Customer mining method and device, terminal equipment and storage medium

Abstract

The application discloses a customer mining method and device, terminal equipment and a storage medium, and the method comprises the following steps: acquiring customer characteristic data; inputting the customer characteristic data into a preset list generation model to acquire a customer list. The application solves the problem that the generation of the customer list is highly dependent on professional data analysis personnel and algorithm personnel and is difficult to meet the rapid iteration business demand in different scenes, and improves the customer mining efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a customer mining method and device, a terminal device and a storage medium. BACKGROUND

[0002] Intelligent marketing technology is a technology that uses machine learning to mine potential marketing customer lists. It can replace the original cumbersome manual rule list distribution process, not only reducing the work cost of frontline personnel in distributing marketing lists, but also achieving more accurate identification of potential target customers, thereby improving marketing success rate and performance.

[0003] At present, the process of using machine learning classification algorithms to build potential customer mining models and systems for precise marketing by the operation implementer often needs data cleaning and processing, customer feature image making, marketing model construction and evaluation, potential customer list mining and many other links for specific scenarios, which has high dependence on professional data analysts and algorithm personnel, long project cycle, and is difficult to meet the rapid iteration needs of frontline operations in different scenarios.

[0004] The above content is only used to assist in understanding the technical solutions of the present application and does not represent the acknowledgement of the above content as prior art. SUMMARY

[0005] The main purpose of the present application is to provide a customer mining method, device, terminal device and storage medium, which aims to solve the technical problem that the generation of customer lists depends on professional data analysts and algorithm personnel, and is difficult to meet the rapid iteration needs of frontline operations in different scenarios.

[0006] To achieve the above purpose, the present application provides a customer mining method, which comprises:

[0007] Obtaining customer feature data;

[0008] Inputting the customer feature data into a preset list generation model to obtain a customer list.

[0009] Optionally, the step of inputting the customer feature data into the preset list generation model to obtain the customer list further comprises:

[0010] Building a list generation model based on a preset strategy database.

[0011] Optionally, the step of building a list generation model based on a preset strategy database comprises:

[0012] Reading the strategy database to obtain task start parameters;

[0013] According to the task start parameters, a list generation model construction task is obtained.

[0014] The list generation model construction task is executed to obtain a list generation model.

[0015] Optionally, the step of reading the strategy database to obtain the task start parameter comprises:

[0016] The strategy database is read to obtain a model construction strategy configuration.

[0017] The task start parameter is obtained according to the model construction strategy configuration.

[0018] Optionally, the step of obtaining the task start parameter according to the model construction strategy configuration comprises:

[0019] The model construction strategy configuration is read to obtain a customer group data parameter, an initial customer feature data parameter and a model generation parameter.

[0020] The customer group data parameter, the initial customer feature data parameter and the model generation parameter are parsed to obtain an initial task start parameter.

[0021] The initial task start parameter is assembled to obtain the task start parameter.

[0022] Optionally, the step of executing the list generation model construction task to obtain a list generation model comprises:

[0023] The task start parameter is parsed to obtain a customer group data parameter, an initial customer feature data parameter and a model generation parameter.

[0024] An initial list generation model is obtained through model construction according to the model generation parameter.

[0025] Sample data, positive sample data and customer feature sample data are generated according to the customer group data parameter and the initial customer feature data parameter.

[0026] The sample data, the positive sample data and the customer feature sample data are input into the initial list generation model for model training to obtain a list generation model.

[0027] Optionally, the step of reading the strategy database to obtain the model construction strategy configuration further comprises:

[0028] A marketing scenario, an in-range customer group, a target customer group and a modeling parameter are obtained based on a preset business scenario.

[0029] The model construction strategy configuration is obtained by matching the marketing scenario, the in-range customer group, the target customer group and the modeling parameter.

[0030] The embodiment of the present application also provides a customer mining device, which comprises:

[0031] An acquisition module is configured to acquire customer feature data.

[0032] A generation module is configured to input the customer feature data into a preset list generation model to acquire a customer list.

[0033] The embodiment of the present application also provides a terminal device, which comprises a memory, a processor and a customer mining program stored in the memory and executable on the processor, and the processor implements the steps of the customer mining method when executing the customer mining program.

[0034] The embodiment of the present application also provides a computer readable storage medium, which stores a customer mining program, and the processor implements the steps of the customer mining method when executing the customer mining program.

