Business operation method, device and readable medium

Abstract

The embodiment of the application relates to the technical field of vehicle enterprise operation, and discloses a business operation method and device, computer equipment and readable medium, the method comprising: acquiring user feature data set; according to the user feature data set, dividing N users into M user groups; determining at least one split line in any feature dimension; according to the value evaluation score of each user group in each feature dimension, respectively determining the comprehensive evaluation score of each user group; respectively determining the business strategy of each user group, and executing business operation on the M user groups according to the business strategy, solving the problem of insufficient data caused by low-frequency interaction in the current vehicle enterprise industry, and giving the corresponding value evaluation score business logic for each user group, pushing high-value customers to the vehicle enterprise, or matching low-value users with potential value mining business logic, so that different resources and directions can be tilted for each user.

Description

Technical Field

[0001] This application relates to the field of vehicle applications, specifically to a business operation method, apparatus, and readable medium. Background Technology

[0002] Currently, in sectors that utilize different user data to conduct business operations according to different business strategies, such as the internet or e-commerce industries, user value models can be used to identify valuable users, and then different business operations can be matched according to different value users. However, in the automotive industry, car purchase is a decision-making behavior with a long cycle, and the frequency of car purchases is different from that of e-commerce products. E-commerce products can be purchased in multiple quantities at the same time, and the purchase behavior can range from dozens to hundreds of times. However, car manufacturers' users generally replace their cars every few years, and the purchase behavior is mostly zero or one time. The original user value model is difficult to apply directly. For car manufacturers, it is impossible to effectively distinguish between relatively high-value users and low-value users, which leads to problems in the subsequent matching of business strategies, thus exhibiting shortcomings. Summary of the Invention

[0003] To address the issue that existing user value models cannot be directly applied to matching automotive companies' user business strategies, embodiments of this application provide a business operation method, apparatus, and readable medium that can identify high-value users from a massive user base, thereby enabling the matching of higher-quality business operations to high-value users and more affordable or economical business operations to low-value users.

[0004] On one hand, embodiments of this application provide a business operation method, including: Obtain a user feature dataset, which includes user feature groups of N users. Each user feature group includes K feature dimensions, which include the number of times the user logs in on the advertising platform where the target object is located, the user's login duration, and the user's payment depth for the target object. Based on the user feature dataset, the N users are divided into M user groups; At least one dividing line is determined on any feature dimension, and the at least one dividing line is used to divide the M user groups into at least two feature intervals on any feature dimension, and each feature interval corresponds to a value assessment score. Based on the value assessment score of each user group in the feature interval of each feature dimension, a comprehensive assessment score is determined for each user group. Based on the comprehensive evaluation score of each user group, a business strategy is determined for each user group, and business operations are performed on the M user groups in accordance with the business strategy.

[0005] On the other hand, embodiments of this application provide a business operation apparatus, including: The acquisition module acquires a user feature dataset, which includes user feature groups of N users. Each user feature group includes K feature dimensions, which include the number of times the user logs in on the advertising platform where the target object is located, the user's login duration, and the user's payment depth for the target object. The user segmentation module divides the N users into M user groups based on the user feature dataset. The dividing line determination module determines at least one dividing line on any feature dimension. The at least one dividing line is used to divide the M user groups into at least two feature intervals on any feature dimension. Each feature interval corresponds to a value assessment score. The evaluation module determines the comprehensive evaluation score for each user group based on the value evaluation score of each user group in the feature interval of each feature dimension. The business operation module determines the business strategy for each user group based on the comprehensive evaluation score of each user group, and executes the business operations for the M user groups according to the business strategy.

[0006] In an optional embodiment, the evaluation module includes: The value assessment score of each user group in each feature interval on each feature dimension is multiplied by the corresponding weight to obtain the weighted score of each feature dimension. For each user group, the sum of all weighted scores is calculated to obtain a comprehensive evaluation score.

