DEA cross sorting-based logistics project risk evaluation and sorting method

Abstract

The invention discloses a logistics project risk evaluation and sorting method based on DEA cross sorting, and belongs to the technical field of logistics risk evaluation, and the method comprises the steps: collecting real data in a third-party logistics platform; according to risk scenes faced by the third-party logistics platform at different nodes, constructing different logistics risk characteristic indexes; carrying out relative evaluation on the risk conditions of the logistics items based on a DEA cross sorting method, and sorting the logistics items according to a risk evaluation result; according to the sorting result, the logistics items with high risks are judged, and corresponding prevention and control measures are taken; according to the logistics project risk evaluation and sorting method based on DEA cross sorting, objectification of logistics projects with different risks and rapid identification of high-risk logistics projects of decision makers are achieved, universality is achieved, and the method can be expanded and applied to performance or risk sorting scenes in other fields.

Description

Technical Field

[0001] This invention relates to the field of logistics risk assessment technology, and in particular to a method for risk assessment and ranking of logistics projects based on DEA cross-ranking. Background Technology

[0002] In the current new retail landscape, logistics platforms effectively integrate resources to provide businesses and consumers with more convenient, efficient, and secure logistics services.

[0003] Logistics project risk assessment is of great significance to logistics platforms: By effectively assessing and classifying logistics project risks, logistics platforms can better manage logistics projects, reduce project risk losses, and thus improve project success rates and investment returns; through risk management, logistics platforms can quickly respond to market changes, adjust logistics strategies, and maintain competitiveness; the results of risk assessment provide a basis for the decision-making of logistics platforms, helping enterprises to make more scientific decisions when facing uncertainty.

[0004] However, logistics platform companies face numerous challenges in risk assessment for logistics projects. First, while these companies possess massive amounts of data, their risk assessment methods may be unscientific or overly simplistic, lacking systematic and quantitative risk assessment tools and models. Second, risk assessment for logistics projects has unique characteristics, including a large number of assessment targets and the potential for decision-maker bias. Therefore, risk assessment models and methods tailored to the characteristics of logistics projects urgently need research. Summary of the Invention

[0005] The purpose of this invention is to provide a method for risk assessment and ranking of logistics projects based on DEA cross-sorting, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, this invention provides a method for risk assessment and ranking of logistics projects based on DEA cross-ranking, comprising the following steps:

[0007] S1. Collect real data from third-party logistics platforms;

[0008] S2. Based on the risk scenarios faced by third-party logistics platforms at different nodes, construct different logistics risk characteristic indicators;

[0009] S3. The risk status of logistics projects is relatively evaluated based on the DEA cross-ranking method, and the logistics projects are ranked according to the risk evaluation results.

[0010] S4. Based on the ranking results, identify logistics projects with higher risks and take corresponding prevention and control measures.

[0011] Preferably, the logistics risk characteristic indicators of S2 include real-time anomaly rate, trajectory missing rate, overlapping anomaly rate, overload rate, contract anomaly rate, defect rate, and settlement anomaly rate, and the calculation logic is as follows:

[0012] Real-time anomaly rate = Number of real-time anomaly orders / Total number of orders;

[0013] Overlapping anomaly rate = Number of overlapping anomaly orders / Total number of orders;

[0014] Track anomaly rate = Number of track anomaly orders / Total number of orders;

[0015] Settlement anomaly rate = Amount of settlement anomaly waybills / Total payment amount;

[0016] Defect rate = Special approval bill payment amount / Total payment amount;

[0017] Contract anomaly rate = Number of contract information change orders / Total number of waybills;

[0018] Overload rate = Number of waybills whose actual load exceeds the vehicle's rated load capacity / Total number of waybills;

[0019] Preferably, the specific steps of S3 are as follows:

[0020] S31. Based on the selected logistics risk characteristic indicators, determine the input variables and output variables. Take the real-time anomaly rate, trajectory missing rate, overlap anomaly rate, overload rate, contract anomaly rate, defect rate and settlement anomaly rate as input variables x, and construct virtual output y for each decision unit DMU, ​​with a value of 1.