[0035] The customer mining method, device and terminal device provided by the embodiment of the present application acquire customer feature data, input the customer feature data into a preset list generation model to acquire a customer list, thereby inputting the customer feature data into the pre-constructed list generation model to acquire the customer list, solving the problem that the generation of the customer list highly depends on professional data analysis personnel and algorithm personnel and is difficult to meet the rapid iteration business demand in different scenes, and improving the efficiency of list generation. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 FIG. 1 is a schematic diagram of a function block of a terminal device to which the customer mining method and device of the present application belong;

[0037] Figure 2 FIG. 2 is a schematic diagram of a flow of an exemplary embodiment of the customer mining method of the present application;

[0038] Figure 3 FIG. 3 is a schematic diagram of an overall flow of the customer mining method of the present application;

[0039] Figure 4 FIG. 4 is a schematic diagram of a flow of another exemplary embodiment of the customer mining method of the present application;

[0040] Figure 5 FIG. 5 is a schematic diagram of a flow of the customer mining method of the present application related to the construction of a list generation model;

[0041] Figure 6 FIG. 6 is a schematic diagram of a flow of the list generation method of the present application related to the acquisition of a task starting parameter;

[0042] Figure 7Another flowchart for the customer mining method of the present application involves obtaining task initiation parameters;

[0043] Figure 8 A flowchart for the list generation method of the present application involves obtaining task initiation parameters;

[0044] Figure 9 A flowchart for the customer mining method of the present application involves obtaining a list generation model;

[0045] Figure 10 A flowchart for the list generation method of the present application involves obtaining a list generation model;

[0046] Figure 11 Another flowchart for the list generation method of the present application involves obtaining a list generation model;

[0047] Figure 12 A flowchart for the customer mining method of the present application involves obtaining a model building strategy configuration;

[0048] Figure 13 A flowchart for the customer mining method of the present application involves configuring a model building strategy;

[0049] Figure 14 A flowchart for the customer mining method of the present application involves zero-code modeling.

[0050] The implementation, functional features and advantages of the present application will be further described with reference to the accompanying drawings. DETAILED DESCRIPTION

[0051] It should be understood that the specific embodiments described herein merely set forth preferred combinations of components and / or other features, and that persons of ordinary skill in the art will appreciate that many modifications are possible and can in fact be desirable within the scope of the application.

[0052] The main solution of the embodiment of the application is: based on the preset strategy database, a list generation model is constructed. The strategy database is read to obtain task start parameters; according to the task start parameters, a list generation model construction task is obtained; the list generation model construction task is executed to obtain a list generation model. The strategy database is read to obtain model construction strategy configuration; according to the model construction strategy configuration, task start parameters are obtained. The model construction strategy configuration is read to obtain customer group data parameters, initial customer feature data parameters and model generation parameters; the customer group data parameters, the initial customer feature data parameters and the model generation parameters are analyzed to obtain initial task start parameters; the initial task start parameters are assembled to obtain task start parameters. The task start parameters are analyzed to obtain customer group data parameters, initial customer feature data parameters and model generation parameters; the model is constructed through the model generation parameters to obtain an initial list generation model; according to the customer group data parameters and the initial customer feature data parameters, sample data, positive sample data and customer feature sample data are generated; the sample data, the positive sample data and the customer feature sample data are input into the initial list generation model for model training to obtain a list generation model. Based on the preset business scenario, marketing scenarios, qualified customer groups, target customer groups and modeling parameters are obtained; the marketing scenarios, the qualified customer groups, the target customer groups and the modeling parameters are matched to obtain model construction strategy configuration. Thus, the problem that customer list generation relies heavily on professional data analysis personnel and algorithm personnel and is difficult to meet the rapid iteration needs of front-line different scenarios is solved, the customer list is obtained, and the efficiency of customer mining is improved. Based on the present application, the customer mining method is designed to solve the problem that customer list generation relies heavily on professional data analysis personnel and algorithm personnel and is difficult to meet the rapid iteration needs of front-line different scenarios, and the effectiveness of the customer mining method is verified during customer mining. Finally, the efficiency of customer mining by the present application is significantly improved.

[0053] Technical terms related to the embodiments of the application:

[0054] Machine learning algorithm: Machine learning algorithm is a class of algorithms for automatically learning patterns and rules from data. They analyze and learn from training data to build prediction or decision models, and are used to predict new data or make decisions. Different machine learning algorithms have wide applications in different types of problems and data sets. Selecting the appropriate machine learning algorithm requires considering data types, problem types, requirements and advantages and disadvantages of algorithms, and conducting experiments and evaluations to obtain the best learning results.

[0055] Spark SQL: Spark SQL is a module in the Apache Spark ecosystem that provides high-level data processing functions for handling structured data and executing SQL queries, the goal of Spark SQL is to enable developers to query and analyze large-scale data using SQL statements and standard data source APIs, Spark SQL can seamlessly integrate with other Spark ecosystem components such as Spark Streaming, MLlib, and GraphX, providing a powerful and flexible data processing and analysis platform that is widely used in various big data scenarios, including data warehouse, data exploration, machine learning, and real-time analysis, etc.

