Customer obtaining analysis method and device based on transaction data, equipment and medium

Abstract

The invention relates to the field of artificial intelligence, and discloses a customer obtaining analysis method, device and equipment based on transaction data and a medium, and the method comprises the steps: collecting the transaction data of a to-be-analyzed customer group, and carrying out the preprocessing, and generating a standard data set; performing feature screening on the standard data set, and performing individual feature extraction and clustering analysis to obtain individual features and group features; according to the individual features, the group features and behavior time sequence features in the transaction data, user portrait data are constructed and obtained; inputting the user portrait data into the trained behavior prediction model, performing behavior prediction processing on the user portrait data according to a time sequence weighting strategy, and generating intention behavior information; and generating a customer marketing strategy according to the intention behavior information, and sending marketing information to the customer through a specified channel at specified triggering time according to the customer marketing strategy. By constructing comprehensive user portraits and utilizing a time sequence weighting strategy to accurately predict future behaviors of customers, accurate customer acquisition analysis is realized, and customer acquisition efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the fields of artificial intelligence technology and financial technology, and in particular to customer acquisition analysis methods, devices, equipment and media based on transaction data. Background Art

[0002] How to acquire customers efficiently and accurately is one of the keys to ensuring the efficient operation of the sales team. In the past, insurance companies acquired customers through the agents' own abilities. Most agents have a certain customer base, but apart from the customer base they operate in depth, they lack accurate customer acquisition analysis for other customer groups with less interaction and not fully understood, making it difficult to conduct accurate personalized marketing and acquire customers through effective channels, thereby reducing customer conversion rates. Summary of the invention

[0003] In view of the above-mentioned deficiencies in the prior art, the purpose of the present invention is to provide a customer acquisition analysis method, device, equipment and medium based on transaction data that can be applied to financial technology or other related fields. Its main purpose is to achieve accurate customer acquisition analysis and provide personalized marketing for different customers to improve customer acquisition efficiency.

[0004] The technical solution of the present invention is as follows:

[0005] A first aspect of the present invention provides a customer acquisition analysis method based on transaction data, comprising:

[0006] Collecting transaction data of the customer group to be analyzed and preprocessing the transaction data to generate a standard data set;

[0007] Performing feature screening on the standard data set, and performing individual feature extraction and cluster analysis on the screened standard data set to obtain corresponding individual features and group features;

[0008] Constructing user portrait data based on the individual characteristics and group characteristics and the behavioral time series characteristics in the transaction data;

[0009] Inputting the user portrait data into a trained behavior prediction model, performing behavior prediction processing on the user portrait data according to a time series weighted strategy through the behavior prediction model to generate intended behavior information;

[0010] A corresponding customer acquisition marketing strategy is generated according to the intended behavior information, and marketing information is sent to the customer through a specified channel at a specified trigger time according to the customer acquisition marketing strategy.

[0011] A second aspect of the present invention provides a customer acquisition analysis device based on transaction data, comprising:

[0012] A collection module, used to collect transaction data of the customer group to be analyzed and pre-process the transaction data to generate a standard data set;

[0013] A feature extraction module is used to perform feature screening on the standard data set, and perform individual feature extraction and cluster analysis on the screened standard data set to obtain corresponding individual features and group features;

[0014] A construction module, used to construct user portrait data according to the individual characteristics and group characteristics and the behavioral time series characteristics in the transaction data;

[0015] A customer acquisition prediction module, which is used to input the user portrait data into a trained behavior prediction model, perform behavior prediction processing on the user portrait data according to a time series weighted strategy through the behavior prediction model, and generate intended behavior information;

[0016] The customer acquisition strategy generation module is used to generate a corresponding customer acquisition marketing strategy based on the intended behavior information, and send marketing information to the customer through a specified channel at a specified trigger time according to the customer acquisition marketing strategy.

[0017] A third aspect of the present invention provides a computer device, comprising at least one processor; and

[0018] a memory communicatively connected to the at least one processor; wherein,

[0019] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above-mentioned customer acquisition analysis method based on transaction data.

[0020] A fourth aspect of the present invention provides a non-volatile computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are executed by one or more processors, the one or more processors can execute the above-mentioned customer acquisition analysis method based on transaction data.

[0021] Beneficial effects: The present invention discloses a customer acquisition analysis method, device, equipment and medium based on transaction data. Compared with the prior art, the embodiment of the present invention generates a standard data set by collecting the transaction data of the customer group to be analyzed and preprocessing the transaction data; performs feature screening on the standard data set, and performs individual feature extraction and cluster analysis on the screened standard data set to obtain corresponding individual features and group features; constructs user portrait data based on the individual features and group features and the behavioral time series features in the transaction data; inputs the user portrait data into a trained behavior prediction model, and performs behavior prediction processing on the user portrait data according to the time series weighting strategy through the behavior prediction model to generate intention behavior information; generates a corresponding customer acquisition marketing strategy based on the intention behavior information, and sends marketing information to customers through a specified channel at a specified trigger time based on the customer acquisition marketing strategy. A comprehensive user portrait is constructed by extracting individual and group features and behavioral time series features, and the behavior prediction model and time series weighting strategy are used to accurately predict the customer's future behavior, automatically generate personalized customer acquisition strategies, achieve accurate customer acquisition analysis, and improve conversion rate and customer satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the scheme in the present invention, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0023] Figure 1 A schematic diagram of an application environment of a customer acquisition analysis method based on transaction data provided by an embodiment of the present invention;

