Potential customer mining method

Abstract

The invention relates to the technical field of customer mining, in particular to a potential customer mining method, which comprises the following steps: S1, constructing a supply chain network of a bank based on enterprise information, a transaction relationship between enterprises, natural person information and enterprise industrial and commercial information, the supply chain network having an entity table and associated edges; s2, constructing a calculation model of the supply strength between any two enterprises; s3, constructing a core enterprise identification model, and identifying potential core enterprises; s4, establishing a marketing clue model, and mining potential first-degree associated enterprises, second-degree associated enterprises and third-degree associated enterprises according to the marketing clue model based on the potential core enterprises; and S5, first-degree associated enterprises are preferentially selected and obtained, if no enterprise with the score reaching the standard exists in the first-degree associated enterprises, second-degree associated enterprises are obtained, and if no enterprise with the score reaching the standard exists in the second-degree associated enterprises, third-degree associated enterprises are obtained. According to the method, potential customers can be mined, and the reach rate and the success rate of associated marketing are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of customer mining, and in particular to a method for mining potential customers. Background Art

[0002] Customer management is the driving force behind the sustainable and high-quality development of banks. Traditional marketing lists and marketing information acquisition mainly rely on manual statistics by business personnel, resulting in low marketing efficiency. In terms of new customer development, there is a lack of effective marketing lists and related marketing auxiliary information, and the marketing success rate is low.

[0003] Therefore, it is necessary to provide a method for mining potential customers to improve the reach and success rate of related marketing. Summary of the invention

[0004] The purpose of the present invention is to provide a method for mining potential customers, so as to mine potential customers and improve the reach and success rate of associated marketing.

[0005] In order to solve the problems existing in the prior art, the present invention provides a method for mining potential customers, comprising the following steps: S1: Building a bank's supply chain network based on enterprise information, inter-enterprise transaction relationships, natural person information, and enterprise industrial and commercial information. The supply chain network has an entity table and associated edges. S2: Construct a calculation model for the supply intensity between any two enterprises. The calculation formula of the model is as follows: Q=X×(1+0.3Y)×(1+0.7Z), where Q is the value of the supply intensity between any two enterprises, X is the total transaction amount between enterprises in a preset time period, Y is the relationship stickiness value between enterprises, and Z is the transaction regularity gain value between enterprises; The calculation method of the relationship stickiness value between enterprises is as follows: ; Among them, P 1 P is the time period from the first transaction to the present, calculated in months; 2 The time period from the last transaction to the present, calculated in months; K=M / P 1 , M is the total number of months in which the transaction occurs; The calculation method of the transaction regularity gain value between enterprises is as follows: if M is less than 3, then Z=0; if M is greater than or equal to 3, then Z=exp(-B / A), where A and B are the mean and standard deviation of the monthly interval list respectively, and the monthly interval list is obtained according to the month in which the transaction occurs; S3: Build a core enterprise identification model and identify potential core enterprises; S4: Establish a marketing clue model, based on potential core enterprises, and mine potential related enterprises according to the marketing clue model. The marketing path score of the first-level related enterprises is calculated as follows: ; The calculation method of the marketing path score of the second-degree related enterprises is as follows: ; The calculation method of the marketing path score of the third-degree related enterprises is as follows: ;D 1 Score the marketing path of the first-degree related enterprise, D 2 Score the marketing path of the second-degree related enterprise, D 3 Score the marketing paths of third-degree related companies. is the weight coefficient of non-local customers, R 1 is the strength value of the relationship between the core enterprise and its first-degree related enterprises, R 2 is the strength value of the relationship between the core enterprise and the secondary related enterprises, R 3 is the strength value of the relationship between the core enterprise and the third-degree related enterprises, R 1 , R 2 and R 3 Based on supply strength, equity relationship, guarantee relationship, same senior management, same legal representative and same actual controller, The stickiness of the related enterprise relationships to be mined; S5: Give priority to first-degree related enterprises. If there is no enterprise with a score that meets the standard among the first-degree related enterprises, obtain second-degree related enterprises. If there is no enterprise with a score that meets the standard among the second-degree related enterprises, obtain third-degree related enterprises.

