CN114386490A - Financial customer grading method based on RFM model generalization characteristics - Google Patents
Financial customer grading method based on RFM model generalization characteristics Download PDFInfo
- Publication number
- CN114386490A CN114386490A CN202111588546.XA CN202111588546A CN114386490A CN 114386490 A CN114386490 A CN 114386490A CN 202111588546 A CN202111588546 A CN 202111588546A CN 114386490 A CN114386490 A CN 114386490A
- Authority
- CN
- China
- Prior art keywords
- data
- model
- rfm
- customer
- financial
- 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
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- 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
Abstract
The invention discloses a financial customer grading method based on RFM model generalization characteristics in the technical field of data classification, which comprises the following steps: s1, collecting client data, and randomly extracting sample data with a preset proportion for cross labeling; wherein the customer data comprises RFM metric data and additional feature data for a particular scenario; s2, carrying out feature processing on the marked data, and carrying out machine learning model training; s3, analyzing and processing the total customer data in a machine learning model in a specific financial place, grading customers by combining the RFM model with the added specific scene characteristics in different financial scenes and combining a classification algorithm in a mode of randomly sampling and marking data, realizing grading of customers in different scenes, promoting the accuracy of scenarized marketing, and being particularly suitable for grading the accuracy of the customers in different financial scenes.
Description
Technical Field
The invention relates to the technical field of data classification, in particular to a financial customer grading method based on RFM model generalization characteristics.
Background
The RFM model is an important tool and means to measure customer value and customer profitability. Among the numerous modes of analysis for Customer Relationship Management (CRM), the RFM model is widely mentioned. The model describes the value status of a customer by its recent purchases, the overall frequency of purchases, and how much 3 dollars are spent. The last consumption (Recency) refers to how long the last purchase has elapsed since the specified time. Frequency of consumption (Frequency) refers to the number of times a customer purchases over a defined period of time. The amount of consumption (money) refers to the difference between the amount of purchase minus the amount of return within a defined period of time, i.e., the net income that the customer brings. RFM measures the value of the client from three dimensions of the recent consumption time, consumption frequency and consumption amount, and then grades the value.
The traditional RFM model performs customer ranking: r represents the last consumption time (Recency), and the interval from the last consumption record to the current time can be taken, such as 7 days, 30 days and 90 days of non-consumption; f denotes the Frequency of consumption (Frequency) over a certain time, typically the Frequency of consumption by the user over a period of time; m represents the accumulated consumption amount (money) in a certain time, and generally, the user consumption amount in a certain time period is taken; essentially, a standard method is found by using three classification dimensions; through the combined calculation of three dimensions, the grading of the client can be judged, and then corresponding measures are taken.
However, in the process of implementing the technical solution of the invention in the embodiments of the present application, the inventors of the present application find that the above-mentioned technology has at least the following technical problems:
1. the traditional RFM model excessively depends on manual work to carry out customer grading;
2. the traditional RFM model has less data characteristics considered in modeling and cannot reflect the grading situation of actual financial customers on specific business.
Based on the above, the invention designs a financial customer grading method based on the RFM model generalization characteristics to solve the above problems.
Disclosure of Invention
In order to solve the technical problems mentioned in the background art, the invention aims to provide a financial customer grading method based on the generalization characteristics of an RFM model.
In order to achieve the purpose, the invention adopts the following technical scheme:
a financial customer grading method based on RFM model generalization characteristics comprises the following steps:
s1, collecting client data, and randomly extracting sample data with a preset proportion for cross labeling; wherein the customer data comprises RFM metric data and additional feature data for a particular scenario;
s2, carrying out feature processing on the marked data, and carrying out machine learning model training;
and S3, analyzing and processing the total amount of customer data in a specific financial place in the machine learning model.
Preferably, the cross labeling comprises:
and randomly disordering sample data to be marked, and marking the extracted client level according to a majority principle.
Preferably, the performing feature processing on the marking data includes:
and (4) carrying out normalization processing on the labeled data to make the data range of each characteristic converge to the same order of magnitude.
Preferably, the machine learning model training comprises:
and inputting the training data into a classification algorithm for model fitting training, and verifying whether the classification of the obtained model is accurate through randomly extracting data, wherein the labeling data comprises the training data and the randomly extracted data.
Preferably, the data volume ratio of the training data to the randomly drawn data is preferably 7: 3.
Preferably, the processing of the analysis of the full-scale customer data in the machine learning model for the specific financial site includes:
and judging whether the classification accuracy reaches a service expected threshold, and if not, improving the classification accuracy.
Preferably, the improving the classification accuracy includes:
redefining the RFM model, adding new features, and performing data processing and model training again to enable the final value of the classification accuracy to reach a desired threshold value.
One or more technical solutions provided in the embodiments of the present invention have at least the following technical effects or advantages:
1. according to the invention, personnel classification is automatically carried out on the client in a machine learning manner, so that the cost consumed by manpower classification is reduced;
2. according to the invention, the RFM model is combined with the added specific scene characteristics under different financial scenes, and the client classification is carried out by combining a classification algorithm in a mode of randomly sampling and carrying out data annotation, so that different client classifications under different scenes are realized, and the accuracy of scene marketing can be promoted;
in conclusion, the method and the device are particularly suitable for the client precision grading treatment in different financial scenes.
