CN113792219A - Customer recommendation model construction method under financing lease scene - Google Patents
Customer recommendation model construction method under financing lease scene Download PDFInfo
- Publication number
- CN113792219A CN113792219A CN202111351119.XA CN202111351119A CN113792219A CN 113792219 A CN113792219 A CN 113792219A CN 202111351119 A CN202111351119 A CN 202111351119A CN 113792219 A CN113792219 A CN 113792219A
- Authority
- CN
- China
- Prior art keywords
- index
- indexes
- influence factors
- consistency
- matrix
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/06—Asset management; Financial planning or analysis
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Finance (AREA)
- Theoretical Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- General Engineering & Computer Science (AREA)
- Game Theory and Decision Science (AREA)
- Human Resources & Organizations (AREA)
- Operations Research (AREA)
- Data Mining & Analysis (AREA)
- Technology Law (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The invention discloses a method for constructing a customer recommendation model in a financing lease scene. The method comprises the steps of obtaining a registration report of the financing lease service from the middle-network-logging, and abstracting a plurality of indexes from registration basic information, lessee information and lease property newly-added in the registration report; constructing a hierarchical structure of the influence factors according to the indexes, wherein the indexes serve as primary influence factors, and the options of each index serve as secondary influence factors; constructing a reciprocal matrix according to the influence factors to quantify the importance degree among different influence factors; carrying out consistency check on the constructed reciprocal matrix; after the consistency check is passed, calculating the weight of the index; and giving option scores to the options of the indexes, setting the product of the weight of the indexes and the option scores as index scores, and taking the sum of the index scores as a client recommendation score. The invention can reduce the range of the customer group, and the recommendation is more targeted, and the customer conversion rate of the recommendation model is better, and has better interpretability, feasibility and iteration capability.
Description
Technical Field
The invention relates to the technical field of customer recommendation in a financing lease scene, in particular to a customer recommendation model construction method in the financing lease scene.
Background
One of the cores of the development of the financing lease service is the customer. How to acquire customers and how to perform accurate marketing is always one of the directions explored in the industry. In the current market, a marketing scheme oriented to the financial field mainly aims at bank customers, mainly aims at large and medium-sized enterprises such as listed companies and the like, and a recommendation model carries out accurate marketing mainly according to the information of the scale, the industry, the bid and the like of the enterprises.
The characteristics of the financing lease service oriented to the guest group are different from those of the common bank customers, and most of the financing lease service is small and micro enterprises. The small and micro enterprises are characterized by small scale, low network exposure rate, few capital channels and the like. The conventional recommendation model cannot be applied to the financing lease service: firstly, a proper small micro-enterprise cannot be recommended; and secondly, the acceptance degree of the clients to the financing lease service is not considered when the clients are recommended, the marketing cost of a client manager is improved, and the marketing success rate of the client manager is influenced.
Disclosure of Invention
The invention aims to provide a customer recommendation model construction method under a financing lease scene aiming at the defects in the prior art.
In order to achieve the above object, the present invention provides a method for constructing a customer recommendation model in a financing lease scenario, comprising:
acquiring a registration report of the financing lease service from the middle-logging network, and abstracting a plurality of indexes from registration basic information, tenant information and lease property newly-added in the registration report;
constructing a hierarchical structure of influence factors according to the indexes, wherein the indexes serve as primary influence factors, and options of each index serve as secondary influence factors;
constructing a reciprocal matrix according to the influence factors to quantify the importance degree among different influence factors;
carrying out consistency check on the constructed reciprocal matrix;
after the consistency check is passed, calculating the weight of the index;
and giving option scores to the options of the indexes, setting the product of the weight of the indexes and the option scores as index scores, and taking the sum of the index scores as a client recommendation score.
Further, the indexes comprise rental expiration conditions, customer activity conditions, lessor classification conditions, rental price conditions and average rental term conditions.
Further, the consistency check of the constructed reciprocal matrix specifically includes:
calculating maximum eigenvalue of reciprocal matrix
And according to the maximum eigenvalue of the inverse matrix
And calculating a consistency index CI according to the dimension n of the inverse matrix:
wherein n is a natural number greater than 1;
searching and matching a set average random consistency index RI according to the dimension n of the inverse matrix;
and calculating a consistency ratio CR according to the consistency index CI and the average random consistency index RI:
if CR is smaller than the set threshold value, the consistency detection of the reciprocal matrix is judged to pass, otherwise, the consistency detection is judged not to pass.
Further, the set threshold is 0.1.
Further, the weight of the index is calculated by any one of an arithmetic average method, a geometric average method, and a characteristic value method.