[0006] Optionally, in the potential customer mining method, abnormal data is marked on the supply chain network, and the abnormal data marking includes a suspected shell company mark, a super node processing mark, and a risk list processing mark.

[0007] Optionally, in the potential customer mining method, the core enterprise identification model is as follows: Score core = Industrial and commercial information characteristic parameters × [(paid-in capital characteristic parameters, establishment years characteristic parameters) × (paid-in capital, establishment years) T ]+Financial data characteristic parameters×[(asset-liability ratio characteristic parameters, enterprise profit margin characteristic parameters, pre-tax profit characteristic parameters, main business income characteristic parameters, credit rating characteristic parameters)×(asset-liability ratio, enterprise profit margin, pre-tax profit, main business income, credit rating) T ]+Business data characteristic parameter×[(Deposit amount characteristic parameter, Financing amount characteristic parameter)×(Deposit amount, Financing amount) T]+characteristic parameter of moving account amount×[(characteristic parameter of maximum expenditure amount, characteristic parameter of maximum income amount, characteristic parameter of average expenditure amount, characteristic parameter of average income amount)×(maximum expenditure amount, maximum income amount, average expenditure amount, average income amount) T ]+graph association data characteristic parameter×[(node ​​degree characteristic parameter, PageRank characteristic parameter, out-degree characteristic parameter, in-degree characteristic parameter)×(node ​​degree, PageRank, out-degree, in-degree) T ]+Negative information characteristic parameters×[(judicial case characteristic parameters, administrative penalty characteristic parameters, abnormal business status characteristic parameters, enterprise tax arrears amount characteristic parameters, whether the enterprise owes taxes characteristic parameters)×(judicial case, administrative penalty, abnormal business status, enterprise tax arrears amount, whether the enterprise owes taxes) T ], where Score core is the core enterprise identification score, T is the matrix transpose; Based on the core enterprise identification score, potential core enterprises are selected using the TOP n method, where n is a natural number.

[0008] Optionally, in the potential customer mining method, each characteristic parameter is set or changed according to actual conditions.

[0009] Optionally, in the potential customer mining method, the strength value of the association relationship between the core enterprise and the corresponding associated enterprise is calculated as follows: R = weight 1 × supply intensity + weight 2 × equity relationship + weight 3 × guarantee relationship + weight 4 × relationship with senior executives + weight 5 × relationship with legal representative + weight 6 × relationship with actual controller; Wherein, R is the calculated strength value of the relationship between the core enterprise and the corresponding associated enterprise, and R is calculated based on the calculation method of the strength value of the relationship between the core enterprise and the corresponding associated enterprise. 1 , R 2 and R 3 .

[0010] Optionally, in the potential customer mining method, the calculation method of the relationship stickiness of the associated enterprises to be mined is as follows: Score loyaty =Weight 7×customer account opening time + Weight 8×product holding quantity + Weight 9×whether it is a basic account + Weight 10×whether it is payment on behalf of others + Weight 11×whether it is a credit customer + Weight 12×whether the current loan balance is not 0 + Weight 13×deposit balance + Weight 14×loan balance + Weight 15×time since last transaction + Weight 16×transaction frequency + Weight 17×transaction amount.

[0011] In the potential customer mining method provided by the present invention, the bank's supply chain network is mainly constructed based on enterprise information, inter-enterprise transaction relationships, natural person information, and enterprise industrial and commercial information, and on this basis, a core enterprise identification model and a marketing clue model are constructed. First, the enterprise supply chain network is constructed based on the production supply chain relationship and enterprise association relationship, and the strength of the supply relationship between enterprises is quantified; then the core enterprise identification model is constructed, and potential core enterprises are mined, and a marketing list is output by sorting the core enterprise identification score values; finally, a marketing clue model is constructed by using a variety of associations between enterprises to screen out the marketing path with the highest marketing reach, thereby improving the reach and success rate of associated marketing. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 A flowchart of a potential customer mining method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0013] The specific implementation of the present invention will be described in more detail below in conjunction with the schematic diagram. The advantages and features of the present invention will become clearer based on the following description. It should be noted that the drawings are all in a very simplified form and are not in exact proportions, and are only used to facilitate and clearly assist in explaining the purpose of the embodiments of the present invention.

