Automotive industry big data customer acquisition methods, devices, electronic equipment and storage media

Abstract

This application provides a method, device, electronic device, and storage medium for acquiring customers through big data in the automotive industry. The electronic device analyzes the differences in access information of various types of applications by users who own vehicles and those who do not, filters out target applications related to the user's vehicle ownership status and access information related to each target application, and generates a current car purchase intention model based on this. The electronic device also filters the original car purchase behavior feature data of vehicle owners and vehicle owners to obtain effective behavior feature data, and combines it with car purchase intention to train a historical car purchase intention model. The electronic device uses the two models of historical car purchase intention and current car purchase intention to jointly judge the car purchase operation of the target user, increasing the influencing factors considered when identifying whether a user has performed a car purchase behavior, improving the recognition accuracy, and thus improving the accuracy of target user profile generation and the precision of marketing activities based on user profiles.

Description

Technical Field

[0001] The embodiments of this application relate to the field of big data application technology, and in particular to a method, apparatus, electronic device and storage medium for acquiring customers through big data in the automotive industry. Background Technology

[0002] User profiles are models that depict users' social attributes, lifestyle habits, and consumption behaviors. Car purchase and ownership are important factors in describing user consumption behavior and are a crucial component of user profiles. The automotive service industry can develop corresponding marketing strategies based on user consumption behaviors identified within these user profiles.

[0003] In existing technologies, servers train neural networks by using a large amount of user browsing, search behavior, and related comments about automotive products as labels for users' car purchase intentions. These neural networks are then used to determine the purchase intentions of target users, creating a user profile so that the server can implement corresponding marketing strategies. However, the labels used in the neural network training process are based on experience and fail to fully reflect the factors users consider when making purchase decisions. This results in inaccuracies in the neural network and biases in the generated user profiles. Summary of the Invention

[0004] This application provides a method, device, electronic device, and storage medium for acquiring customers through big data in the automotive industry, in order to solve the technical problem that neural networks trained based on experience-defined labels do not accurately judge the car purchase intention of target users.

[0005] Firstly, this application provides a big data-driven customer acquisition method for the automotive industry, the method comprising:

[0006] The system acquires access information and original car purchase behavior data of a first user who owns a vehicle and a second user who does not own a vehicle to various types of applications; the access information includes access time range and operation information, and the original car purchase behavior data includes various offline behavior characteristics and various online behavior characteristics.

[0007] For each type of application, a difference analysis is performed on the operation information of the first user and the second user within the same access time range, generating analysis results. Based on the analysis results, a rule model is constructed to generate a current car purchase intention model. The current car purchase intention model includes a first current car purchase intention model and a second current car purchase intention model, and the analysis results include the target application.

[0008] The offline and online behavioral feature data are cleaned and preprocessed to determine the valid behavioral feature data.

[0009] The effective behavioral feature data is used to train a logistic regression model to obtain a historical car purchase intention model;

[0010] The system acquires access information and effective behavioral feature data of the target user to the various types of applications. Based on the current car purchase intention model, the access information of the target application, the historical car purchase intention model, and the effective behavioral feature data, it generates a user profile of the target user and makes corresponding information recommendations based on the user profile.

[0011] In the above technical solution, the electronic device analyzes the differences in access information of various types of applications by users who own vehicles and those who do not, filters out target applications related to the user's vehicle ownership status and access information related to each target application, and generates a current car purchase intention model based on this. The electronic device also filters the original car purchase behavior feature data of vehicle owners and vehicle owners to obtain effective behavior feature data, and combines it with the car purchase intention to train a historical car purchase intention model. The electronic device uses the historical car purchase intention model and the current car purchase intention model to jointly judge the car purchase operation of the target user, which increases the influencing factors considered when identifying whether the user has performed a car purchase behavior, improves the recognition accuracy, and thus improves the accuracy of target user profile generation and the precision of marketing activities based on user profiles.

[0012] Optionally, for various applications, a difference analysis is performed on the operation information of the first user and the second user within the same access time range, generating analysis results, specifically including:

[0013] For various applications, the operation information of the first user and the second user within the same access time range is used as a quantitative variable, and the user's vehicle ownership status is used as a qualitative variable. An independent samples t-test is performed to determine the analysis results. The analysis results include the target applications related to the differences in various applications and the target operation information corresponding to each target application.

[0014] Optionally, each application includes multiple original applications; the operation information of the first user and the second user within the same access time range is used as a quantitative variable, and the user's ownership status of the vehicle is used as a categorical variable. An independent samples t-test is performed to determine the analysis results, specifically including:

[0015] For each of the original applications, count the number of times the first user and the second user perform the operation information within the same access time range;

[0016] Valid operation information is selected according to a preset number of times, and this valid operation information is determined as a quantitative variable; the valid operation information is operation information whose number of operations is within the preset number of times.

[0017] The user's ownership status of the vehicle is used as a categorical variable. An independent samples t-test is performed on the quantitative and categorical variables to calculate the significance index.

[0018] When the significance index is less than the preset significance threshold, the original application is identified as the target application, and the minimum value of the preset number range is identified as the effective behavior threshold corresponding to the target operation information.

[0019] Optionally, a rule model is constructed based on the analysis results to generate a current car purchase intention model, specifically including:

[0020] A vehicle holding status recognition model is generated based on the target operation information of the target application and its corresponding effective behavior threshold.

[0021] Based on the target operation information of all target applications within a first preset time range and a first preset threshold, a first current car purchase constraint is constructed; the first current car purchase constraint is that the total amount of all target operation information generated by the user is greater than the first preset threshold.