[0007] In an optional embodiment, the business operation module includes: If the overall evaluation score of a user group is lower than the first threshold, the business strategy for the user group is determined to be to push low-end vehicle product introduction advertisements, and low-end vehicle product introduction advertisements are pushed to the user group at appropriate times. If the overall evaluation score of a user group is lower than the second threshold but not lower than the first threshold, the business strategy for the user group is determined to be to push information about car company test drive activities, and the information about car company test drive activities is pushed to the user group at an appropriate time. If the overall evaluation score of a user group is lower than the third threshold but not lower than the second threshold, the business strategy for the user group is determined to be to push the latest promotional activities and high-end car information, and the latest promotional activities and high-end car information will be pushed to the user group at appropriate times.

[0008] In an optional embodiment, it further includes: The contour coefficient and relative entropy determination module determines the contour coefficient and relative entropy of each user group based on the data of all users in each user group. The separation degree determination module determines the group separation degree of each user group based on the profile coefficient and relative entropy of each user group; The user separation module separates at least one user from each user group based on the group separation degree of each user group.

[0009] In an optional embodiment, it further includes: The feature distance determination module, for each separated user, obtains the feature distance between the user and each user group based on the user data of the user and the user data of at least one user in each user group; The classification module reclassifies the separated users into one of the user groups based on the feature distance between the separated users and each user group.

[0010] In an optional embodiment, the contour coefficient and relative entropy determination module includes: The first calculation module calculates the product of the first weight and the silhouette coefficient, and calculates the product of the second weight and the relative entropy; The separation degree calculation module uses the sum of the two products as the group separation degree of the user group.

[0011] In an optional embodiment, the clustering module includes: The model input unit inputs user interaction data into the user value model to obtain the three-dimensional coordinates of each user in the model space, wherein each feature dimension coordinate axis of the model space represents an interaction feature index of a feature dimension. The clustering unit clusters the users based on their three-dimensional coordinates in the model space to obtain the multiple user groups.

[0012] In an optional embodiment, the value assessment score increases or decreases along the coordinate axis on the corresponding feature dimension coordinate. The segmentation module includes: The coordinate lookup unit searches for the coordinate range of all user coordinates on the corresponding coordinate axis for each feature dimension. The partitioning unit divides the coordinate range into multiple partitions according to a set number of layers, and obtains the dividing line between each pair of adjacent partitions.

[0013] In another aspect, embodiments of this application provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method described above.

[0014] In another aspect, embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0015] Beneficial effects This application provides a business operation method, apparatus, computer equipment, and readable medium. By combining user groups and user value models, each user group is divided into different intervals, and then the dividing line of each interval is obtained. Since each coordinate in the model space of the user value model corresponds to a value assessment score, after determining each interval dividing line, the assessment score of each user group on each single feature dimension can be determined by combining the coordinates of each interval dividing line. Then, by summing the assessment scores of the three feature dimensions, the high-value user groups can be identified. Although the interaction data of an individual user is relatively small in this application, the data of each user group is used as the unit after dividing the user groups, so the data of each individual analysis object increases accordingly. The value assessment score of each interval is determined according to the coordinates of the interval dividing line, which is equivalent to equating the interaction data of car company users in one interval with a user in other high-frequency interaction industries. This solves the problem of insufficient data volume in the current car company industry due to low-frequency interaction. It can provide business logic with corresponding value assessment scores for each user group, and can push high-value customers to car companies or match low-value users with business logic for potential value mining, so that different resources and directions can be tilted towards each user. Attached Figure Description

[0016] Figure 1 A flowchart illustrating the business operation method provided in the embodiments of this application; Figure 2 This is a flowchart illustrating one of the sub-steps of the business operation method provided in the embodiments of this application; Figure 3 The second flowchart illustrates the sub-steps of the business operation method provided in the embodiments of this application; Figure 4 The third flowchart illustrating the sub-steps of the business operation method provided in the embodiments of this application; Figure 5 This is a schematic diagram of the structure of the model space and user group provided in the embodiments of this application; Figure 6 The fourth flowchart illustrating the sub-steps of the business operation method provided in the embodiments of this application; Figure 7 A schematic diagram of the module of the business operation device provided in the embodiments of this application; Figure 8 A schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the specific technical solutions of this application will be further described in detail below with reference to the accompanying drawings of the embodiments of this application. The following embodiments are used to illustrate this application, but are not intended to limit the scope of this application.