Drawings
The invention is described in further detail below with reference to the following figures and detailed description:
FIG. 1 is a flow chart of the classification process of the classification method of the present invention;
FIG. 2 is a table diagram of sample data according to the present invention
Detailed Description
The following description of the embodiments of the present invention is provided for illustrative purposes, and other advantages and effects of the present invention will become apparent to those skilled in the art from the present disclosure.
Example one
Referring to fig. 1 to 2, the present invention provides a technical solution: a financial customer grading method based on RFM model generalization characteristics comprises the following steps:
s1, collecting client data, and randomly extracting sample data with a preset proportion for cross labeling; wherein the customer data comprises RFM metric data and additional feature data for a particular scenario;
s2, carrying out feature processing on the marked data, and carrying out machine learning model training;
and S3, analyzing and processing the total amount of customer data in a specific financial place in the machine learning model.
Through the steps, it is not difficult to find that in the financial client grading method, financial scenes are respectively provided with a traditional RFM model and a specific financial scene, namely the obtained client data comprise RFM index data corresponding to the traditional RFM and additional characteristic data corresponding to the specific financial scene, and the specific financial scene is graded automatically by combining a model algorithm in a mode of adding different additional characteristic data according to different financial scenes, so that a business department is assisted in completing targeted marketing analysis on clients.
The RFM index data includes the latest accounting interval (R), the accounting frequency (F) on the last 30 days, and the average settlement deposit (M) on the last 30 days.
It should be further noted that different scene characteristics are added for different financial scenes, that is, the scene characteristics are represented by RFM +; taking the loan scenario as an example, in addition to the three basic features represented by the commonly defined RFM, other features such as the number of overdue, the amount of borrowed money, and the like may be added as the scenario features.
In order to better realize the labeling processing on the sample data, the cross labeling comprises the following steps:
and randomly disordering sample data to be marked, and marking the extracted client level according to a majority principle.
In this embodiment, 10% of sample data (4500 pieces of sample data if 45000 pieces of sample data) is randomly extracted from the sampled customer data, the sample data includes RFM index data and additional feature data (e.g., overdue number and loan amount in the loan scene in fig. 2) for a specific scene, and then the sample data is cross-labeled by means of business personnel or machine equipment identification, and the sample data to be labeled is randomly disordered due to the fact that the data labeled by different business personnel may be the same, and the customer level is rapidly labeled by adopting a majority principle.
It should be noted that the financial clients include, but are not limited to, the following 8 levels: 1. important value users; 2. an important development customer; 3. important protection customers; 4. an important potential customer; 5. a general value user; 6. a general development client; 7. generally protecting the customer; 8. the customer is generally saved; indicated by the numbers 1-8, respectively.
In order to implement preprocessing of labeled data before model training, the performing feature processing on the labeled data includes:
and (4) carrying out normalization processing on the labeled data to make the data range of each characteristic converge to the same order of magnitude.
To implement model training, the machine learning model training includes:
and inputting the training data into a classification algorithm for model fitting training, and verifying whether the classification of the obtained model is accurate through randomly extracting data, wherein the labeling data comprises the training data and the randomly extracted data.
In this embodiment, after normalization processing is performed on the marked data, the marked data are used as training data for model training and randomly extracted data for verifying model accuracy, wherein the data volume of the model training data is preferably larger than the randomly extracted data volume.
It is noted that the classification model is preferably a DeepForest algorithm (forest algorithm), but is not limited to this type of algorithm, and for example, logistic regression, XGBOOST, etc. can be used to implement the present invention instead of the DeepForest algorithm.
In order to achieve model training and accuracy verification more reasonably, the data volume ratio of the training data to the randomly drawn data is preferably 7: 3.
For example, with the implementation contents, the labeled data is normalized to make the data range of each feature converge to the same order of magnitude, and at the same time, 30% of labeled and normalized data is randomly extracted as model accuracy verification test data, 70% of labeled and normalized data is used as model training data, the training data is input into a DeepForest algorithm to perform model fitting training, and the accuracy of model classification is verified through the test data after the model is obtained.
In order to better realize the client grading processing of a specific financial place, the financial place-specific analysis processing of the full-scale client data in the machine learning model comprises the following steps:
and judging whether the classification accuracy reaches a service expected threshold, and if not, improving the classification accuracy.
In this embodiment, when the classification accuracy reaches the service expectation threshold, the classification of the customer in different financial scenes is preferably automatically distinguished by feeding RFM + data.
It is added that the traffic expectation threshold is preferably 95% accuracy.
To enable processing of classification accuracy, the improving classification accuracy comprises:
redefining the RFM model, adding new features, and performing data processing and model training again to enable the final value of the classification accuracy to reach a desired threshold value.