[0022] Based on the target operation information of all target applications within a first preset time range and a second preset threshold, a second current car purchase constraint is constructed; the second current car purchase constraint is that the total amount of all target operation information generated by the user is greater than the product of the average value of all target operation information generated by the user within the second preset time range and a preset multiple.

[0023] Based on the first current vehicle purchase constraint and the vehicle ownership status recognition model, a first vehicle purchase intention model is generated;

[0024] A second vehicle purchase intention model is generated based on the second current vehicle purchase constraint and the vehicle ownership status recognition model.

[0025] Optionally, data cleaning and preprocessing operations are performed on the offline and online behavioral feature data to determine valid behavioral feature data, specifically including:

[0026] Pearson correlation analysis was performed on the offline and online behavioral feature data, and redundant feature data was removed from the original car purchase behavior feature data based on the analysis results to obtain the first car purchase behavior data.

[0027] The first car purchase behavior data is normalized to generate valid behavioral feature data.

[0028] Optionally, a user profile of the target user is generated based on the current car purchase intention model, the access information of the target application, the historical car purchase intention model, and the effective behavioral feature data, specifically including:

[0029] The current car purchase intention model is used to process the access information of the target application to determine whether the target user has a current car purchase intention;

[0030] The effective behavioral feature data is processed using the historical car purchase intention model to determine whether the target user has a history of car purchase intention;

[0031] Based on the current purchase intention and the historical purchase intention, determine the implementation status of the target user's purchase behavior;

[0032] Based on the status of the car purchase behavior, a user profile of the target user is generated.

[0033] Optionally, the purchase behavior status of the target user is determined based on the current purchase intention and the historical purchase intention, specifically including:

[0034] When the target user has both the current intention to purchase a car and the historical intention to purchase a car, it is determined that the target user has engaged in car purchase behavior.

[0035] Optionally, the current car purchase intention model is used to process the access information of the target application to determine whether the target user has a current car purchase intention, specifically including:

[0036] The access information of the target application is processed using a vehicle ownership status recognition model to determine whether the target user has purchased a vehicle.

[0037] When the target user has not purchased a car and meets the first current car purchase constraint, it is determined that the target user has a current car purchase intention;

[0038] When the target user has already purchased a vehicle and meets the second current vehicle purchase constraint, it is determined that the target user has a current vehicle purchase intention.

[0039] Optionally, the access information of the target application is processed using a vehicle ownership status recognition model to determine whether the target user has purchased a vehicle, specifically including:

[0040] Based on the access information of the target applications, count the number of corresponding target operations performed by the target users on each target application;

[0041] The number of target operations performed on each target application is compared with the effective behavior threshold corresponding to the target operation information. Valid applications and their access volumes are then selected from each target application. Valid applications are those whose number of target operations exceeds the corresponding effective behavior threshold.

[0042] Based on the target number of operations for each effective application, calculate the total number of visits for each type of application.

[0043] When the total number of visits to all application classes exceeds the effective access threshold, it is determined that the target user has purchased a car.

[0044] In the above technical solution, the electronic device uses an independent samples t-test to analyze the access information of various application types of the first user with a car and the second user without a car, to screen out the differential variables related to vehicle ownership status and the effective behavioral thresholds to ensure the differential. Combining the behavioral differences between car-owning and carless users before and after purchasing a car, a current car purchase intention model is generated. The electronic device also uses Pearson correlation analysis to screen out effective behavioral feature data based on the correlation between the original car purchase behavior data of the first and second users and their car purchase intention, ensuring the independence between the feature data and reducing the complexity of the logistic regression model generated based on this. The electronic device identifies the car purchase behavior of the target user through two models trained with diverse and non-repeating data, improving the accuracy of the identification results and thus improving the accuracy of the corresponding user profile.

[0045] Secondly, this application provides a big data customer acquisition device for the automotive industry, comprising:

[0046] The acquisition module is used to acquire access information and original car purchase behavior characteristic data of a first user who owns a vehicle and a second user who does not own a vehicle to multiple types of applications; the access information includes access time range and operation information, and the original car purchase behavior characteristic data includes multiple offline behavior characteristic data and multiple online behavior characteristic data.

[0047] The processing module is used to perform differential analysis on the operation information of the first user and the second user within the same access time range for various types of applications, generate analysis results, and construct a rule model based on the analysis results to generate a current car purchase intention model; the current car purchase intention model includes a first current car purchase intention model and a second current car purchase intention model, and the analysis results include the target application;

[0048] The processing module is also used to perform data cleaning and preprocessing operations on the offline behavioral feature data and online behavioral feature data to determine valid behavioral feature data.

[0049] The processing module is also used to train a logistic regression model using the effective behavioral feature data to obtain a historical car purchase intention model;

[0050] The processing module is also used to obtain the target user's access information and effective behavioral feature data of the multiple types of applications, generate a user profile of the target user based on the current car purchase intention model, the access information of the target application, the historical car purchase intention model and the effective behavioral feature data, and take corresponding information recommendations based on the user profile.

[0051] Thirdly, this application provides an electronic device, including: a processor and a memory communicatively connected to the processor;

[0052] The memory stores computer instructions;

[0053] The processor executes computer instructions to implement the big data customer acquisition methods for the automotive industry involved in the first aspect.

[0054] Fourthly, this application provides a computer-readable storage medium storing computer instructions, which, when executed by a processor, are used to implement the automotive industry big data customer acquisition method involved in the first aspect.