[0056] HDFS: HDFS (Hadoop Distributed File System) is a distributed file system of Apache Hadoop, which is a distributed storage solution built on reliable hardware, the design goal of HDFS is to accommodate large-scale data sets and provide high-throughput data access, HDFS is widely used in the field of big data, especially combined with other components in the Apache Hadoop ecosystem (such as MapReduce, Hive, Spark), for storing and processing large-scale structured and unstructured data, it provides a reliable, high-throughput distributed file storage solution, suitable for data warehouse, data backup, data analysis, and large-scale data processing applications, etc.

[0057] The embodiment of the present application considers that the related art has the problem of high dependence on professional data analysis personnel and algorithm personnel in the process of generating a list when mining customers, and the project cycle is long, which is difficult to meet the rapid iteration of business needs in different scenarios.

[0058] Therefore, the embodiment of the present application starts from the problem that the generation of the customer list in reality depends on professional data analysis personnel and algorithm personnel, and it is difficult to meet the rapid iteration of business needs in different scenarios, designs a customer mining method, and verifies the effectiveness of the customer mining method of the present application when mining customers, and finally the efficiency of customer mining by the method of the present application is obviously improved.

[0059] Specifically, referring to Figure 1 , Figure 1 It is a functional block diagram of the terminal device to which the customer mining device of the present application belongs. The customer mining device can be a device that can independently mine customers, which can be carried on the terminal device in the form of hardware or software. The terminal device can be a smart mobile device with data processing function such as mobile phone, tablet computer, etc., and can also be a fixed terminal device or server with data processing function, etc.

[0060] In the embodiment, the terminal device to which the customer mining apparatus belongs at least comprises an output module 110, a processor 120, a memory 130 and a communication module 140.

[0061] The memory 130 stores an operating system and a customer mining program, and the customer mining apparatus can acquire customer characteristic data; the customer characteristic data is input into a preset list generation model to acquire a customer list. The list generation program is used for list generation, and the mining result and other information are stored in the memory 130; the output module 110 can be a display screen and the like. The communication module 140 can include a WIFI module, a mobile communication module, a Bluetooth module and the like, and communicates with external devices or servers through the communication module 140.

[0062] When the customer mining program in the memory 130 is executed by the processor, the following steps are implemented:

[0063] Acquire customer characteristic data;

[0064] Input the customer characteristic data into a preset list generation model to acquire a customer list.

[0065] Further, when the customer mining program in the memory 130 is executed by the processor, the following steps are implemented:

[0066] Construct a list generation model based on a preset strategy database.

[0067] Further, when the customer mining program in the memory 130 is executed by the processor, the following steps are implemented:

[0068] Read the strategy database to acquire task start parameters;

[0069] Acquire a list generation model construction task according to the task start parameters;

[0070] Execute the list generation model construction task to acquire a list generation model.

[0071] Further, when the customer mining program in the memory 130 is executed by the processor, the following steps are implemented:

[0072] Read the strategy database to acquire a model construction strategy configuration;

[0073] Acquire task start parameters according to the model construction strategy configuration.

[0074] Further, when the customer mining program in the memory 130 is executed by the processor, the following steps are implemented:

[0075] Read the model construction strategy configuration to obtain customer group data parameters, initial customer feature data parameters and model generation parameters;

[0076] Parse the customer group data parameters, initial customer feature data parameters and model generation parameters to obtain initial task start parameters;

[0077] Assemble the initial task start parameters to obtain task start parameters.

[0078] Further, the customer mining in the memory 130 is further implemented when executed by the processor to implement the following steps:

[0079] Parse the task start parameters to obtain customer group data parameters, initial customer feature data parameters and model generation parameters;

[0080] Construct a model through the model generation parameters to obtain an initial list generation model;

[0081] Generate sample data, positive sample data and customer feature sample data according to the customer group data parameters and initial customer feature data parameters;

[0082] Input the sample data, positive sample data and customer feature sample data into the initial list generation model for model training to obtain a list generation model.

[0083] Further, the customer mining program in the memory 130 is further implemented when executed by the processor to implement the following steps:

[0084] Obtain a marketing scenario, an inducted customer group, a target customer group and modeling parameters based on a preset business scenario;

[0085] Match the marketing scenario, the inducted customer group, the target customer group and the modeling parameters to obtain a model construction strategy configuration.

[0086] The above scheme is used to obtain customer feature data, input the customer feature data into a preset list generation model, and obtain a customer list. The problem that customer list generation is highly dependent on professional data analysis personnel and algorithm personnel and is difficult to meet the needs of rapid iteration in different scenarios can be solved. Based on the present invention, a customer mining method is designed to solve the problem that customer list generation is highly dependent on professional data analysis personnel and algorithm personnel and is difficult to meet the needs of rapid iteration in different scenarios. The effectiveness of the customer mining method is verified during customer mining. Finally, the efficiency of customer mining using the present invention is significantly improved.

[0087] Based on the above terminal device architecture but not limited to the above framework, the present invention method embodiment is proposed.