[0024] Figure 2 A flow chart of a customer acquisition analysis method based on transaction data provided by an embodiment of the present invention;

[0025] Figure 3 A flow chart of step S201 in the customer acquisition analysis method based on transaction data provided by an embodiment of the present invention;

[0026] Figure 4 A flow chart of step S202 in the customer acquisition analysis method based on transaction data provided by an embodiment of the present invention;

[0027] Figure 5 A flow chart of step S204 in the customer acquisition analysis method based on transaction data provided by an embodiment of the present invention;

[0028] Figure 6 A schematic diagram of functional modules of a customer acquisition analysis device based on transaction data provided by an embodiment of the present invention;

[0029] Figure 7 A schematic diagram of the hardware structure of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0030] In order to make the purpose, technical solution and effect of the present invention clearer and more specific, the present invention is further described in detail below. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. The embodiments of the present invention are described below in conjunction with the accompanying drawings.

[0031] The customer acquisition analysis method based on transaction data provided by the embodiment of the present invention can be applied in the following aspects: Figure 1 In the application environment of FIG. 1 , it includes a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104 and a server 105. The network 104 is used to provide a medium for a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103 and the server 105. The network 104 may include various connection types, such as wired and / or wireless communication links, etc.

[0032] The user can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as knowledge reading applications, web browser applications, search applications, instant messaging tools, email clients, and / or social platform software, etc. (only as examples).

[0033] The first terminal device 101, the second terminal device 102, and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0034] The server 105 may be a server that provides various services, such as a background server that provides support for the content browsed by the user using the first terminal device 101, the second terminal device 102, and the third terminal device 103 (for example only). The background server may analyze and process the received data such as the user request, and feed back the processing results (such as web pages, information, or data obtained or generated according to the user request) to the terminal device. The server 105 may be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services ("Virtual Private Server", or "VPS" for short). The server 105 may also be a server for a distributed system, or a server combined with a blockchain.

[0035] It should be noted that the customer acquisition analysis method based on transaction data provided in the embodiment of the present application can generally be executed by the first terminal device 101, the second terminal device 102 or the third terminal device 103. Accordingly, the customer acquisition analysis device based on transaction data provided in the embodiment of the present invention can also be set in the first terminal device 101, the second terminal device 102 or the third terminal device 103. Alternatively, the customer acquisition analysis method based on transaction data provided in the embodiment of the present invention can also generally be executed by the server 105. Accordingly, the customer acquisition analysis device based on transaction data provided in the embodiment of the present invention can generally be set in the server 105.

[0036] It should be understood that the numbers of the above terminal devices, networks and servers are only illustrative. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0037] like Figure 2 As shown, the customer acquisition analysis method based on transaction data provided by the embodiment of the present invention specifically includes the following steps:

[0038] S201. Collect transaction data of the customer group to be analyzed and pre-process the transaction data to generate a standard data set.

[0039] In this embodiment, the transaction data of the customer group to be analyzed is collected as the basis for customer acquisition analysis. The specific customer group to be analyzed refers to new customers with transaction data or customers who have less interaction with the enterprise and insufficient data collection, making it difficult to conduct accurate personalized marketing for them. In the financial insurance scenario, the acquired transaction data may include, for example, policy information, basic customer information, claims records, etc. Since these transaction data may contain missing values, outliers, and inconsistent formats, the collected transaction data is pre-processed, such as data cleaning (such as removing spaces, standardizing city names, etc.), processing missing values ​​(filling medians or modes), and outlier processing (replacing or deleting with specific values), so as to generate a standard data set. The format of the pre-processed standard data set is unified, which provides a high-quality data foundation for subsequent customer acquisition analysis and improves the accuracy and robustness of predictive analysis.

[0040] S202: Perform feature screening on the standard data set, and perform individual feature extraction and cluster analysis on the screened standard data set to obtain corresponding individual features and group features.

[0041] In this embodiment, feature engineering and selection are further performed based on the standard data set obtained by preprocessing, and the most valuable features for analysis are extracted and selected from the original data, such as policy information (such as insurance type, insured amount, insurance period, etc.), customer statistical information (such as age, gender, occupation, etc.), key analysis factors (such as "coverage", "liability limit", etc.), etc. These features provide a basis for understanding customer needs and preferences. By performing individual feature extraction and cluster analysis on the screened standard data set, corresponding individual features and group features are obtained, where individual features refer to independent features of each customer, such as individual age, gender, occupation, etc., and group features refer to the behavior patterns and preferences of customer groups, such as different groups may have different purchasing habits and needs, and different groups may have common preferences for certain characteristics of insurance products, thereby providing more comprehensive and rich feature data for building user portraits, so as to conduct more accurate marketing and product recommendations.

[0042] S203: Construct user portrait data based on the individual characteristics and group characteristics and the behavioral time series characteristics in the transaction data.