[0055] This application provides a method, device, electronic device, and storage medium for acquiring customers through big data in the automotive industry. The electronic device analyzes the differences in access information of various types of applications by users who own vehicles and those who do not, filters out target applications related to the user's vehicle ownership status and access information related to each target application, and generates a current car purchase intention model based on this. The electronic device also filters the original car purchase behavior feature data of vehicle owners and vehicle owners to obtain effective behavior feature data, and combines it with car purchase intention to train a historical car purchase intention model. The electronic device uses the two models of historical car purchase intention and current car purchase intention to jointly judge the car purchase operation of the target user, increasing the influencing factors considered when identifying whether a user has performed a car purchase behavior, improving the recognition accuracy, and thus improving the accuracy of target user profile generation and the precision of marketing activities based on user profiles. Attached Figure Description

[0056] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0057] Figure 1 This is a flowchart illustrating a customer acquisition method for the automotive industry using big data, provided according to an exemplary embodiment of this application.

[0058] Figure 2This is a flowchart illustrating a customer acquisition method for the automotive industry using big data, provided in another exemplary embodiment of this application.

[0059] Figure 3 This is a schematic diagram of the structure of a big data customer acquisition device for the automotive industry provided according to an embodiment of this application;

[0060] Figure 4 This is a schematic diagram of the structure of an electronic device provided according to an embodiment of this application.

[0061] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0062] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0063] User profiles are models that depict users' social attributes, lifestyle habits, and consumption behaviors. Car purchase and ownership are important factors in describing user consumption behavior and are a crucial component of user profiles. The automotive service industry can develop corresponding marketing strategies based on user consumption behaviors identified within these user profiles.

[0064] In existing technologies, servers train neural networks by using a large amount of user browsing, search behavior, and related comments about automotive products as labels for users' car purchase intentions. These neural networks are then used to determine the purchase intentions of target users, creating a user profile so that the server can implement corresponding marketing strategies. However, the labels used in the neural network training process are based on experience and fail to fully reflect the factors users consider when making purchase decisions. This results in inaccuracies in the neural network and biases in the generated user profiles.

[0065] To address the aforementioned technical problems, this application provides a method, device, electronic device, and storage medium for acquiring customers through big data in the automotive industry. It aims to solve the technical problem of inaccurate judgments of target users' car purchase intentions by neural networks trained with experience-based labels. The technical concept of this application is as follows: The electronic device analyzes the differences in access information of various types of applications by users who own vehicles and those who do not, filtering out target applications related to the user's vehicle ownership status and access information related to each target application. Based on this, a current car purchase intention model is generated. The electronic device also filters the original car purchase behavior feature data of vehicle owners and vehicle-owning users to obtain effective behavioral feature data. This data is then combined with the car purchase intention model to obtain a historical car purchase intention model. The electronic device uses both the historical and current car purchase intention models to jointly judge the target user's car purchase action, increasing the influencing factors considered when identifying whether a user has performed a car purchase action, improving the recognition accuracy, and consequently improving the accuracy of target user profile generation and the precision of marketing activities based on user profiles.

[0066] The automotive industry big data customer acquisition method provided in this application can be applied to scenarios including electronic devices and databases. The electronic device retrieves access information and raw behavioral characteristic data of a first user and a second user (who owns a vehicle) from the database for various application types. For each application type, the electronic device performs differential analysis on the operational information of the first and second users within the same access time range, generates analysis results, and constructs a rule model based on these results to generate a current car purchase intention model. The electronic device also performs data cleaning and preprocessing on offline and online behavioral characteristic data from the raw car purchase behavior characteristic data to determine effective behavioral characteristic data. This effective behavioral characteristic data is then used to train a logistic regression model to obtain a historical car purchase intention model.

[0067] When determining whether a target user has a car purchase intention, the electronic device obtains access information and effective behavioral feature data of various types of applications from its internally stored data or its input unit, based on the data required by the current car purchase intention model and the historical car purchase intention model. It then processes the access information of various types of applications using the current car purchase intention model and processes the effective behavioral feature data using the historical car purchase intention model. Based on the processing results of the two models, it determines whether the target user has a car purchase intention and generates a corresponding user profile, so that the electronic device can make corresponding information recommendations based on the user profile.

[0068] Based on the above application scenarios and in conjunction with the accompanying drawings, some embodiments of this application will be described in detail below. Where there is no conflict between the embodiments, the following embodiments and features can be combined with each other. Furthermore, the timing of the steps in the following method embodiments is merely an example and not a strict limitation.

[0069] Figure 1 This is a flowchart illustrating a customer acquisition method for the automotive industry using big data, provided according to an exemplary embodiment of this application. Figure 1 As shown, customer acquisition methods using big data in the automotive industry include:

[0070] S101. The electronic device acquires access information and original vehicle purchase behavior characteristics data of a first user who owns a vehicle and a second user who does not own a vehicle to multiple types of applications.

[0071] Multiple application types refer to various applications related to vehicles. Examples include: car owner service applications, refueling service applications, car maintenance applications, traffic violation inquiry applications, and car information applications.

[0072] Access information for various application types represents information about a user's access to different applications and their actions during that access. This access information includes the access time range and the actions performed.

[0073] The original car purchase behavior data represents some essential behavioral characteristics taken by car owners before and after purchasing a car. This original car purchase behavior data includes various offline behavioral characteristics and various online behavioral characteristics. Among them, offline behavior refers to the actual offline operations conducted by visiting 4S stores and interacting with sales staff or vehicle-related personnel, while online behavior refers to vehicle-related behaviors performed by users through applications or websites.

[0074] When electronic devices obtain relevant information from the first user and the second user, the number of first users and the number of second users obtained are the same to prevent uneven data distribution from affecting subsequent processing and analysis results. However, the number of first users obtaining access information from multiple application types may differ from the number of first users obtaining the original car purchase behavior characteristic data.

[0075] S102. For each type of application, the electronic device performs a differential analysis on the operation information of the first user and the second user within the same access time range, generates analysis results, and constructs a rule model based on the analysis results to generate a current car purchase intention model.