[0014] Hereinafter, if the method described herein includes a series of steps, the order in which the steps are presented herein is not necessarily the only order in which the steps may be performed, and some of the steps described may be omitted and / or some other steps not described herein may be added to the method.

[0015] Customer management is the driving force behind the sustainable and high-quality development of banks. Traditional marketing lists and marketing information acquisition mainly rely on manual statistics by business personnel, resulting in low marketing efficiency. In terms of new customer development, there is a lack of effective marketing lists and related marketing auxiliary information, and the marketing success rate is low.

[0016] In order to solve the problems existing in the prior art, the present invention provides a method for mining potential customers, such as Figure 1 As shown, the method comprises the following steps: S1: Building a bank's supply chain network based on enterprise information, inter-enterprise transaction relationships, natural person information, and enterprise industrial and commercial information. The supply chain network has an entity table and associated edges. Specifically, the structures of the entity table and the associated edge table are as follows.

[0017] Table 1 Entity information table

[0018] Table 2 Association edge information table

[0019] Preferably, the supply chain network is marked with abnormal data, including suspected shell enterprise marks, super node processing marks, and risk list processing marks. The abnormal data marking method is as follows: During the network construction process, there are entities that do not meet the marketing target requirements (such as suspected shell companies, blacklisted companies, etc.). These entities will have a certain negative impact on the core enterprise identification model and the marketing lead model. Therefore, these entities need to be marked in advance and eliminated in the subsequent modeling process.

[0020] (1) Suspected shell company mark Shell companies refer to companies that have established a corporate entity and a company name, but do not have any business operations or debts. Although such companies do not pose risks in most cases, their corporate nature means that they currently have no marketing value. However, when using relationships such as "common legal entities" to mine potential customers, such companies will affect the final effect of the model, so five rules are constructed to mark shell companies.

[0021] Table 3 Rules for suspected shell companies

[0022] (2) Super Node Processing Super nodes refer to nodes in the network whose out-degree or in-degree is much larger than the mean. Such nodes will destroy the stability and reliability of the model. In the enterprise relationship network, super nodes are mainly platform-type, agency-type, and trading-type enterprises. Specifically, there are many corporate counterparties containing keywords such as "trade, commerce, industry, and economic trade". Some trading companies even have more than 10,000 counterparties in the basic trading network. At the same time, there is a high probability that the counterparties of these trading companies also include similar super node companies, which leads to the alternation of relationship links between trading companies, which is not conducive to building production and manufacturing relationships in the supply chain. Therefore, the present invention sets three rules to limit such super nodes: For super node enterprises, the top 20 counterparties in terms of transaction amount are retained, and other counterparties with relatively low transaction amounts are eliminated.

[0023] According to business rules, the transaction links of enterprises without core enterprise certification scores within the three dimensions of TradeLink will be eliminated.

[0024] If the number of network connections is still too large after the above two methods, the Louvain algorithm can be used to split the graph structure.

[0025] (3) Risk list processing Enterprises on the risk list in the transaction network need to be marked, including the bank's blacklist, dishonest debtors, cancellation, revocation, liquidation of dishonest executors by the High Court, bankruptcy liquidation, gray and black lists involved in telecommunications fraud, anti-money laundering high-risk customer list, and overdue credit customers within the bank.