[0021] S32. Calculate the self-assessment risk coefficient;

[0022] S33. Calculate the risk coefficient of other assessments and construct the interval cross risk coefficient matrix;

[0023] S34. Introduce a risk ranking function, calculate the cross-ranking probability of each logistics project based on the interval cross-risk coefficient matrix, and construct a cross-ranking matrix.

[0024] S35. Construct the preference matrix;

[0025] S36. The clustering preference matrix yields the final ranking result.

[0026] Preferably, the formula for calculating the self-assessed risk coefficient of logistics project a in S32 is as follows:

[0027]

[0028] Among them, 1-P aa P represents the self-assessed risk coefficient. aa u represents the self-evaluation efficiency calculated by the DEA model. a The weight vector representing the output of logistics project a. Representing DMU a The input weight vector, X a Indicates DMU a The risk indicator vector is the a-th column of the m x n risk indicator matrix X. X j It is DMU j The risk indicator vector is the j-th column of the risk indicator matrix.

[0029] Preferably, the formula for calculating the risk coefficient in S33 is as follows:

[0030]

[0031]

[0032] in, This represents the lower bound of the risk coefficient of logistics project a for evaluating logistics project k, assuming the self-assessed risk coefficient of logistics project a remains unchanged. 1- P ak

[0033] This represents the upper bound of the risk coefficient of logistics project a for evaluating logistics project k, assuming the self-assessed risk coefficient of logistics project a remains unchanged.

[0034] The interval risk coefficient matrix A is:

[0035]

[0036] In this matrix, the a-th row represents the risk coefficient range of n logistics projects evaluated by logistics project a.

[0037] Preferably, S34 specifically includes:

[0038] Under the evaluation of logistics project a, the risk ranking function for logistics project j is:

[0039] rank(ξ j )=1+∑ k ρ(ξ j <ξ k (8)

[0040] Where, ξ j For the evaluation of logistics project a, the cross-risk coefficient of logistics project j can be considered as being within the interval It follows a uniform distribution and has a density function of f. j (ξ j ξ is an independent random variable, ρ(·) is an indicator function, and ξ is an independent random variable. j <ξ k When true, ρ(true) = 1; otherwise, ρ(false) = 0.

[0041] The cross-order probability of logistics item j is:

[0042]

[0043] in, Let ξ represent the probability that logistics project j is ranked rth under the evaluation of logistics project a, and D represent a set of n logistics projects evaluated by logistics project a. f(ξ) is the joint density function, calculated as follows:

[0044]

[0045] Among them, f j (ξ j Let ξ be the density function, and ξ be ξ1, ξ2, ..., ξ3. n ;

[0046] The cross-sort matrix is:

[0047]

[0048] Where n is the number of logistics projects.

[0049] Preferably, the specific formula for the preference matrix in S35 is as follows:

[0050]

[0051] in,

[0052] Preferably, the S36 clustering preference matrix specifically includes:

[0053]

[0054] μ1≥2μ2≥3μ3≥…≥nμ n (15)

[0055]

[0056] Where, β j μ is the comprehensive preference value for logistics project j, used for final risk ranking. r The weight is the ranking position r.

[0057] Preferably, the specific steps of S4 are as follows:

[0058] S41. Based on the comprehensive preference value β j Screening high-risk logistics projects and selecting β j Large-value logistics projects are considered high-risk targets;

[0059] S42. For the selected high-risk logistics projects, check their corresponding waybill data to locate the key nodes of the risk source;

[0060] S43. Take corresponding control measures for the nodes where risks originate.

[0061] Therefore, the present invention employs the above-mentioned method for risk assessment and ranking of logistics projects based on DEA cross-ranking, which has the following beneficial effects:

[0062] (1) By depicting the real business scenarios of third-party logistics platform companies, a series of risk characteristic indicators have been constructed, which can fully reflect the risks faced by each key node in the business scenario and have authenticity and scientificity.

