the weights w of each link in the direction of reducing the mean square error (MSE), and terminate the training when the mean square error (MSE) is less than the given allowable output threshold (θ).

6. A deposit-free online leasing method according to claim 5, characterized in that: The normalization of the personal data of historical rental users to convert the personal data of historical rental users into sample data of unified standards includes the following steps: for non-numeric data, each personal data is assigned a different numeric value in the range of [0, 1] according to its influence on personal credit assessment in actual work; for numeric data, the personal data is converted into a numeric value distributed in (0, 1) by using the normal distribution S function.

7. A deposit-free online leasing method according to claim 1, characterized in that: After inputting the personal data of the new rental user into the trained neural network model and calculating the output result, the rental default rate of the new rental user is calculated through the verification unit, and the second credit risk assessment prediction result of the new rental user is determined. If the second credit risk assessment prediction result is the same as the first credit risk assessment prediction result, the new rental user is matched with a mobile phone model available for rental based on the first credit risk assessment prediction result. Otherwise, the third credit risk assessment prediction result of the new rental user is calculated through the review unit, and the one of the first credit risk assessment prediction result and the second credit risk assessment prediction result that is the same as the third credit risk assessment prediction result is selected as the basis for matching the new rental user with a mobile phone model available for rental.

8. A deposit-free online leasing method according to claim 7, characterized in that: The verification unit calculates the lease default rate of the new lease user The method includes the following steps: substituting the personal data of historical rental users into the Logit prediction model, selecting the optimal personal data weight according to SPSS software, substituting the personal data of new rental users into the Logit distribution function, and calculating the rental default rate of new rental users; The Logit prediction model is: logitP=a0+a1y1+a2y2+...+a m y m , Where P is the rental default rate of historical rental users, y1, y2, ..., y m is the personal data of historical rental users, a0, a1, ..., a m The weight of personal data; The Logit distribution function is: Where P is the rental default rate of new tenants, y1, y2, ..., y m is the personal data of the new rental user, a0, a1, ..., a m is the optimal individual data weight selected by SPSS software.

9. A deposit-free online leasing method according to claim 7, characterized in that: The calculation of the third credit risk assessment prediction result of the new lease user by the review unit comprises the following steps: Assume that the credit risk assessment discriminant function is: Y = C1X1 + C2X2 + ... + C p X p , Where Y is the credit risk assessment indicator, X1, X2, ..., X P The personal data of the rental user, C1, C2, ..., C P The weight of personal data; The historical rental user samples are divided into two groups according to the credit risk assessment of good and poor. The group with good credit risk assessment is group A, the sample size is n1 groups, and the average credit risk assessment index is The group with poor credit risk assessment is group B, with a sample size of n2 groups and an average credit risk assessment index of Calculate the weighted average of the average credit risk assessment index Y c : The personal data of the new rental user X 0 1, X 0 2, ..., X 0 P Substitute the credit risk assessment discriminant function to obtain the credit risk assessment index of the new rental user: Y0=C1X 0 1+C2X 0 2+...+C p X 0 p ; when When Y0>Y c , then the credit risk assessment of the new lease user is excellent. If Y0 <Y c , then the credit risk assessment of the new lease user is poor; when When Y0 <Y c , then the credit risk assessment of the new lease user is excellent. If Y0>Y c , then the credit risk assessment of the new rental user is poor.

10. A deposit-free online rental system, characterized in that: include: An information collection unit, used to collect personal data of historical rental users and rental default situations; A first credit risk assessment prediction unit, used to train and learn personal data of historical rental users through a neural network model, collect personal data of new rental users, and calculate a first credit risk assessment prediction result of the new rental user; The rental matching unit is used to receive the rental demand of the new rental user, match the mobile phone models available for rental for the new rental user, and generate a rental order according to the selection of the new rental user.

Images