[0088] Referring to Figure 2 , Figure 2 is a flowchart of an exemplary embodiment of a customer mining method. The list generation method comprises:

[0089] Step S01, obtaining customer characteristic data;

[0090] The execution subject of the method of the embodiment can be a customer mining device, a customer mining terminal device or a server. The embodiment is exemplified by a customer mining device, which can be integrated on a terminal device with data processing function.

[0091] In order to obtain customer characteristic data, the following steps are implemented:

[0092] First, in the embodiment, the customer characteristic data refers to data describing and recording various attributes, characteristics and information of customers. These data are mainly used for analyzing customer groups, personalized services and decision making, etc. Collecting and analyzing customer characteristic data can help enterprises better understand and meet customer needs;

[0093] Finally, the customer characteristic data is obtained according to the actual business operation and activity strategy, wherein the obtained data includes but is not limited to personal information, geographic location, purchase behavior, interest, social media data and customer feedback, etc.

[0094] Step S03, inputting the customer characteristic data into a preset list generation model to obtain a customer list.

[0095] After obtaining the customer characteristic data, the customer list is obtained by the following steps:

[0096] First, input the customer characteristic data into the pre-built list generation model, wherein the list generation model is built by task-driven mode, which is not limited by place and personnel, and is constructed by pre-configured task starting parameters, including but not limited to marketing scene, shortlisted customer group, target customer group and modeling parameter;

[0097] Finally, the corresponding customer list is obtained by predicting the customer characteristic data through the list generation model.

[0098] Specifically, as shown in Figure 3 , Figure 3 is a flowchart of the overall process of the customer mining method.

[0099] First, the list creation module provides intelligent lead generation process services tailored to specific business scenarios. This primarily involves standardizing the configuration steps for lead generation, enabling zero-threshold AI modeling and list generation. Specifically, this includes: configuring the business scenario (selecting specific marketing scenarios through the module's scenario system, such as Didi Chuxing, offline catering events, wealth management product promotions, and micro-loan promotions); configuring the initial customer group (configuring rules to filter the initial customer group using the module's basic customer tag library, such as customers aged 20 to 60); and configuring the target customer group (configuring the target customer group using the module's customer behavior tag library or event tracking library). Success criteria for targeting the marketing audience include, for example, identifying financial product purchases within 7 days of campaign launch; configuring AI algorithm templates, utilizing the module's algorithm template library to configure specific model processing templates for the current list mining campaign, where AI algorithms include, but are not limited to, machine learning algorithms such as LR, Random Forest, XGBoost, and LightGBM; configuring the customer feature list for AI analysis, using the module's customer feature library to configure the feature list for input to the model; initiating AI modeling and generating the list: based on the above business configuration, initiating the AI ​​model training and evaluation task, and after the model is generated, outputting high-intent potential list data based on model scoring;

[0100] Then, the model template service module provides algorithm template information to the list creation service module. At the same time, it can transform the business configuration in the list creation activity into the input parameters of the model processing task, such as: transforming the shortlisted customer group label screening rules into sample data reading Spark SQL statements for model input.

[0101] Then, the scenario service module provides the list creation service module with scenario classification information, customer tag information, customer group tracking event information, and customer feature profile information.

[0102] Then, the task management module starts or stops the model processing task, monitors the task execution progress on the engine side, and provides task log query service.

[0103] Then, the list delivery module is used to deliver the potential customer list created by the user to the downstream application system;

[0104] Then, the platform management module provides business configuration (such as user registration) and system operation report services;

[0105] Then, a front-end web access service is provided through the web service module;

[0106] Then, the monitoring and alarm service module provides the system with system monitoring and alarm services such as main link error logs, model effect decay, and data anomalies to ensure the reliability of the system.

[0107] Then, the model processing task queuing and the task execution after the resource is ready are completed through the task scheduling engine service.

[0108] Then, the authentication and trust of the big data processing cluster are performed through the model processing front-end service, the model processing workflow job script deployment node is deployed, and the job execution is started by the scheduling engine service.

[0109] Finally, the distributed engine for model processing is completed through the big data processing cluster, including but not limited to Hadoop, Spark, Yarn and other big data cluster components.

[0110] The embodiment obtains customer feature data, inputs the customer feature data into a preset list generation model, and obtains a customer list. Thus, the generation of the customer list is completed, the problem that the generation of the customer list highly depends on professional data analysis personnel and algorithm personnel and cannot meet the rapid iteration business demand in different scenes is solved, and the efficiency of customer mining is improved.

[0111] Referring to Figure 4 , Figure 4 FIG. 2 is a flowchart of another exemplary embodiment of the customer mining method of the present application.

[0112] Based on the above Figure 2 embodiment, before the step S03 of inputting the customer feature data into a preset list generation model and obtaining a customer list, the step further includes:

[0113] In step S02, a list generation model is constructed based on a preset strategy database.

[0114] Specifically, the construction of the list generation model is completed through the following steps:

[0115] First, the construction of the list generation model is based on task start parameters to generate a task.