In the embodiment, the customer data is processed and trained again in a manner of adding new features, so that the expected threshold value of the classification accuracy gradual region is realized until the expected threshold value is reached, and the customer accurate grading based on the financial scene is completed.
The foregoing embodiments are merely illustrative of the principles and utilities of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or change the above-mentioned embodiments without departing from the spirit and scope of the present invention. Accordingly, it is intended that all equivalent modifications or changes which can be made by those skilled in the art without departing from the spirit and technical spirit of the present invention be covered by the claims of the present invention.
Claims (7)
1. A financial customer grading method based on RFM model generalization characteristics is characterized by comprising the following steps:
s1, collecting client data, and randomly extracting sample data with a preset proportion for cross labeling; wherein the customer data comprises RFM metric data and additional feature data for a particular scenario;
s2, carrying out feature processing on the marked data, and carrying out machine learning model training;
and S3, analyzing and processing the total amount of customer data in a specific financial place in the machine learning model.
2. The RFM-model-generalization-feature-based financial-class customer grading method according to claim 1, wherein the cross-labeling comprises:
and randomly disordering sample data to be marked, and marking the extracted client level according to a majority principle.
3. The RFM-model-generalization-feature-based financial-like customer grading method according to claim 1, wherein said feature processing of tagged data comprises:
and (4) carrying out normalization processing on the labeled data to make the data range of each characteristic converge to the same order of magnitude.
4. The RFM-model-generalization-feature-based financial-class customer grading method according to claim 1 or 3, wherein the machine learning model training comprises:
and inputting the training data into a classification algorithm for model fitting training, and verifying whether the classification of the obtained model is accurate through randomly extracting data, wherein the labeling data comprises the training data and the randomly extracted data.
5. The RFM-model-generalization-feature-based financial customer grading method according to claim 4, wherein the data volume ratio of the training data to the randomly drawn data is preferably 7: 3.
6. The RFM-model-generalization-feature-based financial-class customer grading method according to claim 1, wherein the financial-site-specific analysis processing of the full-scale customer data in the machine learning model comprises:
and judging whether the classification accuracy reaches a service expected threshold, and if not, improving the classification accuracy.
7. The RFM generalization-based financial customer grading method according to claim 6, wherein said improving classification accuracy comprises:
redefining the RFM model, adding new features, and performing data processing and model training again to enable the final value of the classification accuracy to reach a desired threshold value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111588546.XA CN114386490A (en) | 2021-12-23 | 2021-12-23 | Financial customer grading method based on RFM model generalization characteristics |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111588546.XA CN114386490A (en) | 2021-12-23 | 2021-12-23 | Financial customer grading method based on RFM model generalization characteristics |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114386490A true CN114386490A (en) | 2022-04-22 |
Family
ID=81196968
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111588546.XA Pending CN114386490A (en) | 2021-12-23 | 2021-12-23 | Financial customer grading method based on RFM model generalization characteristics |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114386490A (en) |
-
2021
- 2021-12-23 CN CN202111588546.XA patent/CN114386490A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107248082B (en) | Card maintenance identification method and device | |
CN111291816B (en) | Method and device for carrying out feature processing aiming at user classification model | |
CN108665159A (en) | A kind of methods of risk assessment, device, terminal device and storage medium | |
CN105184315A (en) | Quality inspection treatment method and system | |
CN109583966B (en) | High-value customer identification method, system, equipment and storage medium | |
CN108133013A (en) | Information processing method, device, computer equipment and storage medium | |
CN106875206A (en) | Acquisition of information, assessment, questionnaire method, device and server | |
CN110163683B (en) | Value user key index determination method, advertisement delivery method and device | |
CN111192133A (en) | Method and device for generating risk model after user loan and electronic equipment | |
CN111882420A (en) | Generation method of response rate, marketing method, model training method and device | |
CN111210332A (en) | Method and device for generating post-loan management strategy and electronic equipment | |
CN112950359B (en) | User identification method and device | |
CN110610415A (en) | Method and device for updating model | |
CN114386490A (en) | Financial customer grading method based on RFM model generalization characteristics | |
KR102336462B1 (en) | Apparatus and method of credit rating | |
CN114626940A (en) | Data analysis method and device and electronic equipment | |
CN110570301B (en) | Risk identification method, device, equipment and medium | |
CN114693428A (en) | Data determination method and device, computer readable storage medium and electronic equipment | |
CN114549035A (en) | Construction method of financial user accurate customer acquisition label based on telecommunication big data | |
CN113870020A (en) | Overdue risk control method and device | |
CN112508689A (en) | Method for realizing decision evaluation based on multiple dimensions | |
CN113191857A (en) | Cloud platform e-commerce data processing method and system based on big data | |
CN110766544A (en) | Credit risk detection method and device, storage medium and electronic device | |
Balash et al. | Comparative analysis of innovation diffusion models: Empirical results and predictive performance on Russian mobile phone propagation data | |
CN110135467A (en) | A kind of model training method, device, system and recording medium based on data splicing |
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 |