[0076] Difference analysis is a commonly used data analysis method used to detect whether there are differences between two sets of data and whether the differences are significant.

[0077] For each type of application, the electronic device performs a difference analysis on the number of target operations performed by the first user and the second user in turn to determine whether there are differences in the performance of target operations by users with cars and users without cars. This determines whether there is a necessary correlation between performing target operations on this type of application and the user's ownership of a vehicle, and obtains the analysis results.

[0078] Based on the above analysis results, the electronic device identifies user-related information that is necessary to link to vehicle ownership, including but not limited to: application type, related target applications, and operation information for each target application. Each type of application contains at least one target application.

[0079] Based on the user relevance information in the analysis results, the electronic device constructs a rule model and, combined with the corresponding current car purchase constraints, generates a current car purchase intention model, which includes a first current car purchase intention model and a second current car purchase intention model.

[0080] More specifically, the electronic device also determines the current car purchase intention conditions of carless users based on the significant operation information before and after the time point when the carless user first makes a car purchase, and generates a first current car purchase intention model by combining the aforementioned user relevance information of carless users.

[0081] The electronic device also determines the current car purchase intention conditions of car owners based on the significant operation information before and after the car purchase time point, and generates a second current car purchase intention model by combining the above-mentioned user relevance information of car owners.

[0082] S103. Electronic devices perform data cleaning and preprocessing operations on offline and online behavioral feature data to determine valid behavioral feature data.

[0083] Data cleaning refers to the process by which electronic devices correct inaccurate information in offline and online behavioral characteristic data.

[0084] Preprocessing refers to the process by which electronic devices transform cleaned data into data that is helpful for subsequent analysis.

[0085] Effective behavioral feature data represents feature data composed of effective behaviors that are related to a user's car purchase intention.

[0086] The electronic device analyzes the correlation of data after the above cleaning and preprocessing operations to identify information related to the user's car purchase intention, and defines this information as valid behavioral feature data. Specifically, if the user's car purchase intention is to perform a purchase action after a preset time range, then the user has a car purchase intention within that preset time range.

[0087] S104. Electronic devices use effective behavioral feature data to train a logistic regression model to obtain a historical car purchase intention model.

[0088] Logistic regression is a type of classification model.

[0089] The historical car purchase intention model trained using effective behavioral feature data is a model that describes the relationship between effective behavioral feature data and car purchase intention.

[0090] S105. The electronic device acquires the target user's access information and effective behavioral characteristic data for multiple types of applications, generates a user profile of the target user based on the current car purchase intention model, the access information of the target application, the historical car purchase intention model, and the effective behavioral characteristic data, and takes corresponding information recommendations based on the user profile.

[0091] When creating a user profile for a target user, the electronic device acquires the user's access information to various applications required for the current car purchase intention model, as well as the effective behavioral feature data required for the historical car purchase intention model. It then processes this data using the aforementioned models to determine whether the target user has engaged in car purchase behavior within a preset timeframe, and generates a user profile based on the implementation status of this purchase behavior. This user profile also includes whether the target user is a first-time car buyer.

[0092] In the above technical solution, the electronic device analyzes the differences in access information of various types of applications by users who own vehicles and those who do not, filters out target applications related to the user's vehicle ownership status and access information related to each target application, and generates a current car purchase intention model based on this. The electronic device also filters the original car purchase behavior feature data of vehicle owners and vehicle owners to obtain effective behavior feature data, and combines it with the car purchase intention to train a historical car purchase intention model. The electronic device uses the two models of historical car purchase intention and current car purchase intention to jointly judge the car purchase operation of the target user, which increases the influencing factors considered when identifying whether the user has performed a car purchase behavior, improves the recognition accuracy, and thus improves the accuracy of target user profile generation and the precision of marketing activities based on user profiles.

[0093] Figure 2 This is a flowchart illustrating a customer acquisition method for the automotive industry using big data, provided in another exemplary embodiment of this application. Figure 2 As shown, customer acquisition methods using big data in the automotive industry include:

[0094] S201. Electronic devices acquire access information and original vehicle purchase behavior characteristics data of a first user who owns a vehicle and a second user who does not own a vehicle to multiple types of applications.

[0095] This step has been explained in detail in step S101, and will not be repeated here.

[0096] S202. For electronic devices, for various applications, the operation information of the first user and the second user within the same access time range is used as a quantitative variable, and the user's ownership status of the vehicle is used as a categorical variable. An independent samples t-test is conducted to determine the analysis results.

[0097] Each application category contains multiple original applications, and the analysis results include target applications related to differences within each application category, as well as target operation information corresponding to each target application.

[0098] More specifically, when determining the analysis results, the electronic device, for each original application, counts the number of times the first user and the second user perform operations corresponding to the operation information within the same access time range. Valid operation information is selected according to a preset frequency range and defined as a quantitative variable. The user's vehicle ownership status is used as a qualitative variable. An independent samples t-test is performed on the quantitative and qualitative variables to calculate a significance index. When the significance index is less than a preset significance threshold, the original application is determined as the target application, and the minimum value of the preset frequency range is determined as the effective behavior threshold corresponding to the target operation information. Valid operation information refers to operation information whose frequency falls within the preset frequency range.

[0099] Taking a refueling service application as an example, the calculation process of the significance index is explained as follows: If the current month is July, the electronic device counts the number of times the first user (who owns a car) and the second user (who does not own a car) accessed the refueling service application within the time range of January to June, and counts the number of people who made the same number of accesses. The electronic device sets three preset access ranges: greater than 1 time, greater than 10 times, and greater than 20 times. The electronic device uses the number of people obtained from the above statistics as a quantitative variable and the user's vehicle ownership status as a qualitative variable. Then, an independent samples t-test analysis is performed within the above three preset access ranges to obtain a significant P-value. When the significant P-value obtained within the preset access range of greater than 1 time is greater than 0.05, there is no difference in access to the refueling service application. However, when the significant P-values ​​obtained within the preset access ranges of greater than 10 times and greater than 20 times are both less than 0.05, then there is a difference in access to the refueling service application within the preset access range of greater than 10 times. Therefore, the refueling service application is identified as the target application, and its corresponding effective behavior threshold is 10 times.