[0116] Then, the task start parameters are stored in the strategy database, mainly including model construction strategy configuration.

[0117] Finally, according to the specific business scenario, the corresponding model construction strategy configuration is obtained, and the construction of the list generation model is completed.

[0118] More specifically, as Figure 5 shown, Figure 5 FIG. 3 is a flowchart of the process of constructing a list generation model involved in the customer mining method of the present application.

[0119] Based on the above Figure 4 embodiment, the step S02 of constructing a list generation model based on a preset strategy database includes:

[0120] Step S021, reading the policy database to obtain task starting parameters;

[0121] Step S022, obtaining a list generation model construction task according to the task starting parameters;

[0122] Step S023, executing the list generation model construction task to obtain a list generation model.

[0123] Further, in order to obtain a list generation model, the following steps are implemented:

[0124] First, the policy database is read to obtain the shortlisted customer group screening rule, the target customer group success criterion, the AI algorithm, and the AI analyzed customer features;

[0125] Then, the shortlisted customer group screening rule, the target customer group success criterion, the AI algorithm, and the AI analyzed customer features are parsed and assembled according to the obtained task starting parameters;

[0126] Then, according to the task starting parameters, the construction task of the list generation model is obtained;

[0127] Finally, the list generation model construction task is executed to obtain the list generation model.

[0128] The above-mentioned scheme is used to construct a list generation model based on a preset policy database. Thus, the construction of the list generation model is realized, the problem of no corresponding model for customer feature data for customer list generation is solved, and the efficiency of customer mining is improved.

[0129] Referring to Figure 6 , Figure 6 The flowchart for obtaining the task starting parameters of the list generation method of the present application is shown.

[0130] Based on the above Figure 5 embodiment, the step S021 of reading the policy database to obtain the task starting parameters comprises:

[0131] Step S0213, reading the policy database to obtain a model construction strategy configuration;

[0132] Step S0214, obtaining the task starting parameters according to the model construction strategy configuration.

[0133] Specifically, the task starting parameters are obtained by the following steps:

[0134] Firstly, the strategy database is read to obtain the business scene, the shortlisted customer group label screening rule, the success criterion of the target customer group, the specific AI algorithm adopted and the customer characteristic list of AI analysis and generate a model construction strategy configuration;

[0135] Finally, the model construction strategy business data configured by the user is converted into the input parameters of the model processing task to obtain the task starting parameters.

[0136] The embodiment obtains the model construction strategy configuration by reading the strategy database, and obtains the task starting parameters according to the model construction strategy configuration. Thus, the task starting parameters are obtained, the problem that there is no corresponding task parameter when the list is generated is solved, and the efficiency of list generation is improved.

[0137] Referring to Figure 7 , Figure 7 Another process schematic diagram for obtaining the task starting parameters for the customer mining method of the present application.

[0138] Based on the above Figure 6 embodiment, the step S0214 of obtaining the task starting parameters according to the model construction strategy configuration includes:

[0139] Step S02141, reading the model construction strategy configuration to obtain customer group data parameters, initial customer feature data parameters and model generation parameters;

[0140] Step S02142, analyzing the customer group data parameters, the initial customer feature data parameters and the model generation parameters to obtain initial task starting parameters;

[0141] Step S02143, assembling the initial task starting parameters to obtain the task starting parameters.

[0142] Specifically, in order to obtain the task starting parameters, the following steps are implemented:

[0143] Firstly, the model construction strategy configuration is read to obtain customer group information, initial customer data parameters and model generation parameters, wherein the customer group information, the initial customer data parameters and the model generation parameters include but are not limited to shortlisted customer group label screening rule information, target customer group success criterion information, AI analysis characteristic list and AI algorithm matching specific algorithm model processing workflow template;

[0144] Then, the shortlisted customer group label screening rule information is analyzed to assemble the read Spark SQL statement of the model training sample data;

[0145] Then, according to the success criterion information of the target customer group, parse the read Spark SQL statement of the training positive sample data of the assembled model after parsing;

[0146] Then, according to the user-selected AI analysis feature list, parse the read Spark SQL statement of the customer feature data of the assembled model after parsing;

[0147] Then, according to the user-selected AI algorithm, match the specific algorithm model processing workflow template;

[0148] Finally, assemble the input parameters of the algorithm model, including: sampling strategy, feature selection strategy, algorithm hyperparameter, etc., to obtain the task start parameters.

[0149] More specifically, as shown in Figure 8 , Figure 8 The list generation method of the present application relates to the schematic diagram of obtaining task start parameters.