[0026] S2: Construct a calculation model for the supply intensity between any two enterprises. S21: The model calculation formula is as follows: Q=X×(1+0.3Y)×(1+0.7Z), where Q is the value of the supply intensity between any two enterprises, X is the total transaction amount between enterprises in a preset time period (for example, X is the total transaction amount between enterprises in the past 18 months), Y is the relationship stickiness value between enterprises, and Z is the transaction regularity gain value between enterprises; The calculation method of the relationship stickiness value between enterprises is as follows: ; Among them, P 1 P is the time period from the first transaction to the present, calculated in months; 2 The time period from the last transaction to the present, calculated in months; K=M / P 1 , M is the total number of months in which the transaction occurs. If the total transaction amount in a month is less than the preset amount, the month in which the transaction occurs is not included in M. In one embodiment, if the total transaction amount in a month is less than 100,000, the transaction in that month is not included in the number of months in which the transaction occurs. The calculation method of the transaction regularity gain value between enterprises is as follows: if M is less than 3, then Z=0; if M is greater than or equal to 3, then Z=exp(-B / A), where A and B are the average and standard deviation of the monthly interval list respectively, and the monthly interval list is obtained according to the month in which the transaction occurs; in one embodiment, the effective month list is [month 1 ,month 2 ,...month n-1 ,month n ], and then get the monthly interval list [month 2 -month 1 ,...,month n -month n-1 ].

[0027] S22: Normalize the value of inter-enterprise supply intensity to obtain the normalized inter-enterprise supply intensity value, denoted by S.

[0028] S3: Build a core enterprise identification model and identify potential core enterprises; Specifically, the core enterprise identification model is as follows: S31: Score core= Industrial and commercial information characteristic parameters × [(paid-in capital characteristic parameters, establishment years characteristic parameters) × (paid-in capital, establishment years) T ]+Financial data characteristic parameters×[(asset-liability ratio characteristic parameters, enterprise profit margin characteristic parameters, pre-tax profit characteristic parameters, main business income characteristic parameters, credit rating characteristic parameters)×(asset-liability ratio, enterprise profit margin, pre-tax profit, main business income, credit rating) T ]+Business data characteristic parameter×[(Deposit amount characteristic parameter, Financing amount characteristic parameter)×(Deposit amount, Financing amount) T ]+characteristic parameter of moving account amount×[(characteristic parameter of maximum expenditure amount, characteristic parameter of maximum income amount, characteristic parameter of average expenditure amount, characteristic parameter of average income amount)×(maximum expenditure amount, maximum income amount, average expenditure amount, average income amount) T ]+graph association data characteristic parameter×[(node ​​degree characteristic parameter, PageRank characteristic parameter, out-degree characteristic parameter, in-degree characteristic parameter)×(node ​​degree, PageRank, out-degree, in-degree) T ]+Negative information characteristic parameters×[(judicial case characteristic parameters, administrative penalty characteristic parameters, abnormal business status characteristic parameters, enterprise tax arrears amount characteristic parameters, whether the enterprise owes taxes characteristic parameters)×(judicial case, administrative penalty, abnormal business status, enterprise tax arrears amount, whether the enterprise owes taxes) T ], where Score core To identify the score value for core enterprises, PageRankPageRank algorithm is the core algorithm used by Google to measure the importance of web pages. It evaluates the level and importance of web pages by analyzing the number and quality of inbound links to the web pages. T is the matrix transpose.

[0029] Preferably, each characteristic parameter is set or changed according to the actual situation. In one embodiment, each characteristic parameter is shown in the following table: Table 4 Core enterprise identification model

[0030] In this embodiment, Score core =0.1×[(0.4,0.6)×(paid-in capital, years of establishment) T ]+0.3×[(0.2,0.15,0.2,0.2,0.25)×(asset-liability ratio, enterprise profit margin, pre-tax profit, main business income, credit rating) T ]+0.15×[(0.65,0.35)×(deposit amount, financing amount) T]+0.15×[(0.2,0.2,0.3,0.3)×(expenditure-maximum value, income-maximum value, expenditure-average value, income-average value) T ]+0.1×[(0.3,0.3,0.2,0.2)×(node ​​degree, PageRank, out-degree, in-degree) T ]+0.2×[(0.2,0.2,0.25,0.15,0.2)×(judicial cases, administrative penalties, abnormal business status, amount of tax owed by the enterprise, whether the enterprise owes tax) T ].

[0031] S32: Based on the core enterprise identification score, potential core enterprises are selected using the TOP n method, where n is a natural number.