[0100] S203. Electronic devices construct rule models based on the analysis results and generate current car purchase intention models.

[0101] When generating a current car purchase intention model, electronic devices generate a vehicle ownership status recognition model based on the target operation information of the target application and its corresponding effective behavior threshold.

[0102] More specifically, the vehicle state recognition model includes:

[0103]

[0104] Among them, S k This represents the total number of target operations performed by users on the k-th type of application, where n represents the total number of application types, and U represents the vehicle ownership value. This represents the number of times a user performs a corresponding target operation on the i-th target application within the k-th application category, where m represents the total number of target applications within the k-th application category. In one embodiment, The data corresponding to the variable is valid when it exceeds the corresponding valid behavior threshold.

[0105] More specifically, when U is greater than 0, it means that the user owns a vehicle; when U is equal to 0, it means that the user does not own a vehicle.

[0106] The electronic device also constructs a first current car purchase constraint based on the target operation information of all target applications within a first preset time range and a first preset threshold. This first current car purchase constraint is that the total amount of all target operation information generated by the user exceeds the first preset threshold. In one embodiment, the first preset time range is the current month, and the first current car purchase constraint is that the total number of times the user performs target operations on all target applications within the current month exceeds the first preset threshold.

[0107] The electronic device constructs a second current car purchase constraint based on the target operation information of all target applications within a first preset time range and a second preset threshold. The second current car purchase constraint is that the total amount of all target operation information generated by the user is greater than the product of the average of all target operation information generated by the user within the second preset time range and a preset multiple. In one embodiment, the first preset time range is the current month, and the second preset event range is the six months closest to the current month. Therefore, the second current car purchase constraint is that the total number of target operations performed by the user in the current month is greater than the product of the average number of target operations performed by the user in the previous six months and a preset multiple. The average number of target operations performed by the user in the previous six months is the quotient of the total number of target operations performed by the user in the previous six months divided by six.

[0108] The electronic device generates a first purchase intention model based on the first current purchase constraints and the vehicle ownership status recognition model, and generates a second purchase intention model based on the second current purchase constraints and the vehicle ownership status recognition model.

[0109] S204. The electronic device performs Pearson correlation analysis on offline and online behavioral characteristic data, and removes redundant characteristic data from the original car purchase behavior characteristic data based on the analysis results to obtain the first car purchase behavior data.

[0110] Before performing Pearson correlation analysis on offline and online behavioral characteristic data, electronic devices need to perform data cleaning operations, including but not limited to: removing duplicate, abnormal, and invalid data, correcting erroneous data, filling in missing values ​​as appropriate, and converting categorical variables into numerical variables. In one embodiment, missing values ​​are all filled with 0.

[0111] After the above data cleaning operation is completed, the electronic device uses Pearson correlation analysis to identify at least two features with strong correlation. The electronic device retains only one feature from the above at least two features and uses all the behavioral data obtained in the final screening as the first car purchase behavior data.

[0112] In one embodiment, when the electronic device selects one feature from the above at least two features to retain, it obtains the priority of each feature and selects the feature with the highest priority to retain.

[0113] S205. The electronic device normalizes the first car purchase behavior data to generate valid behavioral feature data.

[0114] Electronic devices use a normalization model to normalize the first car purchase behavior data.

[0115] More specifically, the electronic device uses a normalization model to normalize data belonging to the same variable in the first car purchase behavior data. The normalization model is as follows:

[0116]

[0117] Where X represents any data point corresponding to the target variable in the first car purchase behavior data, X min X represents the minimum value of the target variable in the first car purchase behavior data. max Y represents the maximum value of the target variable in the first car purchase behavior data, and Y represents the effective behavioral feature data corresponding to the target variable in the normalized first car purchase behavior data.

[0118] Electronic devices use the label information of whether or not a user has the intention to purchase a car as valid behavioral feature data to generate a sample set, and then split the sample set into a training sample set and a test sample set according to a preset ratio.

[0119] S206. Electronic devices use effective behavioral feature data to train a logistic regression model to obtain a historical car purchase intention model.

[0120] Specifically, the logistic regression model is represented as follows:

[0121]

[0122] in,

[0123] θ is the feature weight vector, x is the feature vector composed of effective behavioral feature data, and h θ (x) represents the logistic regression model with respect to x, g(θ) T x) represents about θ T Logistic regression model for x.

[0124] The electronic device trains the logistic regression model using the sample set obtained in step S205, and evaluates the trained model using the test sample set. More specifically, the electronic device evaluates the training performance of the model using precision and recall.

[0125] When the accuracy and recall of the trained model are both within the corresponding preset evaluation range, the model is determined as the historical car purchase intention model.

[0126] S207. The electronic device uses the current car purchase intention model to process the access information of the target application and determine whether the target user has the current car purchase intention.

[0127] More specifically, the vehicle ownership status recognition model is used to process the access information of the target application to determine whether the target user has purchased a vehicle. This vehicle ownership status recognition model is trained in step S203.

[0128] When the target user has not purchased a car and meets the first current car purchase constraint, it is determined that the target user has the current car purchase intention; when the target user has already purchased a car and meets the second current car purchase constraint, it is determined that the target user has the current car purchase intention.