[0043] In this embodiment, based on the extracted individual characteristics and group characteristics, and further combined with the behavioral time series characteristics in the transaction data, user portrait data is jointly constructed, wherein individual characteristics include age, gender, occupation, income, risk preference, etc., groups include behavioral group behavior patterns, group preferences, etc., and behavioral time series characteristics include user behavior records over time, such as purchase history, access behavior, interactive behavior, etc. Individual characteristics help to identify customers' personal needs and preferences, thereby predicting which products they may be interested in, while group characteristics help to identify customer groups with similar behaviors and preferences, so as to conduct more targeted marketing, and behavioral time series characteristics help to predict customers' future behavior, such as whether they may make purchases at a specific time point. Therefore, comprehensive user portrait data can provide a description of customer characteristics and preferences, as well as behavioral patterns and trends, so that customers' future behavior can be predicted, and personalized customer acquisition marketing strategies can be provided to improve customer acquisition efficiency and accuracy.

[0044] S204: input the user portrait data into a trained behavior prediction model, and use the behavior prediction model to perform behavior prediction processing on the user portrait data according to a time series weighting strategy to generate intended behavior information.

[0045] In this embodiment, behavior prediction is performed on user portrait data through a pre-constructed and trained behavior prediction model. The behavior prediction model is specifically constructed and trained based on an LSTM network with an attention mechanism. In the behavior prediction model, behavior prediction processing is performed on user portrait data through a time series weighted strategy. The strategy dynamically adjusts the weights of each time point in the sequence according to the timeliness and historical similarity of the user behavior. For example, if the most recent behavior has a greater impact on the prediction result, its weight is increased. Conversely, if the earlier behavior has a smaller impact on the prediction result, its weight is reduced, etc., thereby improving the prediction accuracy and obtaining accurate intended behavior information.

[0046] S205: Generate a corresponding customer acquisition marketing strategy based on the intended behavior information, and send marketing information to the customer through a specified channel at a specified trigger time according to the customer acquisition marketing strategy.

[0047] In this embodiment, personalized customer acquisition marketing strategies can be formulated for different customers based on the intention behavior information generated by the behavior prediction model. For example, for customers who have a high purchase intention for a certain product and are regularly active in a certain channel according to the prediction results, customized insurance product recommendations can be sent via email, text messages or mobile applications during the most active time of the customer. The personalized marketing strategy improves the targeting and effectiveness of marketing activities, reduces the time cost of customer acquisition, and improves customer experience and satisfaction.

[0048] In the above embodiment, the present invention discloses a customer acquisition analysis method based on transaction data, which generates a standard data set by collecting the transaction data of the customer group to be analyzed and preprocessing the transaction data; performing feature screening on the standard data set, and performing individual feature extraction and cluster analysis on the screened standard data set to obtain corresponding individual features and group features; constructing user portrait data based on the individual features and group features and the behavioral time series features in the transaction data; inputting the user portrait data into a trained behavior prediction model, and performing behavior prediction processing on the user portrait data according to the time series weighting strategy through the behavior prediction model to generate intention behavior information; generating a corresponding customer acquisition marketing strategy based on the intention behavior information, and sending marketing information to customers through a specified channel at a specified trigger time based on the customer acquisition marketing strategy. A comprehensive user portrait is constructed by extracting individual and group features and behavioral time series features, and the behavior prediction model and time series weighting strategy are used to accurately predict the customer's future behavior, automatically generate personalized customer acquisition strategies, achieve accurate customer acquisition analysis, and improve conversion rate and customer satisfaction.

[0049] In one embodiment, Figure 3 As shown, step S201 includes:

[0050] S301, collecting transaction data of the customer group to be analyzed from a preset database;

[0051] S302, performing data cleaning on the transaction data to filter out noise data in the transaction data;

[0052] S303: Standardize the transaction data from which the noise data has been filtered out, convert field data in different formats into a specified format, and generate a standard data set.

[0053] In this embodiment, the transaction data of the customer group to be analyzed is collected from the preset database. For example, in the financial insurance scenario, the transaction data of the customer can be collected from the customer relationship management system (CRM) database of the insurance company, including the basic information of the customer, policy details, claim records, etc. For example, the following transaction data can be collected: customer ID, age, gender, occupation, annual income, type of insurance product purchased, policy effective date, number and amount of claims, official website visit history, marketing interaction history, etc. After that, the transaction data is cleaned to identify and process the noise data, including missing values, outliers, duplicate records, irrelevant information, etc., to improve the data quality of the data set. After data cleaning, the transaction data with noise data filtered out is further standardized, and field data in different formats are converted to the specified format. For example, the date field may exist in different formats (such as YYYY-MM-DD, DD / MM / YYYY, etc.), so all date fields are uniformly converted to the ISO standard format (YYYY-MM-DD). Similarly, for the gender field, there may be different representation methods (such as M / F, Male / Female, etc.), so the gender field is unified into a standard format (such as 0 / 1 or Male / Female). The format of the data set after standardization is unified, which is convenient for data analysis and improves data readability and processing efficiency.

[0054] In one embodiment, Figure 4 As shown, step S202 includes:

[0055] S401, extracting basic features from the standard data set, and identifying key risk features from the standard data set;

[0056] S402, performing feature screening on the standard data set according to the basic features and key risk features to obtain a key data set;

[0057] S403, extracting individual features from the data of each customer in the key data set to obtain individual features of each customer;

[0058] S404: Perform cluster analysis on the data of all customers in the key data set according to a preset clustering algorithm to obtain the group to which each customer belongs and the corresponding group characteristics.