[0018] In the internet or e-commerce industry, user value models can be used to identify valuable users. This is because the amount of interactive data generated by users in the internet or e-commerce industry is large, and high-value users can be identified by simply processing the data using conventional user value models. However, for car manufacturers, users often purchase cars only once every few years, resulting in a small amount of data available for input.

[0019] Based on this, this application first combines user groups and user value models to divide each user group into different intervals, and then obtains the dividing line for each interval. Since each coordinate in the model space of the user value model corresponds to a value assessment score, after determining each interval dividing line, the assessment score of each user group on each single feature dimension can be determined by combining the coordinates of each interval dividing line. Then, by summing the assessment scores of the three feature dimensions, the high-value user groups can be identified. Although the interaction data of a single user is relatively small in this application, the data of each user group is used as the unit after dividing the user groups, so the data of a single analysis object increases accordingly. Based on the coordinates of the interval dividing line, the value assessment score of each interval is determined, which is equivalent to equating the interaction data of car company users in one interval with a user in other high-frequency interaction industries. This solves the problem of insufficient data volume in the current car company industry due to low-frequency interaction. It can provide business logic for giving corresponding value assessment scores to each user group, and can push high-value customers to car companies or match low-value users with business logic for potential value mining, so that different resources and directions can be tilted towards each user.

[0020] Reference Figure 1 This is a flowchart illustrating a business operation method provided in an embodiment of this application. Figure 1 As shown, the business operation method includes: S1: Obtain a user feature dataset, which includes user feature groups of N users. Each user feature group includes K feature dimensions. The K feature dimensions include the number of times the user logs in on the advertising platform where the target object is located, the user's login duration, and the user's payment depth for the target object. S2: Divide the N users into M user groups based on the user feature dataset; S3: Determine at least one dividing line on any feature dimension, the at least one dividing line being used to divide the M user groups into at least two feature intervals on any feature dimension, each feature interval corresponding to a value assessment score; S4: Determine the comprehensive evaluation score for each user group based on the value assessment score of each user group in the feature interval of each feature dimension; S5: Based on the comprehensive evaluation score of each user group, determine the business strategy for each user group, and execute business operations on the M user groups according to the business strategy.

[0021] In this embodiment, by combining user groups and user value models, each user group is divided into different intervals, and then the dividing line of each interval is obtained. Since each coordinate in the model space of the user value model corresponds to a value assessment score, after determining each interval dividing line, the assessment score of each user group on each single feature dimension can be determined by combining the coordinates of each interval dividing line. Then, by summing the assessment scores of the three feature dimensions, the high-value user groups can be identified. Although the interaction data of a single user is relatively small in this application, the data of a single analysis object is increased accordingly after dividing the user groups into user groups. The value assessment score of each interval is determined according to the coordinates of the interval dividing line, which is equivalent to equating the interaction data of car company users in one interval with a user in other high-frequency interaction industries. This solves the problem of insufficient data volume in the current car company industry due to low-frequency interaction. It can provide business logic for giving corresponding value assessment scores to each user group, and can push high-value customers to car companies or match low-value users with business logic for potential value mining, so that different resources and directions can be tilted towards each user.

[0022] In this application, the target object can be a car, and the advertising platform can be a customized program by the car manufacturer or a third-party advertising program, such as a mobile car manufacturer terminal.

[0023] In this embodiment, the comprehensive score can be calculated using the weight of each feature dimension. That is, based on the value assessment score of each user group in the feature interval of each feature dimension, the comprehensive assessment score of each user group is determined, including: The value assessment score of each user group in each feature interval on each feature dimension is multiplied by the corresponding weight to obtain the weighted score of each feature dimension. For each user group, the sum of all weighted scores is calculated to obtain a comprehensive evaluation score.

[0024] In this embodiment, since each feature dimension corresponds to a weight, the influence of each feature dimension can be weighed by adjusting the weight size. Thus, the feature dimensions that have a greater impact on the final car purchase behavior are given a larger weight, while the feature dimensions that have a smaller impact on the car purchase are given a smaller weight.