[0063] (2) The DEA cross-ranking method is adopted, and self-assessment risk coefficient and other-assessment risk coefficient are introduced. The possibility of evaluation results under each weight is fully considered, making the evaluation results more realistic and objective.

[0064] (3) Based on the evaluation results obtained by self-evaluation and peer evaluation, a ranking function, cross-ranking matrix, and preference matrix are further introduced, and the final result of the risk ranking of logistics projects is obtained through the preference aggregation model, making the results more intuitive and enabling decision-makers to quickly screen out high-risk logistics projects and take corresponding measures.

[0065] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0066] Figure 1 This is a flowchart of a logistics project risk assessment and ranking method based on DEA cross-sorting, as described in this invention.

[0067] Figure 2 This is a schematic diagram of a real business process of a third-party logistics platform for a logistics project risk assessment and ranking method based on DEA cross-sorting, as described in this invention.

[0068] Figure 3 This is a schematic diagram illustrating the steps of a logistics project risk assessment and ranking method based on DEA cross-sorting, according to the present invention. Detailed Implementation

[0069] The following detailed description of embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0070] Example

[0071] like Figure 1 As shown, this invention provides a method for risk assessment and ranking of logistics projects based on DEA cross-ranking, including the following steps:

[0072] S1. Collect real data from third-party logistics platforms.

[0073] In this embodiment, all collected data are real data from third-party logistics platform business scenarios. Figure 2 It showcases the actual business processes of a third-party logistics platform, and the collected data includes key nodes in the business processes, making it representative.

[0074] S2. Based on the risk scenarios faced by third-party logistics platforms at different nodes, construct different logistics risk characteristic indicators.

[0075] First, identify the risk scenarios faced at each key node. Among the three key nodes of the shipper creating an order on the platform, the driver accepting the order, and the departure, the risk scenarios that may occur are: the driver accepting the order later than the departure time (late order acceptance) or the shipper creating the order later than the driver's departure time (late order creation). These risks are all abnormalities in the time sequence logic.

[0076] Between the two critical points of origin and destination, the following risk scenarios may occur: missing vehicle transport routes, failing to fully cover both the origin and destination; overlapping transport times for the same vehicle in two waybills, leading the platform to prohibit carpooling; and the actual weight of goods transported by the driver exceeding the vehicle's rated load capacity.

[0077] Between the two key points of the driver uploading the waybill and the platform paying the driver the freight, the following risk scenarios may occur: the transportation agreement may be changed, and the information such as freight, driver information, license plate, transportation route, quantity and name of goods at the time of order creation may be inconsistent with the waybill uploaded by the driver; third-party logistics platform companies may manually approve some flawed waybills due to business relationships; when the platform pays the freight to the driver, the payee is not the driver or vehicle owner, which may involve the return of funds.

[0078] The constructed risk characteristic indicators are based on the project level. A logistics project contains multiple orders. Addressing the potential risk scenarios at each of the key nodes, the logistics risk characteristic indicators include real-time anomaly rate, trajectory missing rate, overlapping anomaly rate, overload rate, contract anomaly rate, defect rate, and settlement anomaly rate. The calculation logic is as follows:

[0079] Real-time anomaly rate = Number of real-time anomaly orders / Total number of orders;

[0080] Overlapping anomaly rate = Number of overlapping anomaly orders / Total number of orders;

[0081] Track anomaly rate = Number of track anomaly orders / Total number of orders;

[0082] Settlement anomaly rate = Amount of settlement anomaly waybills / Total payment amount;

[0083] Defect rate = Special approval bill payment amount / Total payment amount;

[0084] Contract anomaly rate = Number of contract information change orders / Total number of waybills;

[0085] Overload rate = Number of waybills whose actual load exceeds the vehicle's rated load capacity / Total number of waybills;

[0086] In this embodiment, the risk assessment indicators are constructed by including the risk scenarios faced by each key node in the real business scenarios of the third-party logistics platform. Since there are significant differences in the number of waybills and payment amounts of various logistics projects, the risk indicators are constructed in the form of ratios, which makes them more realistic.