[0059] In this embodiment, when performing feature screening, the basic features in the standard data set are first extracted. The basic features include basic information of the customer such as age, gender, occupation, annual income, etc., to provide basic customer features for the construction of user portraits. And the key risk features in the standard data set are also identified. The key risk features may involve policy coverage, liability limits, etc. These features can reflect the customer's risk preference. For example, a policy with a high liability limit may indicate that the customer has a high tolerance for risk. The standard data set is feature screened by extracting and identifying the basic features and key risk features, that is, retaining the habitual data of the basic features and key risk features in the standard data set, filtering out other data that is less relevant to customer acquisition analysis, thereby obtaining a key data set to reduce the amount of data for subsequent analysis and improve analysis efficiency.

[0060] The filtered key data set is analyzed in depth, and individual features are extracted based on the data of each customer in the key data set to obtain the individual features of each customer. For example, on the basis of directly extracting the characteristics of the customer's age, gender, occupation, income, etc., it is also possible to extract the individual features of each customer's risk preference, purchasing power, and insurance demand by analyzing the customer's purchase history and claim settlement record. In addition to individual features, this embodiment also performs cluster analysis on the data of all customers in the key data set according to a preset clustering algorithm to obtain the group to which each customer belongs and the corresponding group features. Specifically, spectral clustering can be used in combination with an adaptive neighborhood graph to perform cluster analysis to handle non-spherical data distribution and clustering problems of different densities. Specifically, a spectral clustering algorithm is used to handle non-spherical data distribution by constructing a similarity matrix and a Laplace matrix. Spectral clustering can identify the inherent clustering structure of the data, rather than clustering based solely on the distance. Combined with an adaptive neighborhood graph, the neighborhood size of each sample point can be dynamically adjusted according to the local density of the data, so as to better handle clustering problems of different densities, and the clustering process can be adaptively optimized to improve the recognition ability of complex data structures. Through cluster analysis, customers can be divided into different groups. Different groups may have different purchasing habits and needs, and different groups may have common preferences for certain characteristics of insurance products, such as young high-risk groups, middle-aged stable groups, etc., which helps to understand the common characteristics and needs of each group. By identifying different customer groups and their common characteristics, in customer acquisition analysis, not only individual needs are considered but also group needs are combined to improve the accuracy of personalized marketing strategies.

[0061] In one embodiment, identifying key risk features in the standard data set includes:

[0062] Identifying risk factors in the standard data set, and analyzing and determining logical relationships between the risk factors;

[0063] According to the importance of the risk factors and the logical relationship between the risk factors, the key risk factors with an importance higher than a preset threshold and having dependencies are screened as the key risk features.

[0064] In this embodiment, when identifying key risk features, risk factors in the standard data set are first identified, such as extracting risk factors such as the customer's credit history and claim frequency, and using NLP technology to automatically extract risk factors such as coverage, liability limit, etc. from the policy terms to convert them into numerical features, and analyze and determine the logical management between risk factors, such as using dependency syntax analysis to determine the grammatical structure between risk factor vocabulary, thereby determining the logical relationship between risk factors. The logical relationship can be determined by the type and direction of the dependency path. For example, if the "liability limit" depends on the "coverage", it can be considered that the "coverage" is a prerequisite for the "liability limit". Based on the identified risk factors and the logical relationship between them, the importance of each risk factor is further evaluated. For example, the weight of each risk factor is determined through the analytic hierarchy process (AHP), and then a threshold is set so that factors with a weight greater than 0.5 are considered important risk factors. In addition, key risk factors with dependencies are screened out from the important risk factors, such as high liability limits and high claims frequency. This allows the logical relationship between risk factors to be considered during subsequent feature selection to ensure that the screened data set can reflect these complex relationships, which helps the model understand how different risk factors work together, thereby more accurately predicting events such as insurance claims and improving the model's predictive ability.

[0065] In one embodiment, Figure 5 As shown, step S204 includes:

[0066] S501, inputting the user portrait data into a trained behavior prediction model, wherein the behavior prediction model is constructed and trained based on an LSTM network of an attention mechanism;

[0067] S502: Performing timeliness analysis and historical similarity analysis of user behavior based on the behavior time series characteristics in the user portrait data;

[0068] S503, assigning corresponding weights to the user behaviors at each time point in the behavior time series feature according to the timeliness analysis results and the historical similarity analysis results, and performing weighted processing according to the weights to obtain weighted sequence features;

[0069] S504: Perform behavior prediction based on the weighted sequence features and the individual features and group features in the user portrait data, and output the purchase intention score of each customer under the corresponding prediction factors.

[0070] In this embodiment, a behavior prediction model is constructed in advance based on the LSTM network of the attention mechanism. The model has been trained with historical user portrait data. When acquiring customers, the constructed user portrait data is input into the trained behavior prediction model. The user portrait data includes individual characteristics, group characteristics, and behavior time series characteristics. These data are used as input and passed to the model for behavior prediction under the time series weighting strategy. The LSTM network based on the attention mechanism can capture the long-term dependencies of user behavior and use the attention mechanism to identify the behavior patterns that have the greatest impact on the prediction results, thereby improving the accuracy and reliability of the prediction. When predicting behavior, for the behavior time series characteristics, it includes the user's behavior records over time, such as purchase history, access behavior, interactive behavior, etc. Therefore, when predicting behavior according to the time series weighting strategy, the timeliness analysis and historical similarity analysis of user behavior are first performed on the behavior time series characteristics. Wherein timeliness refers to the fact that recent behavior has a greater impact on future behavior. For example, if a customer recently visited an insurance product page or searched for related keywords, the impact of this behavior on future purchase decisions may be greater than that of a few months ago. Historical similarity refers to the behavior patterns in similar situations in history, which can provide a reference for predicting current behavior. For example, if a customer has purchased insurance products at the end of each financial year in the past, this historical pattern can be used to predict future behavior.