[0025] In this application embodiment, users whose comprehensive evaluation score is higher than a set threshold are the objects of attention, and those whose comprehensive evaluation score is lower than the set threshold are the objects to be ignored or potential objects to be explored. The value of the set threshold can be based on different business definitions, and this application does not limit it.

[0026] Specifically, based on the comprehensive evaluation score of each user group, a business strategy is determined for each user group, and business operations are performed on the M user groups according to the business strategy. For example, this may include: If the overall evaluation score of a user group is lower than the first threshold, the business strategy for the user group is determined to be to push low-end vehicle product introduction advertisements, and low-end vehicle product introduction advertisements are pushed to the user group at appropriate times. If the overall evaluation score of a user group is lower than the second threshold but not lower than the first threshold, the business strategy for the user group is determined to be to push information about car company test drive activities, and the information about car company test drive activities is pushed to the user group at an appropriate time. If the overall evaluation score of a user group is lower than the third threshold but not lower than the second threshold, the business strategy for the user group is determined to be to push the latest promotional activities and high-end car information, and the latest promotional activities and high-end car information will be pushed to the user group at appropriate times.

[0027] It is important to understand that the first threshold is less than the second threshold, and the second threshold is less than the third threshold.

[0028] If the threshold is less than the first threshold, it means that the user group is indifferent to the vehicle or does not want to buy it temporarily due to economic pressure. In this case, you can increase the push of introductions to lower-end models to generate their desire to buy. If the threshold is less than the second threshold, it means that the user group has a relatively strong desire to buy, but is still in the decision-making period due to short-term economic constraints. Therefore, you can push test drive activities, etc. If the threshold is greater than the second threshold, it means that the user group will definitely buy, but is currently in the process of choosing a model. In this case, you can push advertisements for higher-end models.

[0029] In step S2 of this embodiment, existing clustering algorithms can be used to initially cluster all available users into M user groups. Then, for each feature dimension, a dividing line can be found that separates the M user groups. Generally, at most two dividing lines can be found for each feature dimension, that is, the M user groups are divided into 3 intervals and 3 feature dimensions. It can be understood that in a three-dimensional coordinate system, the three feature dimensions correspond to a three-dimensional space. After the three feature dimensions are divided, a total of 3*3*3=27 "quadrants" can be divided. Each "quadrant" is formed by the combination of intervals of the three dimensions. The purpose is to assign the M user groups to the corresponding quadrants.

[0030] Optionally, in this application, each quadrant contains only one or zero user groups, and each user group corresponds to a business strategy. For example, the quadrant with the overall lower values ​​of the three feature dimensions is ignored; the quadrant with the overall higher values ​​of the three feature dimensions is the focus. Based on the business strategy, different business operations are adopted for user groups in different quadrants, such as different recommendation strategies and different marketing content.

[0031] Taking the above three feature dimensions as an example, there are Kx intervals, namely K1 intervals, K2 intervals and K3 intervals. K1, K2 and K3 can be the same or different, and this application does not impose any restrictions on this.

[0032] It should be noted that in the embodiments of this application, the user groups may "overlap" on each single feature dimension. That is, for a specific point on a single feature dimension, it may be included by both user group a and user group b. In this case, based on the user value model, since the value values ​​of user group a and user group b are similar, user group a and user group b can be divided into one interval.

[0033] It is further important to understand that in the embodiments of this application, each dividing line is used to divide the M user groups into Kx intervals. That is, a single user group is divided as the smallest divisible unit. In other words, there is no situation where a part of a user group is divided into one partition and another part is divided into another partition.

[0034] The following is a description of this application. Figure 1 The illustrated embodiments will be described in detail.

[0035] A user value model is a model that can evaluate multi-dimensional characteristic data of users. Specifically, the user value model in this application can adopt one of the following: RFM value model, CLV value model, and customer social value model.

[0036] The RFM value model is a classic model in user value research in the marketing industry. RFM refers to three indicators: Recency, Frequency, and Monetary. Combining these three indicators forms a three-dimensional space, where each user has three corresponding coordinates. Users can be clustered in this three-dimensional space to find groups with high similarity. Each group has a corresponding score based on its position in the space.