[0129] More specifically, when determining whether a target user has purchased a car, the electronic device counts the number of corresponding target operations performed by the target user on each target application based on the access information of the target application.

[0130] The number of target operations for each target application will be compared with the effective behavior threshold corresponding to the target operation information. Effective applications and their access volume will be selected from each target application. Among them, effective applications are those whose number of target operations is greater than the corresponding effective behavior threshold.

[0131] Based on the target number of operations for each effective application, calculate the total number of visits for each type of application.

[0132] When the total number of visits to all application classes exceeds the effective access threshold, it is determined that the target user has purchased a car.

[0133] S208. Electronic devices use historical car purchase intention models to process effective behavioral feature data to determine whether target users have historical car purchase intentions.

[0134] S209. Electronic devices determine the implementation status of the target user's car purchase behavior based on the current car purchase intention and historical car purchase intention.

[0135] More specifically, when a target user simultaneously exhibits both current and historical purchase intentions, it is determined that the target user has engaged in car purchase behavior. Otherwise, it is determined that the target user has not engaged in car purchase behavior.

[0136] S210: The electronic device generates a user profile of the target user based on the status of the car purchase behavior.

[0137] In the above technical solution, the electronic device uses an independent samples t-test to analyze and process the access information of multiple types of applications of the first user with a car and the second user without a car, to screen out the differential variables related to vehicle ownership status and the effective behavioral thresholds to ensure the differential. Combining the behavioral differences between car-owning and car-less users before and after purchasing a car, a current car purchase intention model is generated. The electronic device also uses Pearson correlation analysis to screen out effective behavioral feature data based on the correlation between the original car purchase behavior data of the first and second users and their car purchase intention, ensuring the independence between the feature data and reducing the complexity of the logistic regression model generated based on this. The electronic device identifies the car purchase behavior of the target user through two models trained with diverse and non-repeating data, improving the accuracy of the identification results and thus improving the accuracy of the corresponding user profile.

[0138] Figure 3 This is a schematic diagram of the structure of a big data customer acquisition device for the automotive industry according to an embodiment of this application. The big data customer acquisition device 300 for the automotive industry includes an acquisition module 301 and a processing module 302, wherein...

[0139] The acquisition module 301 is used to acquire access information and original car purchase behavior feature data of a first user who owns a vehicle and a second user who does not own a vehicle to multiple types of applications; the access information includes the access time range and operation information, and the original car purchase behavior feature data includes multiple offline behavior feature data and multiple online behavior feature data.

[0140] The processing module 302 is used to perform differential analysis on the operation information of the first user and the second user within the same access time range for various types of applications, generate analysis results, and construct a rule model based on the analysis results to generate a current car purchase intention model; the current car purchase intention model includes a first current car purchase intention model and a second current car purchase intention model, and the analysis results include the target application.

[0141] The processing module 302 is also used to perform data cleaning and preprocessing operations on offline and online behavioral feature data to determine valid behavioral feature data.

[0142] Processing module 302 is also used to train a logistic regression model using effective behavioral feature data to obtain a historical car purchase intention model.

[0143] The processing module 302 is also used to obtain the target user's access information and effective behavioral feature data for multiple types of applications, generate a user profile of the target user based on the current car purchase intention model, the access information of the target application, the historical car purchase intention model and the effective behavioral feature data, and take corresponding information recommendations based on the user profile.

[0144] In one embodiment, the processing module 302 is specifically used for:

[0145] For various applications, the operation information of the first user and the second user within the same access time range is used as a quantitative variable, and the user's ownership status of the vehicle is used as a qualitative variable. An independent samples t-test is conducted to determine the analysis results. The analysis results include the target applications related to the differences in various applications and the target operation information corresponding to each target application.

[0146] In one embodiment, the processing module 302 is specifically used for:

[0147] For each original application, count the number of times the first user and the second user performed operations on information within the same access time range;

[0148] Select valid operation information according to a preset number of times, and define the valid operation information as a quantitative variable; valid operation information is operation information whose number of operations is within the preset number of times.

[0149] The user's ownership status of the vehicle is used as a categorical variable. Independent samples t-tests are performed on the quantitative and categorical variables to calculate significance indicators.

[0150] When the significance index is less than the preset significance threshold, the original application is identified as the target application, and the minimum value of the preset number of times is identified as the effective behavior threshold corresponding to the target operation information.

[0151] In one embodiment, the processing module 302 is specifically used for:

[0152] Based on the target operation information of the target application and its corresponding effective behavior threshold, a vehicle holding status recognition model is generated.

[0153] Based on the target operation information of all target applications within a first preset time range and a first preset threshold, a first current car purchase constraint is constructed; the first current car purchase constraint is that the total amount of all target operation information generated by the user is greater than the first preset threshold.

[0154] Based on the target operation information of all target applications within the first preset time range and the second preset threshold, a second current car purchase constraint is constructed; the second current car purchase constraint is that the total amount of all target operation information generated by the user is greater than the product of the average of all target operation information generated by the user within the second preset time range and a preset multiple.

[0155] Based on the first current car purchase constraints and the vehicle ownership status recognition model, a first car purchase intention model is generated;

[0156] A second vehicle purchase intention model is generated based on the second current vehicle purchase constraints and the vehicle ownership status recognition model.

[0157] In one embodiment, the processing module 302 is specifically used for:

[0158] Pearson correlation analysis was performed on offline and online behavioral characteristic data, and redundant characteristic data was removed from the original car purchase behavior characteristic data based on the analysis results to obtain the first car purchase behavior data.

[0159] The data on the first car purchase behavior is normalized to generate valid behavioral feature data.

[0160] In one embodiment, the processing module 302 is specifically used for:

[0161] The current car purchase intention model is used to process the access information of the target application to determine whether the target user has the current car purchase intention;

[0162] By using historical car purchase intention models to process effective behavioral feature data, it can be determined whether target users have a history of car purchase intentions.