[0150] First, read the list making strategy from the database, obtain the shortlisted customer group screening rules, success criteria of the target customer group, AI algorithm and customer features of AI analysis;

[0151] Then, according to the sample data read Spark SQL, positive sample data read Spark SQL, feature data read Spark SQL, and algorithm template and algorithm parameters obtained by the shortlisted customer group screening rules, success criteria of the target customer group, AI algorithm and customer features of AI analysis;

[0152] Finally, according to the sample data read Spark SQL, positive sample data read Spark SQL, feature data read Spark SQL, and algorithm template and algorithm parameters, assemble to obtain the task start parameters.

[0153] The embodiment obtains the customer group data parameters, initial customer feature data parameters and model generation parameters by reading the model construction strategy configuration; parses the customer group data parameters, initial customer feature data parameters and model generation parameters to obtain the initial task start parameters; and assembles the initial task start parameters to obtain the task start parameters. Thus, the task start parameters are obtained, the problem of high dependence of customer list generation on professional data analysis personnel and algorithm personnel is solved, and the efficiency of customer mining is improved.

[0154] Referring to Figure 9 , Figure 9 The customer mining method of the present application relates to the flowchart of obtaining the list generation model.

[0155] Based on the above Figure 5In the illustrated embodiment, the step S023 of performing the list generation model construction task includes the following steps of:

[0156] In step S0231, the task start parameters are parsed to obtain the customer group data parameters, initial customer feature data parameters, and model generation parameters.

[0157] In step S0232, the model is constructed by using the model generation parameters to obtain an initial list generation model.

[0158] In step S0233, sample data, positive sample data, and customer feature sample data are generated based on the customer group data parameters and the initial customer feature data parameters.

[0159] In step S0234, the sample data, positive sample data, and customer feature sample data are input into the initial list generation model for model training to obtain a list generation model.

[0160] Specifically, in order to obtain a list generation model for customer list acquisition, the following steps are implemented:

[0161] First, the task start parameters are parsed to obtain sample data, positive sample data, customer feature sample data, and model generation parameters for model training. The positive sample data refers to training sample data representing the target category to be identified or predicted in a classification problem.

[0162] Then, the initial list generation model is obtained by using the model generation parameters.

[0163] Then, the sample data, positive sample data, and customer feature sample data are input into the initial list generation model for model training to obtain a list generation model.

[0164] Further, as shown in Figure 10 , Figure 10 The list generation method of the present application relates to a schematic diagram for obtaining a list generation model.

[0165] First, the model construction task is scheduled and queued.

[0166] Then, the model construction task is started, and the initial list generation model is generated by using the model generation parameters.

[0167] Then, the input parameters of the task are parsed to obtain the positive sample data, sample data, and customer feature sample data.

[0168] Then, the positive sample data, sample data, and customer feature sample data are filtered.

[0169] Then, the screened data is used for model training and evaluation to obtain the evaluation result of the model;

[0170] Then, the evaluation result of the model is analyzed, and when the evaluation result of the model meets the accuracy of the business requirement, the model is output as a list generation model;

[0171] Finally, the latest customer feature data is input into the trained list generation model to obtain a customer list.

[0172] Further, as shown in Figure 11 , Figure 11 is a schematic diagram for another embodiment of the list generation method of the present application.

[0173] First, data processing is performed, and the specific steps include obtaining full sample data, positive sample data, and feature data, merging the data, performing null value processing, adaptive sampling, and automatic feature screening, etc.

[0174] Then, the processed data is input into the initial model for training, and the XGBoost model is used in the embodiment, and the specific steps include sample splitting, automatic feature conversion, XGBoost binary classification training, prediction using the XGBoost model, and evaluation of the current model effect, etc., wherein XGBoost (eXtreme Gradient Boosting) is a machine learning model based on gradient boosting algorithm. It is an efficient, flexible and widely used ensemble learning method;

[0175] Finally, the trained model is used for result output, and the specific steps include data processing, reading full sample and feature data, merging the data and performing null value processing; using the model for prediction, loading the XGBoost model, feature information loading, using the XGBoost model for prediction, and converting the model result; extracting the result, extracting the list, evaluating the business effect, analyzing the crowd portrait, generating the list explanatory language, and outputting the result.

[0176] The embodiment obtains the customer group data parameter, the initial customer feature data parameter and the model generation parameter by analyzing the task starting parameter, constructs a model by the model generation parameter, obtains an initial list generation model, generates sample data, positive sample data and customer feature sample data according to the customer group data parameter and the initial customer feature data parameter, inputs the sample data, the positive sample data and the customer feature sample data into the initial list generation model for model training, and obtains a list generation model. Thus, the list generation model is obtained, the problem that the customer list generation is highly dependent on professional data analysis personnel and algorithm personnel and is difficult to meet the rapid iteration needs of different scenes in the front line is solved, and the customer mining efficiency is improved.

[0177] Referring to Figure 12 , Figure 12 The customer mining method of the present application relates to a process schematic diagram for obtaining a model construction strategy configuration.