[0037] The CLV (Customer Lifetime Value) model is an effective prediction of a customer's future profits. It measures how much value a customer (user) brings to a company over a period of time. The CLV model can help car companies distinguish between the best customers and those that are not worth investing in. The goal is to extend the customer's lifecycle as much as possible and generate business value during that lifecycle.

[0038] The customer social value model is used to differentiate the social activity of brands. In the current new business environment, the relationship between brands and users is not just about consumption and shopping, and product promotion, but more about interaction and mutual understanding. The inherent dissemination attributes and influence of individual users are receiving more and more attention and application.

[0039] The user value model of this application can also adopt other value models. This application does not limit this. It should be understood that, regardless of the value model, since it includes multiple feature dimensions, it can form a "model space", that is, vectorize the feature dimensions to form "coordinate axes". User data can determine the user's coordinates based on the feature values ​​of the feature data. For a two-dimensional model space, the user coordinates can be represented by (x, y). For a three-dimensional model space, the user coordinates can be represented by (x, y, z).

[0040] This application divides user groups into user value models based on feature dimensions. Specifically, assuming that the clustered user groups include a, b, c, and d, and the three dimensions are x1, x2, and x3 respectively, then in the x1 dimension, for example, the interval dividing line is between a and b; in the x2 dimension, for example, the interval dividing line is between c and d; and in the x3 dimension, the interval dividing line is between b and c, the value of each interval can be evaluated on a single feature dimension by corresponding to the coordinates of each dividing line.

[0041] In some embodiments of this application, the value assessment scores of each feature dimension can be added together. For example, in the above embodiment, the score of user group a can be the sum of the value assessment scores of a in dimension x1, a in dimension x2, and a in dimension x3.

[0042] In other feasible embodiments, the value assessment score of each feature dimension can be multiplied by the corresponding weight and then summed. For example, in the above embodiment, the score of user group a can be the sum of the product m1 of the first weight k1 and the value assessment score of a in the x1 dimension, the product m2 of the second weight k2 and the value assessment score of a in the x2 dimension, and the product m3 of the third weight k3 and the value assessment score of a in the x3 dimension, that is: the value assessment score of user group a = m1 + m2 + m3. This application will not elaborate on this.

[0043] The above embodiments provide a method for calculating the value assessment score of the user group in this application. Since the embodiments of this application allocate a certain assessment score to the interval where the user group is located, compared with the current method of assessing a score for a single user, since the user group includes multiple users after classification, their characteristics have strong similarity and consistency, and can be regarded as "one user". Thus, the data volume of a user group is used as the assessment object, so the amount of assessment data is guaranteed. Therefore, for car manufacturers, an accurate value assessment score can be obtained, which solves the current problem that car manufacturers cannot be assessed or the assessment is inaccurate due to insufficient data.

[0044] In a preferred embodiment, this application can improve the user value model. Specifically, taking the RFM model as an example, the RFM model is suitable for high-frequency online platforms and high-frequency consumer goods, such as vegetables, fresh produce, and clothing. The user visit duration requirement is not high, but the user stickiness requirement is high. Since car company apps are low-frequency payment scenarios, and the information update frequency of car categories is lower than that of daily consumer goods, we have improved the dimensions of the user value model based on the characteristics of car company app scenarios. Recency has been improved to interaction depth, that is, the average visit duration of users in the past 30 days. Therefore, the calculation logic of the user value model has been adjusted from RFM to DFM, which stands for Depth, Frequency, and Monetary. This model is more in line with the decision-making habits of car company buyers. The user's use of the car company app is not based on single short-term stickiness, but on the depth of understanding of the car category information in the app and the frequency of repeated access. Finally, the user value score is calculated by combining the amount of payment made by the user to the car company.

[0045] As can be seen, in this embodiment, the three feature dimensions in the model space are activity level, payment depth, and interaction depth. Taking car company private domain users as an example, Table 1 shows the user segmentation table using this DFM. Activity level (frequency) is calculated using "the number of days the user has logged in in the past 30 days," payment depth (monetary) is represented by "the type of payment the user makes with the brand," and interaction depth (Depth) is calculated using "the average login time the user has spent in the past 30 days." • F dimension: Number of days a user has logged in over the past 30 days • Dimension D: Average login time of users in the past 30 days • M-dimensional: Adding to wishlist / Wishlist deposit / Deposit / Pre-order behavior After using simulated data, the stratified data statistics in Table 1 can be obtained, namely: Table 1 - User DFM Value Stratification Table

[0046] This table shows the number of users after stratifying the data using the F and M dimensions. For this group of users, those facing the bottom right corner of the table have higher user value. Once determined, these users can be prioritized for push notifications to those with large orders of 5 days or more.