[0071] Based on the results of timeliness analysis and historical similarity analysis, the corresponding weights are assigned to user behaviors at each time point in the behavior time series features. That is, the model will assign weights according to the timeliness of the behavior. For example, user behaviors in the last week may be given a higher weight than those in the previous month. And the model will assign weights according to historical similarity. For example, if the current time point is highly similar to the time point when the customer purchased insurance in history, the weights of these historical time points will increase. Therefore, the weights of each time point in the time series can be dynamically adjusted according to the timeliness and historical similarity of user behavior. Specifically, the weights of different time points can be calculated by the formula, w(t) = α·exp(-λ·(tT) 2)+β·S(t), where w(t) is the weight of time point t, T is the current time point, α and λ are timeliness parameters, S(t) is the historical similarity score, and β is the historical similarity parameter. The behavioral time series features are weighted based on the weights of different time points to obtain weighted sequence features. Through dynamic weighting, the model pays more attention to those time points with higher weights when predicting, improves the responsiveness of the prediction model to recent changes, and makes the prediction results more consistent with the current behavior trends of users. Combining weighted sequence features, individual characteristics (such as age, gender, occupation) and group characteristics (such as group preferences), the purchase intention of each customer under the existing prediction factors is predicted to obtain the purchase intention score. Specific prediction factors may include, for example, purchase time, marketing channel, target product, etc. For example, the model may output a score indicating the possibility of a user purchasing a specific insurance product in the next month. By comprehensively considering the user's historical behavior, individual attributes and group attributes, a more comprehensive and accurate purchase intention prediction is provided. This helps insurance companies to more accurately locate potential customers, thereby conducting precise customer acquisition marketing and improving customer conversion rates.

[0072] In one embodiment, after step S205, the method further includes:

[0073] receiving execution feedback information of the intended behavior information;

[0074] Performing difference analysis based on the intention behavior information and the execution feedback information;

[0075] The behavior prediction model is subjected to multi-objective reinforcement learning according to the difference analysis results and preset marketing goals, and the behavior prediction model is optimized.

[0076] In this embodiment, after personalized customer acquisition marketing is performed based on the intention behavior information, the marketing personnel can upload the execution feedback information of the intention behavior information predicted by the model, or can also automatically obtain the execution feedback information by tracking and recording the behavior of the customer to be analyzed. For example, an insurance company sends a customized insurance product recommendation to the customer by email. The customer's feedback may be opening the email, clicking on the link to learn more, actually purchasing insurance, or no response. These feedback information can be tracked and recorded through the customer management system. By collecting and analyzing these execution feedback information, the actual effect of customer acquisition marketing activities can be understood and the success rate of different marketing strategies can be evaluated. Difference analysis is performed on the intention behavior information and the execution feedback information. For example, the model predicts that customer C has a high purchase intention, but the actual customer C has not purchased any products. By comparing the prediction results and the actual behavior, the deviation of the model prediction can be identified, providing guidance for model optimization. The behavior prediction model is optimized based on the results of the difference analysis, specifically, multi-objective reinforcement learning is performed in combination with several preset marketing goals. The model not only considers improving the prediction accuracy, but also considers other marketing goals, such as customer satisfaction, marketing costs, and success rate. By optimizing the model through multi-objective reinforcement learning, multiple marketing objectives can be integrated into the optimal adjustment strategy.

[0077] In one embodiment, after step S205, the method further includes:

[0078] New training data is acquired according to a preset time period, and parameters of the behavior prediction model are adjusted according to the new training data to adapt to the data distribution in the current time period.

[0079] In this embodiment, in order to adapt to market changes and changes in customer behavior, new training data is obtained according to a preset time period. For example, a quarter can be set as a cycle to obtain the latest customer interaction data from the customer relationship management system, including new policy purchase records, claims, and customer service interactions. These data reflect the latest market trends and customer preferences. The behavior prediction model is adjusted through new training data so that the behavior prediction model can be regularly updated to adapt to the data distribution in the current time period. By regularly introducing new training data, the model can capture the latest data distribution characteristics, reduce the impact of outdated data on model performance, improve its adaptability and prediction accuracy in the current market environment, and improve the stability and reliability of the model.

[0080] It should be noted that there is not necessarily a certain order between the above-mentioned steps. A person skilled in the art can understand, based on the description of the embodiments of the present invention, that in different embodiments, the above-mentioned steps may have different execution orders, that is, they may be executed in parallel, may be executed interchangeably, and so on.