[0032] S4: Establish a marketing clue model, based on potential core enterprises, and mine potential related enterprises according to the marketing clue model. The marketing path score of the first-level related enterprises is calculated as follows: ; The calculation method of the marketing path score of the second-degree related enterprises is as follows: ; The calculation method of the marketing path score of the third-degree related enterprises is as follows: ;D 1 Score the marketing path of the first-degree related enterprise, D 2 Score the marketing path of the second-degree related enterprise, D 3 Score the marketing paths of third-degree related companies. is the weight coefficient of non-local customers, R 1 is the strength value of the relationship between the core enterprise and its first-degree related enterprises, R 2 is the strength value of the relationship between the core enterprise and the secondary related enterprises, R 3 is the strength value of the relationship between the core enterprise and the third-degree related enterprises, R 1 , R 2 and R 3 Based on supply strength, equity relationship, guarantee relationship, same senior management, same legal representative and same actual controller, The stickiness of the related enterprise relationships to be mined; S41: The calculation method for the strength value of the relationship between the core enterprise and the corresponding associated enterprise is as follows: R = weight 1 × supply intensity + weight 2 × equity relationship + weight 3 × guarantee relationship + weight 4 × relationship with senior executives + weight 5 × relationship with legal representative + weight 6 × relationship with actual controller; Wherein, R is the calculated strength value of the relationship between the core enterprise and the corresponding associated enterprise, and R is calculated based on the calculation method of the strength value of the relationship between the core enterprise and the corresponding associated enterprise. 1 , R2 and R 3 . . If the core enterprise and the corresponding affiliated enterprise have the same supply intensity, equity relationship, guarantee relationship, same senior executives, same legal representative and same actual controller, the calculation is based on the calculation method of the strength value of the relationship. If not, the corresponding parameter is 0. For example, if the core enterprise and the second-degree affiliated enterprise do not have the same legal representative, the parameter of the same legal representative is 0; for example, if the core enterprise and the third-degree affiliated enterprise do not have the same supply intensity, equity relationship, guarantee relationship, same senior executives, same legal representative and same actual controller, R is not calculated. 3 .

[0033] In one embodiment, the calculation method of supply intensity, equity relationship, guarantee relationship, same senior executives, same legal representative and same actual controller is as follows: Table 5 Relationship Name Rating Definition The final weights in the model Supply Strength <![CDATA[Rsupply: 1. If it is a one-way transaction: (1) If the referral enterprise is the buyer and the main settlement method with the target enterprise is buyer-dominated; or it is the seller and the main settlement method with the target enterprise is seller-dominated, then Rsupply = S (refer to step S22). Here, the referral enterprise is an internal enterprise of the bank. (2) If the referral enterprise is the seller and the main settlement method with the target enterprise is buyer-dominated; or it is the buyer and the main settlement method with the target enterprise is seller-dominated, then Rsupply = S × 0.7. 2. If it is a two-way mutual transaction: Rsupply = 1 - (1 - S 1 )(1 - S 2 ), where S 1 is the normalized inter-enterprise supply intensity value when flowing out of the potential core enterprise, and S 2 is the normalized inter-enterprise supply intensity value when flowing into the potential core enterprise.]]> 21% Equity Relations Requity: 1. If there is one-way holding: (1) the referring enterprise is the investor, then Requity = 1; (2) the referring enterprise is the investee, then Requity = 0.8; 2. If there are cross-holdings: then Requity = 1. 24% Guarantee relationship Rguarantee: 1. If it is a one-way guarantee: if the referring enterprise is the guarantor, then Rguarantee=1; if the referring enterprise is the guaranteed party, then Rguarantee=0.6; 2. If it is a mutual guarantee, then Rguarantee=1. 15% Same executives Rsenior: If yes, then Rsenior=1; if no, then Rsenior=0. 10% Same as legal representative Rperson: Yes, then Rperson=1 No, then Rperson=0 14% Same actual controller Rcontrol:Yes, Rcontrol=1No, Rcontrol=0 16% In this embodiment, R=21%×Rsupply+24%×Requity+15%×Rguarantee+10%×Rsenior+14%×Rperson+16%×Rcontrol.