[0087] S3. The risk status of logistics projects is relatively evaluated based on the DEA cross-ranking method, and the logistics projects are ranked according to the risk evaluation results.

[0088] like Figure 3 The diagram shown is a framework diagram of the DEA cross-ranking method; it can be divided into the following steps: determining input-output indicators, calculating self-assessment risk coefficients, calculating other-assessment risk coefficients and constructing an interval cross-ranking risk coefficient matrix, calculating cross-ranking and constructing a cross-ranking matrix, constructing a preference matrix, and aggregating the preference matrix to obtain the final ranking result.

[0089] The specific steps are as follows:

[0090] S31. Based on the selected logistics risk characteristic indicators, determine the input variables and output variables. In this embodiment, since the defined risk characteristic indicators are all negative indicators, the real-time anomaly rate, trajectory missing rate, overlap anomaly rate, overload rate, contract anomaly rate, defect rate, and settlement anomaly rate are used as input variables x. In this embodiment, there are no naturally positive indicators. A virtual output y is constructed for each decision unit DMU, ​​with a value of 1, indicating that all DMUs complete the same task.

[0091] S32. Calculate the self-assessment risk coefficient.

[0092] Select n logistics projects as decision-making units (DMUs) and calculate the self-assessed risk coefficient of logistics project a:

[0093]

[0094] Among them, 1-P aa P represents the self-assessed risk coefficient. aa u represents the self-evaluation efficiency calculated by the DEA model. aThe weight vector representing the output of logistics project a. Representing DMU a The input weight vector, X a Indicates DMU a The risk indicator vector is the a-th column of the m x n risk indicator matrix X. X j It is DMU j The risk indicator vector is the j-th column of the risk indicator matrix.

[0095] In this embodiment, the input indicators are seven risk characteristic indicators, so m = 7; the fractional programming problem can be converted into a linear programming problem and solved using MATLAB; solving the fractional programming problem n times, where a = 1,...n, in this embodiment, the optimal weights that minimize the risk coefficient for each logistics project during self-assessment can be obtained.

[0096] S33. Calculate the risk coefficient of the other party and construct the interval cross risk coefficient matrix.

[0097] 1-P calculated from the above formula aa In this embodiment, each logistics project selects the weight that minimizes its own risk coefficient when conducting self-assessment. This results in many logistics projects having a self-assessment coefficient of 0, making it impossible to further distinguish the risk levels of these projects. Therefore, external assessment is introduced to calculate the external assessment risk coefficient, as shown in the following formula:

[0098]

[0099] in, This represents the lower bound of the risk coefficient of logistics project a for evaluating logistics project k, assuming the self-assessed risk coefficient of logistics project a remains unchanged. 1- P ak This represents the upper bound of the risk coefficient of logistics project a for evaluating logistics project k, assuming the self-assessed risk coefficient of logistics project a remains unchanged.

[0100] In this embodiment, the optimal weight obtained by logistics project a during self-evaluation is... It is not unique; there may be multiple optimal solutions.

[0101] In the above formula It is the minimum risk coefficient of logistics project k obtained by logistics project a in evaluating logistics project k, while keeping its self-assessed risk coefficient unchanged; 1- P akThis refers to the maximum risk coefficient of logistics project k obtained by logistics project a when evaluating logistics project k, assuming that the self-assessed risk coefficient remains unchanged. In this embodiment, the risk coefficient range for logistics project a's evaluation of logistics project k is...

[0102] Let a = 1,...,n, k = 1,...,n, then we can obtain an n×n interval cross-risk coefficient matrix; for example, when a = 1, k = 1,...,n, we can obtain the risk coefficient intervals obtained when logistics project 1 evaluates all logistics projects (including itself), which is obviously...

[0103] The constructed interval risk coefficient matrix A is shown below:

[0104]

[0105] In this matrix, the a-th row represents the risk coefficient range of n logistics projects evaluated by logistics project a.