[0081] Further references Figure 6 , as a response to the above Figure 2 The present invention provides an embodiment of a customer acquisition analysis device based on transaction data. Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0082] like Figure 6 As shown, the customer acquisition analysis device 60 based on transaction data described in this embodiment includes:

[0083] The collection module 601 is used to collect the transaction data of the customer group to be analyzed and pre-process the transaction data to generate a standard data set;

[0084] The feature extraction module 602 is used to perform feature screening on the standard data set, and perform individual feature extraction and cluster analysis on the screened standard data set to obtain corresponding individual features and group features;

[0085] A construction module 603 is used to construct user portrait data according to the individual characteristics and group characteristics and the behavior time series characteristics in the transaction data;

[0086] The customer acquisition prediction module 604 is used to input the user portrait data into the trained behavior prediction model, and perform behavior prediction processing on the user portrait data according to the time series weighting strategy through the behavior prediction model to generate intended behavior information;

[0087] The customer acquisition strategy generation module 605 is used to generate a corresponding customer acquisition marketing strategy according to the intention behavior information, and send marketing information to the customer through a specified channel at a specified trigger time according to the customer acquisition marketing strategy.

[0088] The module referred to in the present invention refers to a series of computer program instruction segments that can perform specific functions, which is more suitable for describing the execution process of customer acquisition analysis based on transaction data than a program. For the specific implementation method of each module, please refer to the corresponding method embodiment above, which will not be repeated here.

[0089] In one embodiment, the acquisition module 601 includes:

[0090] A collection unit, used to collect transaction data of the customer group to be analyzed from a preset database;

[0091] A data cleaning unit, used to clean the transaction data and filter out noise data in the transaction data;

[0092] The standardization unit is used to standardize the transaction data after filtering out the noise data, convert the field data in different formats into a specified format, and generate a standard data set.

[0093] In one embodiment, the feature extraction module 602 includes:

[0094] An extraction and identification unit, used to extract basic features in the standard data set and identify key risk features in the standard data set;

[0095] A feature screening unit, configured to perform feature screening on the standard data set according to the basic features and key risk features to obtain a key data set;

[0096] An individual feature extraction unit, used to extract individual features from the data of each customer in the key data set to obtain individual features of each customer;

[0097] The cluster analysis unit is used to perform cluster analysis on the data of all customers in the key data set according to a preset clustering algorithm to obtain the group to which each customer belongs and the corresponding group characteristics.

[0098] In one embodiment, the extraction and identification unit includes:

[0099] A logic analysis unit, used to identify risk factors in the standard data set, and analyze and determine the logical relationship between the risk factors;

[0100] The key risk feature extraction unit is used to screen the key risk factors with an importance higher than a preset threshold and having dependencies as the key risk features according to the importance of the risk factors and the logical relationship between the risk factors.

[0101] In one embodiment, the customer acquisition prediction module 604 includes:

[0102] An input unit, used to input the user portrait data into a trained behavior prediction model, wherein the behavior prediction model is constructed and trained based on an LSTM network with an attention mechanism;

[0103] An analysis unit, used to perform timeliness analysis and historical similarity analysis of user behavior based on the behavior time series characteristics in the user portrait data;

[0104] A weighting unit, used to assign corresponding weights to user behaviors at each time point in the behavior time series feature according to the timeliness analysis result and the historical similarity analysis result, and perform weighted processing according to the weights to obtain a weighted sequence feature;

[0105] The prediction unit is used to perform behavior prediction based on the weighted sequence characteristics and the individual characteristics and group characteristics in the user portrait data, and output the purchase intention score of each customer under the corresponding prediction factors.

[0106] In one embodiment, the device 60 further includes:

[0107] A feedback receiving module, configured to receive execution feedback information of the intended behavior information;

[0108] A difference analysis module, used for performing difference analysis based on the intended behavior information and the execution feedback information;

[0109] The model optimization module is used to perform multi-objective reinforcement learning on the behavior prediction model according to the difference analysis results and preset marketing goals, and optimize the behavior prediction model.

[0110] In one embodiment, the device 60 further includes:

[0111] The continuous training module is used to obtain new training data according to a preset time period, and adjust the parameters of the behavior prediction model through the new training data to adapt to the data distribution in the current time period.

[0112] In the above embodiment, the present invention discloses a customer acquisition analysis device based on transaction data, which generates a standard data set by collecting the transaction data of the customer group to be analyzed and preprocessing the transaction data; performs feature screening on the standard data set, and performs individual feature extraction and cluster analysis on the screened standard data set to obtain corresponding individual features and group features; constructs user portrait data based on the individual features and group features and the behavioral time series features in the transaction data; inputs the user portrait data into a trained behavior prediction model, and performs behavior prediction processing on the user portrait data according to the time series weighting strategy through the behavior prediction model to generate intention behavior information; generates a corresponding customer acquisition marketing strategy based on the intention behavior information, and sends marketing information to customers through a specified channel at a specified trigger time based on the customer acquisition marketing strategy. A comprehensive user portrait is constructed by extracting individual and group features and behavioral time series features, and the behavior prediction model and time series weighting strategy are used to accurately predict the customer's future behavior, automatically generate personalized customer acquisition strategies, achieve accurate customer acquisition analysis, and improve conversion rate and customer satisfaction.

[0113] Another embodiment of the present invention provides a computer device, such as Figure 7 As shown, the computer device 70 includes:

[0114] One or more processors 701 and memory 702, Figure 7 A processor 701 is used as an example for description. The processor 701 and the memory 702 may be connected via a bus or other means. Figure 7 The example of connecting through bus is taken in the following.