Further, the calculating the weight of the index by using an arithmetic mean method specifically includes:
normalizing the reciprocal matrix according to columns;
summing the normalized matrix according to rows;
and dividing each element in the vector obtained after summation by the dimension n of the inverse matrix to obtain the weight of each index.
Has the advantages that: according to the invention, the user recommendation model is constructed based on the analytic hierarchy process by introducing the in-process network-surfing data construction index, and compared with the conventional user recommendation model, the client recommendation model has the following advantages:
1. the client group range is narrowed, and the recommendation is more targeted;
2. the model index factors and the service scenes are highly managed, and the customer conversion rate of the recommended model is better;
3. expert experience in the field is introduced in the model construction, and the model is more interpretable;
4. the model is built by fully utilizing the public data source, and the model has higher feasibility and iteration capability.
Drawings
FIG. 1 is a schematic diagram of a hierarchy of influencing factors constructed from metrics;
FIG. 2 is a schematic diagram of a reciprocal matrix constructed from influencing factors;
FIG. 3 is a schematic diagram of summing the normalized matrix by rows;
FIG. 4 is a schematic diagram of the weights calculated for each index;
FIG. 5 is a diagram of a scoring results table for a customer recommendation model.
Detailed Description
The present invention will be further illustrated with reference to the accompanying drawings and specific examples, which are carried out on the premise of the technical solution of the present invention, and it should be understood that these examples are only for illustrating the present invention and are not intended to limit the scope of the present invention.
The embodiment of the invention provides a method for constructing a customer recommendation model in a financing lease scene, which comprises the following steps:
and acquiring a registration report of the financing lease service from the middle-logging network, and abstracting a plurality of indexes from registration basic information, tenant information and lease property newly-added in the registration report. The indicators may include rental expiration, customer liveness, lessor classification, rental price, and average rental period.
And constructing a hierarchical structure of the influence factors according to the indexes, wherein the indexes serve as primary influence factors, and the options of each index serve as secondary influence factors. Specifically, referring to fig. 1, fig. 1 illustrates a constructed influence hierarchy, where a rental expiration condition, a customer activity condition, a lessor classification condition, a rental price condition, and an average rental period condition are used as primary influence factors, and options under each index are used as secondary influence factors. Options such as rental expiration include the presence of a financing lease registration that is about to expire in approximately 1 month, the presence of a financing lease registration that is about to expire in approximately 1-3 months, the presence of a financing lease registration that is about to expire in approximately 3-6 months, the presence of a financing lease registration that is about to expire in approximately 6-12 months, and others, totaling 5 options, with 5 options all being secondary influencing factors.
And constructing a reciprocal matrix according to the influence factors so as to quantify the importance degree among different influence factors. The 1-9 scale proposed by professor t.l.saaty is generally used, as shown in table 1:
TABLE 1
As can be seen from table 1, when a: b =3, it can be said that a is slightly more important than B, whereas B: when a =1/7, a is more important than B. In order to increase the contrast, scales 3, 5, 7 and 9 are usually selected for comparison, the number of indexes suitable for comparison is less than or equal to 5, and once the number of indexes is more than 5, scales 2, 4, 6 and 8 are added. The invention provides a reciprocal matrix of a customer recommendation model shown in table 2 according to the above theory and by combining with business experience:
TABLE 2
It should be noted that the correlation scale of the reciprocal matrix in table 2 can be actually fine-tuned according to the expert experience of the customer manager.
And carrying out consistency check on the constructed reciprocal matrix. Specifically, the consistency check of the constructed reciprocal matrix specifically includes:
calculating maximum eigenvalue of reciprocal matrix
And according to the maximum eigenvalue of the inverse matrix
And calculating a consistency index CI according to the dimension n of the inverse matrix:
wherein n is a natural number greater than 1 and is equal to the index number;
and searching and matching the set average random consistency index RI according to the dimension n of the inverse matrix.
And calculating a consistency ratio CR according to the consistency index CI and the average random consistency index RI:
if CR is smaller than the set threshold value, the consistency detection of the reciprocal matrix is judged to pass, otherwise, the consistency detection is judged not to pass. The set threshold is preferably 0.1, and if CR <0.1, it is considered that the consistency of the reciprocal matrix is acceptable, and it is determined that the consistency detection is passed; otherwise, the consistency detection is judged not to pass, and the reciprocal matrix needs to be corrected. By consistency detection, the reciprocal matrix can be used.
The average random consistency index RI and the dimension n can be set according to the values in table 3:
TABLE 3
It should be noted that in practical applications, the number of dimensions (the number of indexes) n rarely exceeds 10, and if the number of dimensions is greater than 10, it is considered that a secondary index body is established.