[0034] S42: The calculation method of the stickiness of the related enterprise relationship to be mined is as follows: Score loyaty =Weight 7×customer account opening time + Weight 8×product holding quantity + Weight 9×whether it is a basic account + Weight 10×whether it is payment on behalf of others + Weight 11×whether it is a credit customer + Weight 12×whether the current loan balance is not 0 + Weight 13×deposit balance + Weight 14×loan balance + Weight 15×time since last transaction + Weight 16×transaction frequency + Weight 17×transaction amount.

[0035] In one embodiment, the calculation method of the customer's account opening time, the number of products held, whether it is a basic account, whether it is an agency payment, whether it is a credit customer, whether the current loan balance is not 0, the deposit balance, the loan balance, the time since the last transaction, the transaction frequency, and the transaction amount is as follows: Table 6

[0036]

[0037] In this embodiment, Score loyaty =10%×Stime+18%×Sproduct+12%×Sc1+6%×Sc2+6%×Sc3+6%×Sc4+6%×Sdeposit+6%×Sloan+9%×Sdiff+12%×Scount+9%×Smoney.

[0038] S43: In one embodiment, the weight coefficient of the local and remote customers The way to obtain is as follows: Table 7

[0039] S5: Give priority to obtaining first-degree related enterprises. If there is no enterprise with a score that meets the standard among the first-degree related enterprises, obtain second-degree related enterprises. If there is no enterprise with a score that meets the standard among the second-degree related enterprises, obtain third-degree related enterprises. If there is no enterprise with a score that meets the standard among the third-degree related enterprises, do not mine related enterprises.

[0040] In summary, in the potential customer mining method provided by the present invention, the bank's supply chain network is mainly constructed based on enterprise information, inter-enterprise transaction relationships, natural person information, and enterprise industrial and commercial information, and on this basis, a core enterprise identification model and a marketing clue model are constructed. First, the enterprise supply chain network is constructed based on the production supply chain relationship and enterprise association relationship, and the strength of the supply relationship between enterprises is quantified; then the core enterprise identification model is constructed, and potential core enterprises are mined, and a marketing list is output by sorting the core enterprise identification score values; finally, a marketing clue model is constructed by using a variety of associations between enterprises to screen out the marketing path with the highest marketing reach, thereby improving the reach and success rate of associated marketing.

[0041] The above is only a preferred embodiment of the present invention and does not limit the present invention in any way. Any technician in the relevant technical field, without departing from the scope of the technical solution of the present invention, makes any form of equivalent replacement or modification to the technical solution and technical content disclosed in the present invention, which does not depart from the content of the technical solution of the present invention and still falls within the protection scope of the present invention.