[0115] Processor 701 is used to complete various control logics of computer device 70, and it can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), a single-chip microcomputer, an ARM (Acorn RISC Machine) or other programmable logic device, a discrete gate or transistor logic, a discrete hardware component or any combination of these components. In addition, processor 701 can also be any traditional processor, microprocessor or state machine. Processor 701 can also be implemented as a combination of computing devices, for example, a combination of a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors combined with a DSP and / or any other such configuration.

[0116] The memory 702 is a non-volatile computer-readable storage medium that can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as program instructions corresponding to the customer acquisition analysis method based on transaction data in the embodiment of the present invention. The processor 701 executes various functional applications and data processing of the computer device 70 by running the non-volatile software programs, instructions and units stored in the memory 702, that is, implementing the customer acquisition analysis method based on transaction data in the above method embodiment.

[0117] The memory 702 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device 70, etc. In addition, the memory 702 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 702 may optionally include a memory remotely arranged relative to the processor 701, and these remote memories may be connected to the computer device 70 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof. One or more units are stored in the memory 702, and when executed by one or more processors 701, the steps of the customer acquisition analysis method based on transaction data in any of the above-mentioned method embodiments are executed.

[0118] In the above embodiment, the present invention discloses a computer device, which generates a standard data set by collecting transaction data of a customer group to be analyzed and preprocessing the transaction data; performs feature screening on the standard data set, and performs individual feature extraction and cluster analysis on the screened standard data set to obtain corresponding individual features and group features; constructs user portrait data based on the individual features and group features and the behavioral time series features in the transaction data; inputs the user portrait data into a trained behavior prediction model, and performs behavior prediction processing on the user portrait data according to a time series weighting strategy through the behavior prediction model to generate intention behavior information; generates a corresponding customer acquisition marketing strategy based on the intention behavior information, and sends marketing information to customers through a specified channel at a specified trigger time based on the customer acquisition marketing strategy. A comprehensive user portrait is constructed by extracting individual and group features and behavioral time series features, and the behavior prediction model and time series weighting strategy are used to accurately predict the customer's future behavior, automatically generate personalized customer acquisition strategies, achieve accurate customer acquisition analysis, and improve conversion rates and customer satisfaction.

[0119] An embodiment of the present invention provides a non-volatile computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are executed by one or more processors, the steps of the customer acquisition analysis method based on transaction data in any of the above method embodiments are executed.

[0120] In the above embodiment, the present invention discloses a non-volatile computer-readable storage medium, which generates a standard data set by collecting transaction data of a customer group to be analyzed and preprocessing the transaction data; performs feature screening on the standard data set, and performs individual feature extraction and cluster analysis on the screened standard data set to obtain corresponding individual features and group features; constructs user portrait data based on the individual features and group features and the behavioral time series features in the transaction data; inputs the user portrait data into a trained behavior prediction model, and performs behavior prediction processing on the user portrait data according to a time series weighting strategy through the behavior prediction model to generate intention behavior information; generates a corresponding customer acquisition marketing strategy based on the intention behavior information, and sends marketing information to customers through a specified channel at a specified trigger time based on the customer acquisition marketing strategy. A comprehensive user portrait is constructed by extracting individual and group features and behavioral time series features, and the behavior prediction model and time series weighting strategy are used to accurately predict the customer's future behavior, automatically generate personalized customer acquisition strategies, achieve accurate customer acquisition analysis, and improve conversion rates and customer satisfaction.

[0121] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0122] The present invention can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present invention can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present invention can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0123] In summary, the customer acquisition analysis method, device, equipment and medium based on transaction data disclosed in the present invention include: collecting the transaction data of the customer group to be analyzed and preprocessing the transaction data to generate a standard data set; performing feature screening on the standard data set, and performing individual feature extraction and cluster analysis on the screened standard data set to obtain corresponding individual features and group features; constructing user portrait data based on the individual features and group features and the behavioral time series features in the transaction data; inputting the user portrait data into a trained behavior prediction model, and performing behavior prediction processing on the user portrait data according to the time series weighting strategy through the behavior prediction model to generate intention behavior information; generating a corresponding customer acquisition marketing strategy based on the intention behavior information, and sending marketing information to customers through a specified channel at a specified trigger time according to the customer acquisition marketing strategy. A comprehensive user portrait is constructed by extracting individual and group features and behavioral time series features, and the behavior prediction model and time series weighting strategy are used to accurately predict the customer's future behavior, automatically generate personalized customer acquisition strategies, achieve accurate customer acquisition analysis, and improve conversion rate and customer satisfaction.

[0124] Of course, those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing related hardware (such as a processor, a controller, etc.) through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium, and the computer program can include the processes of the above-mentioned method embodiments when executed. The storage medium can be a memory, a disk, a floppy disk, a flash memory, an optical storage device, etc.

[0125] It should be noted that if any software tools or components other than those of the Company appear in the embodiments of the present application, they are merely used for illustration and do not represent actual use. It should be understood that the application of the present invention is not limited to the above examples, and that those skilled in the art can make improvements or changes based on the above description, and all such improvements and changes shall fall within the scope of protection of the claims attached to the present invention.