According to the above process and parameters, the maximum eigenvalue of the reciprocal matrix in the customer recommendation model
5.237194118217367, the consistency index CI has a value of 0.05929852955434178 and the consistency ratio CR has a value of 0.05294511567351944, thereby determining that the bit consistency check passed.
After the consistency check passes, the weight of the index is calculated. There are three methods for calculating the weight, which are an arithmetic average method, a geometric average method and a characteristic value method, and any one of them can be used for calculation. Taking an arithmetic mean method as an example for explanation, the method specifically comprises the following steps:
and normalizing the reciprocal matrix by columns. After treatment as shown in figure 2.
And summing the normalized matrix according to rows. After summing as shown in fig. 3.
And dividing each element in the vector obtained after summation by the dimension n of the inverse matrix to obtain the weight of each index. The weights of rental expiration, customer liveness, rental category, rental price, and average rental period are shown in fig. 4. For the calculated 5 weights, the weights may be sorted from large to small, as shown in table 4:
TABLE 4
And (3) respectively giving option scores to the options of the indexes according to expert experience, setting the product of the weight of the indexes and the option scores as the index score, and taking the sum of the index scores as the client recommendation score, so as to form a scoring result table of the client recommendation model shown in fig. 5. Referring to fig. 5, corresponding index scores are respectively calculated according to the rental expiration condition, the customer activity condition, the lessor classification condition, the rental price condition and the average rental term condition, and then added to obtain the customer recommendation score. It should be noted that, in the four indexes of the customer liveness condition, the lessor classification condition, the rental price condition and the average rental term condition, only one option can be matched with the customer, so that finally, only one index score can be obtained by each of the four indexes. However, the rental expiration condition may be various, for example, a certain customer has a financing rental registration which will expire in about 1 month and a financing rental registration which will expire in about 1-3 months, so when calculating the index score of the rental expiration condition, it is preferable to take the maximum value of the index score. In actual operation, this is achieved by program logic. Specifically, in the judgment, the judgment is performed according to the option scores from high to low, if whether the financing lease registration about to expire in the last 1 month exists is judged, if yes, the index score is calculated according to the option score of the financing lease registration about to expire in the last 1 month, and then the judgment is skipped; if not, it is determined whether there is a financing lease registration that is about to expire in about 1 to 3 months, if so, the index score is calculated by the option score of "the financing lease registration that is about to expire in about 1 to 3 months", and if not, the process is sequentially performed. Or multi-thread simultaneous calculation is carried out, and finally the maximum value is taken.
The foregoing is only a preferred embodiment of the present invention, and it should be noted that other parts not specifically described are within the prior art or common general knowledge to those of ordinary skill in the art. Without departing from the principle of the invention, several improvements and modifications can be made, and these improvements and modifications should also be construed as the scope of the invention.
Claims (6)
1. A customer recommendation model construction method under a financing lease scene is characterized by comprising the following steps:
acquiring a registration report of the financing lease service from the middle-logging network, and abstracting a plurality of indexes from registration basic information, tenant information and lease property newly-added in the registration report;
constructing a hierarchical structure of influence factors according to the indexes, wherein the indexes serve as primary influence factors, and options of each index serve as secondary influence factors;
constructing a reciprocal matrix according to the influence factors to quantify the importance degree among different influence factors;
carrying out consistency check on the constructed reciprocal matrix;
after the consistency check is passed, calculating the weight of the index;
and giving option scores to the options of the indexes, setting the product of the weight of the indexes and the option scores as index scores, and taking the sum of the index scores as a client recommendation score.
2. The method for building the customer recommendation model in the financing rental scenario of claim 1, wherein the indicators include rental expiration status, customer activity status, rental classification status, rental price status, and average rental term status.
3. The method for building a customer recommendation model in a financing lease scenario according to claim 1, wherein the consistency check of the constructed reciprocal matrix specifically comprises:
calculating maximum eigenvalue of reciprocal matrix
And according to the maximum eigenvalue of the inverse matrix
And calculating a consistency index CI according to the dimension n of the inverse matrix:
wherein n is a natural number greater than 1;
searching and matching a set average random consistency index RI according to the dimension n of the inverse matrix;
and calculating a consistency ratio CR according to the consistency index CI and the average random consistency index RI:
if CR is smaller than the set threshold value, the consistency detection of the reciprocal matrix is judged to pass, otherwise, the consistency detection is judged not to pass.