[0106] S34. The interval cross risk coefficient matrix obtained in the above steps can only show the score range where the risk coefficients of other logistics projects may be located under the evaluation of a certain logistics project, and cannot give the actual ranking; introduce a risk ranking function, calculate the cross ranking probability of each logistics project based on the interval cross risk coefficient matrix, and construct a cross ranking matrix.

[0107] In this embodiment, under the evaluation of logistics project a, the risk ranking of each logistics project as an integer is defined by a risk ranking function. The risk ranking function for logistics project j is:

[0108] rank(ξ j )=1+∑ k ρ(ξ j <ξ k (8)

[0109] Where, ξ j Let be the cross-risk coefficient of logistics project j under the evaluation of logistics project a with specific weight selection, and ρ(·) be the indicator function. Check the condition (ξ) within the parentheses. j <ξ k Whether ξ is true or false j <ξ k ρ(true) = 1 when true, otherwise ρ(false) = 0; ∑ k ρ(ξ j <ξ k This summation term calculates the number of other logistics projects with a risk coefficient strictly greater than that of logistics project j, under this specific weight selection; 1+∑ k ρ(ξj <ξ k This represents the risk ranking of logistics project j under this specific weight selection. It should be noted that the higher the ranking, the greater the risk coefficient and the higher the risk of the logistics project.

[0110] Under the evaluation of logistics project a (a=1,...,n), let the cross-risk coefficient of logistics project j (j=1,...,n) be ξ. j ξ j The range of values ​​is

[0111] Assume ξ j Is Let f be an independent random variable that follows a uniform distribution, and its probability density function be f. j (ξ j );

[0112] The cross-order probability of logistics item j is:

[0113]

[0114] in, Let represent the probability that logistics project j is ranked rth under the evaluation of logistics project a, and D represent a set of n logistics projects evaluated by logistics project a.

[0115] Integration domain (D:rank(ξ)) j )=r) means that this integral is in all rank(ξ) satisfying the condition. j The evaluation is performed on the space of an n-dimensional vector ξ of r, where D represents the space of possible risk coefficient values ​​for all n logistics projects under the evaluation of logistics project a. In this embodiment, logistics project a evaluates all logistics projects, and the resulting interval risk coefficient can be obtained from the a-th row of the interval risk coefficient matrix A; f(ξ) is the joint density function, calculated as follows:

[0116]

[0117] Among them, f j (ξ j Let ξ be the density function, and ξ be ξ1, ξ2, ..., ξ3. n ;

[0118] The integral limitation of this embodiment only considers combinations of ξ where logistics project j (j = 1, ..., n) is ranked r under the evaluation of logistics project a (a = 1, ..., n); in this embodiment, formula (9) means the total probability that the risk ranking of logistics project j (j = 1, ..., n) is exactly r among all the possible risk coefficients that logistics project j (j = 1, ..., n) can obtain under the evaluation of logistics project a (a = 1, ..., n); obviously, In this embodiment, under the evaluation of logistics project a, the probabilities of all possible risk rankings r of logistics project j are summed, and the sum of the probabilities of all rankings is 1.

[0119] In this embodiment, under the evaluation of logistics project a, the total probability of all logistics projects obtaining a specific ranking r is 1; therefore, the cross-ranking matrix is ​​as follows:

[0120]

[0121] Where n is the number of logistics projects.

[0122] S35. Construct the preference matrix.

[0123] Based on the cross-ranking matrix given above, a preference matrix is ​​constructed. In this embodiment, the purpose of constructing the preference matrix is ​​to summarize the probability information of all logistics items a (a = 1, ..., n) acting as evaluators for all evaluated logistics items j (j = 1, ..., n) at each ranking position r, forming a global preference view; the specific formula of the preference matrix is:

[0124]

[0125] in, In this embodiment w jr This indicates the sum of the total probabilities that logistics project j (j=1,...,n) is ranked in the r-th position under the evaluation of all logistics projects; it should be noted that w here jr ∈[0,n], because it is the summation of n probability values.

[0126] S36. The clustering preference matrix yields the final ranking result.