Claims

1. A customer acquisition analysis method based on transaction data, characterized in that: include: Collecting transaction data of the customer group to be analyzed and preprocessing the transaction data to generate a standard data set; Performing feature screening on the standard data set, and performing individual feature extraction and cluster analysis on the screened standard data set to obtain corresponding individual features and group features; Constructing user portrait data based on the individual characteristics and group characteristics and the behavioral time series characteristics in the transaction data; Inputting the user portrait data into a trained behavior prediction model, performing behavior prediction processing on the user portrait data according to a time series weighted strategy through the behavior prediction model to generate intended behavior information; A corresponding customer acquisition marketing strategy is generated according to the intended behavior information, and marketing information is sent to the customer through a specified channel at a specified trigger time according to the customer acquisition marketing strategy.

2. The customer acquisition analysis method based on transaction data according to claim 1, characterized in that: The collecting of transaction data of the customer group to be analyzed and preprocessing of the transaction data to generate a standard data set includes: Collect transaction data of the customer group to be analyzed from the preset database; Performing data cleaning on the transaction data to filter out noise data in the transaction data; The transaction data from which the noise data has been filtered out is standardized, and field data in different formats are converted into a specified format to generate a standard data set.

3. The customer acquisition analysis method based on transaction data according to claim 1, characterized in that: The feature screening of the standard data set, and the individual feature extraction and cluster analysis of the screened standard data set to obtain corresponding individual features and group features include: Extracting basic features from the standard data set and identifying key risk features from the standard data set; Performing feature screening on the standard data set according to the basic features and key risk features to obtain a key data set; Extracting individual features from the data of each customer in the key data set to obtain individual features of each customer; A cluster analysis is performed on the data of all customers in the key data set according to a preset clustering algorithm to obtain the group to which each customer belongs and the corresponding group characteristics.

4. The customer acquisition analysis method based on transaction data according to claim 3 is characterized in that: The identifying key risk features in the standard data set includes: Identifying risk factors in the standard data set, and analyzing and determining logical relationships between the risk factors; According to the importance of the risk factors and the logical relationship between the risk factors, the key risk factors with an importance higher than a preset threshold and having dependencies are screened as the key risk features.

5. The customer acquisition analysis method based on transaction data according to claim 1, characterized in that: The step of inputting the user portrait data into a trained behavior prediction model, performing behavior prediction processing on the user portrait data according to a time series weighting strategy through the behavior prediction model, and generating intended behavior information includes: Inputting the user portrait data into a trained behavior prediction model, wherein the behavior prediction model is constructed and trained based on an LSTM network with an attention mechanism; Performing timeliness analysis and historical similarity analysis of user behavior based on the behavior time series characteristics in the user portrait data; According to the timeliness analysis results and the historical similarity analysis results, corresponding weights are assigned to the user behaviors at each time point in the behavior time series feature, and weighted processing is performed according to the weights to obtain weighted sequence features; Behavior prediction is performed based on the weighted sequence features and the individual features and group features in the user portrait data, and the purchase intention score of each customer under the corresponding prediction factors is output.

6. The customer acquisition analysis method based on transaction data according to claim 1, characterized in that: After inputting the user portrait data into the trained behavior prediction model, performing behavior prediction processing on the user portrait data according to the time series weighting strategy through the behavior prediction model to generate the intended behavior information, the method further includes: receiving execution feedback information of the intended behavior information; Performing difference analysis based on the intention behavior information and the execution feedback information; The behavior prediction model is subjected to multi-objective reinforcement learning according to the difference analysis results and preset marketing goals, and the behavior prediction model is optimized.

7. The customer acquisition analysis method based on transaction data according to claim 1, characterized in that: After inputting the user portrait data into the trained behavior prediction model, performing behavior prediction processing on the user portrait data according to the time series weighting strategy through the behavior prediction model to generate the intended behavior information, the method further includes: New training data is acquired according to a preset time period, and parameters of the behavior prediction model are adjusted according to the new training data to adapt to the data distribution in the current time period.

8. A customer acquisition analysis device based on transaction data, characterized in that: include: A collection module, used to collect transaction data of the customer group to be analyzed and pre-process the transaction data to generate a standard data set; A feature extraction module is used to perform feature screening on the standard data set, and perform individual feature extraction and cluster analysis on the screened standard data set to obtain corresponding individual features and group features; A construction module, used to construct user portrait data according to the individual characteristics and group characteristics and the behavioral time series characteristics in the transaction data; A customer acquisition prediction module, which is used to input the user portrait data into a trained behavior prediction model, and perform behavior prediction processing on the user portrait data according to a time series weighted strategy through the behavior prediction model to generate intended behavior information; The customer acquisition strategy generation module is used to generate a corresponding customer acquisition marketing strategy based on the intended behavior information, and send marketing information to the customer through a specified channel at a specified trigger time according to the customer acquisition marketing strategy.

9. A computer device, characterized in that: comprising at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the customer acquisition analysis method based on transaction data as described in any one of claims 1-7.

10. A non-volatile computer-readable storage medium, characterized in that: The non-volatile computer-readable storage medium stores computer-executable instructions, which, when executed by one or more processors, enable the one or more processors to execute the customer acquisition analysis method based on transaction data as described in any one of claims 1-7.

Images