[0163] Determine the implementation status of the target user's car purchase behavior based on current and historical car purchase intentions;

[0164] Based on the status of the car purchase behavior, a user profile of the target user is generated.

[0165] In one embodiment, the processing module 302 is specifically used for:

[0166] When a target user has both current and historical car purchase intentions, it is determined that the target user has engaged in car purchase behavior.

[0167] In one embodiment, the processing module 302 is specifically used for:

[0168] The vehicle ownership status recognition model is used to process the access information of the target application to determine whether the target user has purchased a vehicle;

[0169] When the target user has not purchased a car and meets the first current car purchase constraint, it is determined that the target user has the current intention to purchase a car;

[0170] When the target user has already purchased a vehicle and meets the second current purchase constraint, it is determined that the target user has the current intention to purchase a vehicle.

[0171] In one embodiment, the processing module 302 is specifically used for:

[0172] Based on the access information of the target applications, count the number of target operations performed by the target users on each target application;

[0173] The number of target operations for each target application will be compared with the effective behavior threshold corresponding to the target operation information. Effective applications and their access volume will be selected from each target application. Among them, effective applications are those whose number of target operations is greater than the corresponding effective behavior threshold.

[0174] Based on the target number of operations for each effective application, calculate the total number of visits for each type of application.

[0175] When the total number of visits to all application classes exceeds the effective access threshold, it is determined that the target user has purchased a car.

[0176] Figure 4 This is a schematic diagram of the structure of a control electronic device according to an embodiment of this application. The electronic device 400 includes a memory 401 and a processor 402. The memory 401 stores computer instructions executable by the processor. The memory 401 may include high-speed random access memory (RAM), and may also include non-volatile memory (NVM), such as at least one disk drive, or a USB flash drive, portable hard drive, read-only memory, magnetic disk, or optical disk, etc.

[0177] When executing computer instructions, processor 402 implements the various steps in the automotive industry big data customer acquisition method with electronic devices as the execution entity in the above embodiments. For details, please refer to the relevant descriptions in the foregoing method embodiments. The processor 402 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. A general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly manifested as being executed by a hardware processor, or executed by a combination of hardware and software modules within the processor.

[0178] Optionally, the memory 401 can be either independent or integrated with the processor 402. When the memory 401 is configured independently, the control device 400 also includes a bus for connecting the memory 401 and the processor 402. This bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be categorized as an address bus, data bus, control bus, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0179] This application also provides a computer-readable storage medium storing computer instructions. When a processor executes the computer instructions, it implements the various steps of the automotive industry big data customer acquisition method described above.

[0180] This application also provides a computer program product, including computer instructions, which, when executed by a processor, implement the various steps of the automotive industry big data customer acquisition method described above.

[0181] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.

[0182] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A big data-driven customer acquisition method for the automotive industry, characterized in that, The method includes: The system acquires access information and original car-buying behavior characteristics of a first user who owns a vehicle and a second user who does not own a vehicle to various types of applications. The access information includes access time ranges and operation information. The original car-buying behavior characteristics include various offline and online behavioral characteristics. Offline behavioral characteristics refer to actual offline interactions with sales personnel or vehicle-related personnel. Online behavioral characteristics refer to vehicle-related behaviors performed by users through applications or web pages. For each type of application, the system performs differential analysis on the operation information of the first and second users within the same access time range, generating analysis results. Based on these results, a rule model is constructed to generate a current car-buying intention model. The current car-buying intention model includes a first current car-buying intention model and a second current car-buying intention model. The analysis results include the target application. The step of constructing a rule model based on the analysis results to generate a current car purchase intention model specifically includes: A vehicle holding status recognition model is generated based on the target operation information of the target application and its corresponding effective behavior threshold. Based on the target operation information of all target applications within a first preset time range and a first preset threshold, a first current car purchase constraint is constructed; the first current car purchase constraint is that the total amount of all target operation information generated by the user is greater than the first preset threshold. Based on the target operation information of all target applications within a first preset time range and a second preset threshold, a second current car purchase constraint is constructed; the second current car purchase constraint is that the total amount of all target operation information generated by the user is greater than the product of the average value of all target operation information generated by the user within the second preset time range and a preset multiple. Based on the first current vehicle purchase constraint and the vehicle ownership status recognition model, a first current vehicle purchase intention model is generated; Based on the second current vehicle purchase constraint and the vehicle ownership status recognition model, a second current vehicle purchase intention model is generated; the vehicle ownership status recognition model is used to determine whether a user owns a vehicle based on the number of operations performed by the user on each target application. The offline and online behavioral feature data are cleaned and preprocessed to determine the valid behavioral feature data. The effective behavioral feature data is used to train a logistic regression model to obtain a historical car purchase intention model; The system acquires access information and effective behavioral feature data of the target user to the various types of applications. Based on the current car purchase intention model, the access information of the target application, the historical car purchase intention model, and the effective behavioral feature data, it generates a user profile of the target user and makes corresponding information recommendations based on the user profile.

2. The method according to claim 1, characterized in that, For various applications, a differential analysis is performed on the operation information of the first user and the second user within the same access time range, generating analysis results, specifically including: For various applications, the operation information of the first user and the second user within the same access time range is used as a quantitative variable, and the user's vehicle ownership status is used as a qualitative variable. An independent samples t-test is performed to determine the analysis results. The analysis results include the target applications related to the differences in various applications and the target operation information corresponding to each target application.