[0127] In this embodiment, based on the preference matrix W containing probability information obtained in the previous step, a comprehensive preference value β is calculated using the following preference aggregation model. j The risk of all logistics projects is finally ranked using a sequence of (j = 1, ..., n); the specific formula is as follows:

[0128]

[0129] μ1≥2μ2≥3μ3≥…≥nμ n (15)

[0130]

[0131] Where, β j μ is the comprehensive preference value for logistics project j, used for final risk ranking. r The weight of risk ranking position r; in this embodiment, the preference value β j The larger the value, the higher the risk level of logistics project j; the greater the risk, the higher the ranking.

[0132] This indicates that for each specific logistics project j, the weight μ of all its risk ranking positions r is calculated. r The total probability w multiplied by the risk ranking position r of the logistics project jr The sum of; where constraints Ensured that the calculated β j Comparability is possible, preventing excessively large weights; constraints: μ1≥2μ2≥3μ3≥…≥nμ n This indicates that in this embodiment, the weight of risk ranking position 1 is at least 2 (2 / 1) times the weight of ranking position 2, the weight of risk ranking position 2 is at least 1.5 (3 / 2) times the weight of ranking position 3, and so on, with the weight of risk ranking position r being at least ((r+1) / r) times the weight of ranking position r+1. (Constraint) Ensure that the weight assigned to the last position (r=n) is a positive number. This guarantees that information about all sorting positions is taken into account, avoiding μ. n =0 leads to the loss of last-digit information; while the objective function This means that the sum is calculated across all possible variables μ. r The minimum value (under the constraints); if β is under the most unfavorable weight distribution. j (j=1,...,n) still achieved a large value, which in this embodiment means that the logistics project j has a high risk ranking and is of high risk.

[0133] S4. Based on the ranking results, identify logistics projects with higher risks and take corresponding prevention and control measures.

[0134] S41. Based on the comprehensive preference value β j Screening high-risk logistics projects and selecting β j Large-value logistics projects are considered high-risk targets;

[0135] S42. For the selected high-risk logistics projects, check their corresponding waybill data to locate the key nodes of the risk source;

[0136] S43. Take corresponding control measures for the nodes where risks originate.

[0137] Therefore, this invention adopts the above-mentioned method for risk assessment and ranking of logistics projects based on DEA cross-ranking. By collecting real data from third-party logistics platforms and constructing multi-node risk characteristic indicators, it uses the DEA cross-ranking method to construct interval cross-risk coefficients, cross-ranking, and preference matrices from the dimensions of self-assessment and other-assessment, and outputs the final ranking. This achieves objectification of logistics projects with different risks and enables decision-makers to quickly identify high-risk logistics projects. It also has universality and can be extended to performance or risk ranking scenarios in other fields.

[0138] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for risk assessment and ranking of logistics projects based on DEA cross-ranking, characterized in that, Includes the following steps: S1. Collect real data from third-party logistics platforms; S2. Based on the risk scenarios faced by third-party logistics platforms at different nodes, construct different logistics risk characteristic indicators; S3. The risk status of logistics projects is relatively evaluated based on the DEA cross-ranking method, and the logistics projects are ranked according to the risk evaluation results. S4. Based on the ranking results, identify logistics projects with higher risks and take corresponding prevention and control measures.

2. The method for risk assessment and ranking of logistics projects based on DEA cross-ranking according to claim 1, characterized in that, S2's logistics risk characteristic indicators include real-time anomaly rate, trajectory missing rate, overlapping anomaly rate, overload rate, contract anomaly rate, defect rate, and settlement anomaly rate. The calculation logic is as follows: Real-time anomaly rate = Number of real-time anomaly orders / Total number of orders; Overlapping anomaly rate = Number of overlapping anomaly orders / Total number of orders; Track anomaly rate = Number of track anomaly orders / Total number of orders; Settlement anomaly rate = Amount of settlement anomaly waybills / Total payment amount; Defect rate = Special approval bill payment amount / Total payment amount; Contract anomaly rate = Number of contract information change orders / Total number of waybills; Overload rate = Number of waybills whose actual load exceeds the vehicle's rated load capacity / Total number of waybills.