4. The method of claim 3, wherein the threshold is 0.1.
5. The method for building a customer recommendation model in a financing lease scenario according to claim 1, wherein the index is weighted by any one of arithmetic mean, geometric mean and eigenvalue.
6. The method of claim 5, wherein the calculating the weight of the index by using an arithmetic mean method specifically comprises:
normalizing the reciprocal matrix according to columns;
summing the normalized matrix according to rows;
and dividing each element in the vector obtained after summation by the dimension n of the inverse matrix to obtain the weight of each index.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111351119.XA CN113792219A (en) | 2021-11-16 | 2021-11-16 | Customer recommendation model construction method under financing lease scene |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111351119.XA CN113792219A (en) | 2021-11-16 | 2021-11-16 | Customer recommendation model construction method under financing lease scene |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113792219A true CN113792219A (en) | 2021-12-14 |
Family
ID=78955362
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111351119.XA Pending CN113792219A (en) | 2021-11-16 | 2021-11-16 | Customer recommendation model construction method under financing lease scene |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113792219A (en) |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111062581A (en) * | 2019-11-26 | 2020-04-24 | 国网浙江省电力有限公司电力科学研究院 | Enterprise client high-voltage value-added service system construction method based on AHP (advanced high Performance packet protocol) hierarchy |
| CN112446776A (en) * | 2019-08-27 | 2021-03-05 | 北京宸信征信有限公司 | Small and medium-sized enterprise credit evaluation system and method based on multi-source docking fusion data |
-
2021
- 2021-11-16 CN CN202111351119.XA patent/CN113792219A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112446776A (en) * | 2019-08-27 | 2021-03-05 | 北京宸信征信有限公司 | Small and medium-sized enterprise credit evaluation system and method based on multi-source docking fusion data |
| CN111062581A (en) * | 2019-11-26 | 2020-04-24 | 国网浙江省电力有限公司电力科学研究院 | Enterprise client high-voltage value-added service system construction method based on AHP (advanced high Performance packet protocol) hierarchy |
Non-Patent Citations (2)
| Title |
|---|
| 刘辉等: "《基于机器学习的精准扶贫问题分析与对策》", 30 August 2020, 科学技术文献出版社 * |
| 李双元: "《模式优化与促进政策研究》", 31 December 2015, 中国经济出版社 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108960833B (en) | Abnormal transaction identification method, equipment and storage medium based on heterogeneous financial characteristics | |
| CN108665159A (en) | A kind of methods of risk assessment, device, terminal device and storage medium | |
| Zinser | Retail Islamic banking and financial services: Determinants of use by Muslims in the USA | |
| CN108304656B (en) | Task acceptance condition simulation method for labor crowdsourcing platform | |
| CN112734161A (en) | Method, equipment and storage medium for accurately identifying empty-shell enterprises | |
| CN110119980A (en) | A kind of anti-fraud method, apparatus, system and recording medium for credit | |
| Susilo | Unlocking the secret of E-loyalty: a study from Tiktok users in China | |
| CN113821827B (en) | Combined modeling method and device for protecting multiparty data privacy | |
| CN112580978B (en) | Power market member credit evaluation and credit label generation method | |
| Isaac et al. | Individual behavior and bidding heterogeneity in sealed bid auctions where the number of bidders is unknown | |
| Arief et al. | Fraud detection based-on data mining on Indonesian E-Procurement System (SPSE) | |
| CN110264364B (en) | Recommendation method for investor | |
| CN116523602A (en) | Financial product potential user recommendation method for multi-party semi-supervised learning | |
| CN113792219A (en) | Customer recommendation model construction method under financing lease scene | |
| CN116468273A (en) | Customer risk identification method and device | |
| CN113177733B (en) | Middle and small micro enterprise data modeling method and system based on convolutional neural network | |
| Mosavi et al. | Customer value analysis in banks using data mining and fuzzy analytic hierarchy processes | |
| CN113792044A (en) | Data fusion platform and neural network model hosting training method | |
| Thaba et al. | Analysis of factors affecting proper functioning of smallholder agricultural cooperatives in the Limpopo Province of South Africa | |
| CN110570301A (en) | Risk identification method, device, equipment and medium | |
| CN111681090A (en) | Account grouping method and device of business system, terminal equipment and storage medium | |
| CN107240019A (en) | Assess customer service preference methods, customer investment risk partiality method and device | |
| CN112035594A (en) | Bidding information extraction result screening system and method | |
| Salgado et al. | Factors influencing the acceptance of financial technology by consumers of traditional financial services in Portugal | |
| CN113052689B (en) | Product recommendation method and device based on decision tree |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20211214 |