[0178] Based on the above Figure 6 The step S0213 of reading the strategy database to obtain the model construction strategy configuration further includes the following steps before the step S0213:

[0179] In step S0211, the marketing scenario, the shortlisted customer group, the target customer group and the modeling parameter are obtained based on the preset business scenario.

[0180] In step S0212, the model construction strategy configuration is obtained by matching the marketing scenario, the shortlisted customer group, the target customer group and the modeling parameter.

[0181] Specifically, in order to obtain the model construction strategy configuration, the following steps are implemented:

[0182] Firstly, the corresponding marketing scenario, shortlisted customer group, target customer group and modeling parameter are obtained according to different business scenarios. For example, the current business scenario is a credit card business, at this time, the shortlisted customer group should be people in the working stage, so according to the characteristics of work, people aged 22 to 55 are selected as the shortlisted customer group, and among these people, the target customer group can be selected according to their income level, and the modeling parameter should be selected according to the specific operation to select the appropriate model type and then obtain the modeling parameter.

[0183] Finally, the marketing scenario, the shortlisted customer group, the target customer group and the modeling parameter are matched to obtain the corresponding list making strategy configuration.

[0184] More specifically, as shown in Figure 13 Figure 13 ​The method for customer mining of the application relates to a schematic diagram of configuring a model construction strategy.

[0185] Firstly, the shortlisted customer screening tag library includes but is not limited to customer basic tags, wealth asset tags and black and gray list tags, etc.

[0186] Then, the target customer screening event library includes but is not limited to product purchase events, activity participation events and page browsing events, etc.

[0187] Finally, the feature library includes but is not limited to access conditions, basic information and transaction behaviors, etc.

[0188] Further, as shown in the figure, Figure 14 the method for customer mining of the application relates to a schematic diagram of zero-code modeling. Figure 14

[0189] Firstly, in this embodiment, the object of modeling can be a specific business personnel, and modeling and list generation are performed according to business requirements.

[0190] Then, the step of creating a task and inputting basic information includes but is not limited to activity name, activity description and activity members, wherein the activity members are non-mandatory items.

[0191] Then, the activity is involved, and the model parameters are selected, wherein the steps include but are not limited to selecting scenarios and purposes, selecting shortlisted customers (sample data), selecting target customers (positive sample data), checking features, selecting algorithm types and selecting time windows.

[0192] Then, model training and model evaluation are performed, wherein the model evaluation is a non-essential operation and the steps thereof include selecting feature data date ranges.

[0193] Then, list generation includes but selects feature data date ranges, and the list is generated through the trained model.

[0194] Finally, the list is delivered to the corresponding system, including but not limited to selecting the channel to be delivered and filling in the delivery list quantity, etc.

[0195] Through the above-mentioned scheme, the marketing scenario, shortlisted customers, target customers and modeling parameters are obtained based on the preset business scenario, the marketing scenario, shortlisted customers, target customers and modeling parameters are matched, and the model construction strategy configuration is obtained. Thus, the model construction strategy configuration is completed, the zero-code list generation model is constructed, the problem of high dependence of customer list generation on professional data analysis personnel and algorithm personnel is solved, and the efficiency of customer mining is improved.

[0196] ​Furthermore, the embodiment of the present application also proposes a customer mining device, the customer mining device comprises:

[0197] The acquisition module is configured to acquire customer feature data.

[0198] The generation module is configured to input the customer feature data into a preset list generation model to obtain a customer list.

[0199] Furthermore, the embodiment of the present application also proposes a terminal device, the terminal device comprises a memory, a processor, and a customer mining program stored in the memory and executable on the processor, and the customer mining program, when executed by the processor, implements the steps of the customer mining method.

[0200] Since the customer mining program, when executed by the processor, adopts all the technical solutions of the aforementioned all embodiments, it at least has all the beneficial effects brought by all the technical solutions of the aforementioned all embodiments, which will not be repeated here.

[0201] Furthermore, the embodiment of the present application also proposes a computer readable storage medium, the computer readable storage medium stores a customer mining program, and the customer mining program, when executed by the processor, implements the steps of the customer mining method.

[0202] Since the customer mining program, when executed by the processor, adopts all the technical solutions of the aforementioned all embodiments, it at least has all the beneficial effects brought by all the technical solutions of the aforementioned all embodiments, which will not be repeated here.

[0203] Compared with the prior art, the customer mining method, device, terminal device, and storage medium proposed by the embodiment of the present application acquire customer feature data; input the customer feature data into a preset list generation model to obtain a customer list. Based on the present application, the customer list generation in reality depends on professional data analysis personnel and algorithm personnel, which is difficult to meet the rapid iteration needs of the front line in different scenarios. The customer mining method is designed, and the effectiveness of the customer mining method of the present application is verified in customer mining. Finally, the efficiency of customer mining by the method of the present application is obviously improved.

[0204] Compared with the prior art, the embodiment of the present application has the following advantages:

[0205] 1. A scenario-based intelligent potential customer mining method and system based on machine learning and big data technology are disclosed, which realizes automatic identification of potential customers in different scenarios.