3. The method according to claim 2, characterized in that, Various applications contain multiple original applications; using the operation information of the first user and the second user within the same access time range as quantitative variables, and the user's vehicle ownership status as a categorical variable, an independent samples t-test analysis is performed to determine the analysis results, specifically including: For each of the original applications, count the number of times the first user and the second user perform the operation information within the same access time range; Valid operation information is selected according to a preset number of times, and this valid operation information is determined as a quantitative variable; the valid operation information is operation information whose number of operations is within the preset number of times. The user's ownership status of the vehicle is used as a categorical variable. An independent samples t-test is performed on the quantitative and categorical variables to calculate the significance index. When the significance index is less than the preset significance threshold, the original application is identified as the target application, and the minimum value of the preset number range is identified as the effective behavior threshold corresponding to the target operation information.

4. The method according to claim 1, characterized in that, The offline and online behavioral feature data are cleaned and preprocessed to determine valid behavioral feature data, specifically including: Pearson correlation analysis was performed on the offline and online behavioral feature data, and redundant feature data was removed from the original car purchase behavior feature data based on the analysis results to obtain the first car purchase behavior data. The first car purchase behavior data is normalized to generate valid behavioral feature data.

5. The method according to claim 1, characterized in that, Based on the current car purchase intention model, the access information of the target application, the historical car purchase intention model, and the effective behavioral feature data, a user profile of the target user is generated, specifically including: The current car purchase intention model is used to process the access information of the target application to determine whether the target user has a current car purchase intention; The effective behavioral feature data is processed using the historical car purchase intention model to determine whether the target user has a history of car purchase intention; Based on the current purchase intention and the historical purchase intention, determine the implementation status of the target user's purchase behavior; Based on the status of the car purchase behavior, a user profile of the target user is generated.

6. The method according to claim 5, characterized in that, Based on the current purchase intention and the historical purchase intention, the implementation status of the target user's car purchase behavior is determined, specifically including: When the target user has both the current intention to purchase a car and the historical intention to purchase a car, it is determined that the target user has engaged in car purchase behavior.

7. The method according to claim 5, characterized in that, The current car purchase intention model is used to process the access information of the target application to determine whether the target user has a current car purchase intention, specifically including: The access information of the target application is processed using a vehicle ownership status recognition model to determine whether the target user has purchased a vehicle. When the target user has not purchased a car and meets the first current car purchase constraint, it is determined that the target user has a current car purchase intention; When the target user has already purchased a vehicle and meets the second current vehicle purchase constraint, it is determined that the target user has a current vehicle purchase intention.

8. The method according to claim 7, characterized in that, The access information of the target application is processed using a vehicle ownership status recognition model to determine whether the target user has purchased a vehicle. Specifically, this includes: Based on the access information of the target applications, count the number of corresponding target operations performed by the target users on each target application; The number of target operations performed on each target application is compared with the effective behavior threshold corresponding to the target operation information. Valid applications and their access volumes are then selected from each target application. Valid applications are those whose number of target operations exceeds the corresponding effective behavior threshold. Based on the target number of operations for each effective application, calculate the total number of visits for each type of application. When the total number of visits to all application classes exceeds the effective access threshold, it is determined that the target user has purchased a car.

9. A big data customer acquisition device for the automotive industry, characterized in that, include: The acquisition module is used to acquire access information and original car purchase behavior characteristic data of a first user who owns a vehicle and a second user who does not own a vehicle to multiple types of applications. The access information includes the access time range and operation information, and the original car purchase behavior characteristic data includes multiple offline behavior characteristic data and multiple online behavior characteristic data. Among them, the offline behavior characteristic data refers to the actual offline operations performed with sales personnel or vehicle-related personnel; the online behavior characteristic data refers to the vehicle-related behaviors performed by users through applications or web pages. The processing module is used to perform differential analysis on the operation information of the first user and the second user within the same access time range for various types of applications, generate analysis results, and construct a rule model based on the analysis results to generate a current car purchase intention model; the current car purchase intention model includes a first current car purchase intention model and a second current car purchase intention model, and the analysis results include the target application; The processing module is also used to perform data cleaning and preprocessing operations on the offline behavioral feature data and online behavioral feature data to determine valid behavioral feature data. The processing module is also used to train a logistic regression model using the effective behavioral feature data to obtain a historical car purchase intention model; The processing module is also used to obtain the target user's access information and effective behavioral feature data of the multiple types of applications, generate a user profile of the target user based on the current car purchase intention model, the access information of the target application, the historical car purchase intention model and the effective behavioral feature data, and take corresponding information recommendations based on the user profile; The processing module is specifically used to generate a vehicle holding status recognition model based on the target operation information of the target application and its corresponding effective behavior threshold. Based on the target operation information of all target applications within a first preset time range and a first preset threshold, a first current car purchase constraint is constructed; the first current car purchase constraint is that the total amount of all target operation information generated by the user is greater than the first preset threshold. Based on the target operation information of all target applications within a first preset time range and a second preset threshold, a second current car purchase constraint is constructed; the second current car purchase constraint is that the total amount of all target operation information generated by the user is greater than the product of the average value of all target operation information generated by the user within the second preset time range and a preset multiple. Based on the first current vehicle purchase constraint and the vehicle ownership status recognition model, a first current vehicle purchase intention model is generated; Based on the second current vehicle purchase constraint and the vehicle ownership status recognition model, a second current vehicle purchase intention model is generated; the vehicle ownership status recognition model is used to determine whether the user owns a vehicle based on the number of operations performed by the user on each target application.

10. An electronic device, characterized in that, include: A processor and a memory communicatively connected to the processor; The memory stores computer instructions; When executing the computer instructions, the processor is used to implement the automotive industry big data customer acquisition method as described in any one of claims 1 to 8.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a processor, are used to implement the automotive industry big data customer acquisition method as described in any one of claims 1 to 8.

Images