Claims

1. A method for mining potential customers, characterized in that: The following steps are involved: S1: Building a bank's supply chain network based on enterprise information, inter-enterprise transaction relationships, natural person information, and enterprise industrial and commercial information. The supply chain network has an entity table and associated edges. S2: Construct a calculation model for the supply intensity between any two enterprises. The calculation formula of the model is as follows: Q=X×(1+0.3Y)×(1+0.7Z), where Q is the value of the supply intensity between any two enterprises, X is the total transaction amount between enterprises in a preset time period, Y is the relationship stickiness value between enterprises, and Z is the transaction regularity gain value between enterprises; The calculation method of the relationship stickiness value between enterprises is as follows: ； Among them, P1 is the time period from the first transaction to the present, calculated in months; P2 is the time period from the last transaction to the present, calculated in months; K=M / P1, M is the total number of months in which transactions occur; The calculation method of the transaction regularity gain value between enterprises is as follows: if M is less than 3, then Z=0; if M is greater than or equal to 3, then Z=exp(-B / A), where A and B are the mean and standard deviation of the monthly interval list respectively, and the monthly interval list is obtained according to the month in which the transaction occurs; S3: Build a core enterprise identification model and identify potential core enterprises; S4: Establish a marketing clue model. Based on the potential core enterprises, explore potential related enterprises according to the marketing clue model. The marketing path score of the first-level related enterprises is calculated as follows: ; The calculation method of the marketing path score of the second-degree related enterprises is as follows: ; The calculation method of the marketing path score of the third-degree related enterprises is as follows: ; D1 is the marketing path score of the first-degree related enterprise, D2 is the marketing path score of the second-degree related enterprise, and D3 is the marketing path score of the third-degree related enterprise. is the weight coefficient of the non-local customers, R1 is the strength value of the relationship between the core enterprise and the first-degree related enterprises, R2 is the strength value of the relationship between the core enterprise and the second-degree related enterprises, and R3 is the strength value of the relationship between the core enterprise and the third-degree related enterprises. R1, R2 and R3 are obtained based on the supply strength, equity relationship, guarantee relationship, the same senior executives, the same legal representative and the same actual controller. is the stickiness of the related enterprise relationship to be mined, and the calculation method of the stickiness of the related enterprise relationship to be mined is as follows: Score loyaty =Weight 7×customer account opening time+Weight 8×product holding quantity+Weight 9×whether it is a basic account+Weight 10×whether it is payment on behalf of others+Weight 11×whether it is a credit customer+Weight 12×whether the current loan balance is not 0+Weight 13×deposit balance+Weight 14×loan balance+Weight 15×time since last transaction+Weight 16×transaction frequency+Weight 17×transaction amount; S5: Give priority to obtaining first-degree related enterprises. If there are no enterprises with qualified scores among the first-degree related enterprises, obtain second-degree related enterprises. If there are no enterprises with qualified scores among the second-degree related enterprises, obtain third-degree related enterprises.

2. The method for mining potential customers as claimed in claim 1, characterized in that: Abnormal data marking is performed on the supply chain network, including suspected shell company marking, super node processing marking and risk list processing marking.

3. The method for mining potential customers as claimed in claim 1, characterized in that: The core enterprise identification model is as follows: Score core = Industrial and commercial information characteristic parameters × [(paid-in capital characteristic parameters, establishment years characteristic parameters) × (paid-in capital, establishment years) T ]+Financial data characteristic parameters×[(asset-liability ratio characteristic parameters, enterprise profit margin characteristic parameters, pre-tax profit characteristic parameters, main business income characteristic parameters, credit rating characteristic parameters)×(asset-liability ratio, enterprise profit margin, pre-tax profit, main business income, credit rating) T ]+Business data characteristic parameter×[(Deposit amount characteristic parameter, Financing amount characteristic parameter)×(Deposit amount, Financing amount) T ]+characteristic parameter of moving account amount×[(characteristic parameter of maximum expenditure amount, characteristic parameter of maximum income amount, characteristic parameter of average expenditure amount, characteristic parameter of average income amount)×(maximum expenditure amount, maximum income amount, average expenditure amount, average income amount) T ]+graph association data characteristic parameter×[(node ​​degree characteristic parameter, PageRank characteristic parameter, out-degree characteristic parameter, in-degree characteristic parameter)×(node ​​degree, PageRank, out-degree, in-degree) T ]+Negative information characteristic parameters×[(judicial case characteristic parameters, administrative penalty characteristic parameters, abnormal business status characteristic parameters, enterprise tax arrears amount characteristic parameters, whether the enterprise owes taxes characteristic parameters)×(judicial case, administrative penalty, abnormal business status, enterprise tax arrears amount, whether the enterprise owes taxes) T ], where Score core is the core enterprise identification score, T is the matrix transpose; Based on the core enterprise identification score, potential core enterprises are selected using the TOP n method, where n is a natural number.

4. The method for mining potential customers as claimed in claim 1, characterized in that: The calculation method of the strength value of the relationship between the core enterprise and the corresponding affiliated enterprises is as follows: R = weight 1 × supply intensity + weight 2 × equity relationship + weight 3 × guarantee relationship + weight 4 × relationship with senior executives + weight 5 × relationship with legal representative + weight 6 × relationship with actual controller; Wherein, R is the calculated strength value of the relationship between the core enterprise and the corresponding associated enterprise, and R1, R2 and R3 are calculated based on the calculation method of the strength value of the relationship between the core enterprise and the corresponding associated enterprise.

Images