3. The method for risk assessment and ranking of logistics projects based on DEA cross-ranking according to claim 1, characterized in that, The specific steps for S3 are as follows: S31. Based on the selected logistics risk characteristic indicators, determine the input variables and output variables. Take the real-time anomaly rate, trajectory missing rate, overlap anomaly rate, overload rate, contract anomaly rate, defect rate and settlement anomaly rate as input variables x, and construct virtual output y for each decision unit DMU, ​​with a value of 1. S32. Calculate the self-assessment risk coefficient; S33. Calculate the risk coefficient of other assessments and construct the interval cross risk coefficient matrix; S34. Introduce a risk ranking function, calculate the cross-ranking probability of each logistics project based on the interval cross-risk coefficient matrix, and construct a cross-ranking matrix. S35. Construct the preference matrix; S36. The clustering preference matrix yields the final ranking result.

4. The method for risk assessment and ranking of logistics projects based on DEA cross-ranking according to claim 3, characterized in that, The formula for calculating the self-assessed risk coefficient of logistics project S32a is as follows: Among them, 1-P aa P represents the self-assessed risk coefficient. aa u represents the self-evaluation efficiency calculated by the DEA model. a The weight vector representing the output of logistics project a. Representing DMU a The input weight vector, X a Indicates DMU a The risk indicator vector is the a-th column of the m x n risk indicator matrix X. X j It is DMU j The risk indicator vector is the j-th column of the risk indicator matrix.

5. The method for risk assessment and ranking of logistics projects based on DEA cross-ranking according to claim 4, characterized in that, The formula for calculating the risk coefficient in S33 is as follows: in, This represents the lower bound of the risk coefficient of logistics project a for evaluating logistics project k, assuming the self-assessed risk coefficient of logistics project a remains unchanged. 1- P ak This represents the upper bound of the risk coefficient of logistics project a for evaluating logistics project k, assuming the self-assessed risk coefficient of logistics project a remains unchanged. The interval risk coefficient matrix A is: In this matrix, the a-th row represents the risk coefficient range of n logistics projects evaluated by logistics project a.

6. The method for risk assessment and ranking of logistics projects based on DEA cross-ranking according to claim 5, characterized in that, S34 specifically includes: Under the evaluation of logistics project a, the risk ranking function for logistics project j is: rank(ξ j )=1+∑ k p(x j <ξ k ) (8) Where, ξ j Let ρ(·) be the cross-risk coefficient of logistics project j under the evaluation of logistics project a, and let ξ be the indicator function. j <ξ k When true, ρ(true) = 1; otherwise, ρ(false) = 0. The cross-order probability of logistics item j is: in, Let ξ represent the probability that logistics project j is ranked rth under the evaluation of logistics project a, and D represent a set of n logistics projects evaluated by logistics project a. f(ξ) is the joint density function, calculated as follows: Among them, f j (ξ j Let ξ be the density function, and ξ be ξ1, ξ2, ..., ξ3. n ; The cross-sort matrix is: Where n is the number of logistics projects.

7. The method for risk assessment and ranking of logistics projects based on DEA cross-ranking according to claim 6, characterized in that, The specific formula for the preference matrix in S35 is as follows: in, 8. The method for risk assessment and ranking of logistics projects based on DEA cross-ranking according to claim 7, characterized in that, The S36 clustering preference matrix specifically includes: μ1≥2μ2≥3μ3≥…≥nμ n (15) Where, β j μ is the comprehensive preference value for logistics project j, used for final risk ranking. r The weight is the ranking position r.

9. The method for risk assessment and ranking of logistics projects based on DEA cross-ranking according to claim 8, characterized in that, The specific steps for S4 are as follows: S41. Based on the comprehensive preference value β j Screening high-risk logistics projects and selecting β j Large-value logistics projects are considered high-risk targets; S42. For the selected high-risk logistics projects, check their corresponding waybill data to locate the key nodes of the risk source; S43. Take corresponding control measures for the nodes where risks originate.

Images