[0206] 2. Disclose a general machine learning modeling processing workflow template and input parameter standard corresponding to the template in the marketing field, which can meet different modeling goals in various scenarios;

[0207] 3. Disclose a zero-code AI modeling method in the marketing field, which enables personnel in non-algorithm positions to perform modeling operations in a process-oriented and low-threshold manner, and realizes the rapid landing of potential customer mining needs.

[0208] It should be noted that in this paper, the term "includes", "contains" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or system including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or includes elements inherent to such process, method, article or system. Without more limitations, the element defined by the sentence "includes a" does not exclude the presence of another identical element in the process, method, article or system including the element.

[0209] The above-mentioned embodiment numbers of the application are only for description, not representing the advantages and disadvantages of the embodiments.

[0210] From the above description of the embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be realized by software and necessary general hardware platform, of course, they can also be realized by hardware, but in many cases the former is a better embodiment. Based on such understanding, the technical solutions of the present application or the part that contributes to the prior art can be embodied in the form of software product, which is stored in the above-mentioned storage medium (such as ROM / RAM, magnetic disc, optical disc), including a plurality of instructions for making a terminal device (which can be a mobile phone, computer, server, controlled terminal or network device, etc.) execute the method of each embodiment of the present application.

[0211] The above is only the preferred embodiment of the present application, and does not limit the patent scope of the present application, and any equivalent structure or equivalent process transformation, or direct or indirect application in other related technical fields based on the content of the present application specification and drawings, are also included in the patent protection scope of the present application.

Claims

1. A customer mining method, characterized in that, The customer acquisition method includes the following steps: Obtain customer characteristic data; The customer feature data is input into a preset list generation model to obtain a customer list. The list generation model is a machine learning model based on the gradient boosting algorithm and is built in a task-driven manner. Before the step of inputting the customer feature data into a preset list generation model to obtain the customer list, the method further includes: A list generation model is built based on a pre-defined strategy database; The steps for constructing the list generation model based on a preset strategy database include: The strategy database is read to obtain task startup parameters, which include marketing scenario, shortlisted customer group, target customer group and modeling parameters; Based on the task startup parameters, obtain the list generation model construction task; Execute the list generation model construction task to obtain the list generation model.

2. The customer mining method according to claim 1, characterized in that, The step of reading the strategy database to obtain task startup parameters includes: The strategy database is read to obtain the model building strategy configuration; Based on the model construction strategy configuration, obtain the task startup parameters.

3. The customer mining method according to claim 2, characterized in that, The step of obtaining task startup parameters based on the model-based strategy configuration includes: The model building strategy configuration is read to obtain customer group data parameters, initial customer characteristic data parameters, and model generation parameters; The customer group data parameters, initial customer characteristic data parameters, and model generation parameters are parsed to obtain the initial task start parameters; The initial task startup parameters are assembled to obtain the task startup parameters.

4. The customer mining method according to claim 3, characterized in that, The steps of executing the list generation model construction task and obtaining the list generation model include: The task startup parameters are parsed to obtain customer group data parameters, initial customer characteristic data parameters, and model generation parameters; The model is constructed using the model generation parameters to obtain the initial list generation model; Based on the customer group data parameters and the initial customer characteristic data parameters, generate sample data, positive sample data, and customer characteristic sample data; The sample data, positive sample data, and customer feature sample data are input into the initial list generation model for model training to obtain the list generation model.

5. The customer mining method according to claim 2, characterized in that, Before the step of reading the strategy database to obtain the model building strategy configuration, the method further includes: Based on the preset business scenario, obtain the marketing scenario, shortlisted customer group, target customer group and modeling parameters; The model building strategy configuration is obtained by matching the marketing scenario, shortlisted customer groups, target customer groups, and modeling parameters.

6. A customer excavation device, characterized in that, The customer discovery device includes: The acquisition module is used to acquire customer characteristic data; The generation module is used to input the customer feature data into a preset list generation model to obtain a customer list; The generation module is further configured to: A list generation model is built based on a pre-defined strategy database. The list generation model is a machine learning model based on the gradient boosting algorithm and is built in a task-driven manner. The generation module is further configured to: The strategy database is read to obtain task startup parameters, which include marketing scenario, shortlisted customer group, target customer group and modeling parameters; Based on the task startup parameters, obtain the list generation model construction task; Execute the list generation model construction task to obtain the list generation model.

7. A terminal device, characterized in that, The terminal device includes a memory, a processor, and a customer mining program stored in the memory and executable on the processor. When the customer mining program is executed by the processor, it implements the steps of the customer mining method as described in any one of claims 1-5.

8. A calculator-readable storage medium, characterized in that, The calculator-readable storage medium stores a customer mining program, which, when executed by a processor, implements the steps of the customer mining method as described in any one of claims 1-5.

Images