[0047] Of course, this table only extracts some features for illustration, and this application is not limited to this. This application can use three feature dimensions through the DFM model, which will not be elaborated on here.

[0048] like Figure 5As shown in the model space of this application, the x-axis can represent the interaction depth, the y-axis can represent the frequency, and the z-axis can represent the amount. As can be seen from the figure, each user group corresponds to a region. By finding the interval dividing line on each coordinate axis, the value evaluation score of each user group in the three dimensions can be summarized by defining the value evaluation score of each interval dividing line.

[0049] Furthermore, in this embodiment, users can be clustered in the model space. Specifically, all users can be input into the model space according to three dimensions, and then clustered based on the Euclidean distance between users. In this embodiment, the specific clustering steps may include: S001: Based on the user value model and user interaction data, users are clustered to obtain multiple user groups.

[0050] More specifically, such as Figure 2 As shown, this step includes: S0011: Input the user's interaction data into the user value model to obtain the three-dimensional coordinates of each user in the model space, wherein each feature dimension coordinate axis of the model space represents an interaction feature index of a feature dimension. S0012: Cluster the users according to the three-dimensional coordinates of each user in the model space to obtain the multiple user groups.

[0051] This embodiment demonstrates how to perform clustering in the model space to evaluate the value of the clustered user groups.

[0052] In the embodiments of this application, the interactive data may be user login, payment, discussion, and other operations, which will not be described in detail here.

[0053] The embodiments of this application can employ partition-based clustering methods, density-based clustering methods, hierarchical clustering methods, etc. Taking partition-based clustering as an example, partition-based clustering methods require specifying the number of clusters or cluster centers in advance, and iterating repeatedly until the goal of "points within clusters are close enough and points between clusters are far enough" is finally achieved. Examples include k-means and its variants k-means++, bi-kmeans, kernel k-means, etc.

[0054] It should be understood that this application does not elaborate on density-based clustering methods and hierarchical clustering methods, but their concepts are similar to partitioning clustering. Furthermore, this application does not limit the specific clustering method, meaning that any clustering method should be within the scope of protection of this application.

[0055] Furthermore, in this embodiment, the separation degree of the clustered groups can be evaluated, and then the user groups can be separated according to the separation degree, thereby improving the similarity of the user groups and making the clustering more accurate. In this embodiment, before determining at least one dividing line on each feature dimension of the user value model, such as Figure 3 As shown, the business operation method further includes: S01: Determine the profile coefficient and relative entropy of each user group based on all user data in each user group; S02: Determine the group separation degree of each user group based on the profile coefficient and relative entropy of each user group; S03: Based on the group separation degree of each user group, separate at least one user from each user group.

[0056] Specifically, the silhouette coefficient is an evaluation metric for clustering performance, used to comprehensively assess the cohesion and separation of clustering results. It can be used to evaluate the impact of different algorithms or different algorithm execution methods on the clustering results, based on the same original data. In clustering algorithms, especially the K-means algorithm, an initial number of clusters K needs to be set. For each vector in a cluster, its silhouette coefficient is calculated. The clustering results of 2 clusters are significantly different from those of 3 clusters. The silhouette coefficient value ranges from [-1, 1], with values ​​closer to 1 indicating relatively better cohesion and separation. The average silhouette coefficient of all points is the total silhouette coefficient of the clustering result.

[0057] Relative entropy, also known as KL divergence, is a metric used to measure the similarity between two probability distributions. From a sampling perspective, KL divergence can be intuitively explained as follows: it describes the encoding loss when using a distribution to estimate the true distribution of data. A lower KL divergence indicates better clustering performance. Its function is similar to the silhouette coefficient, but the evaluation perspective differs. The silhouette coefficient focuses more on inter-class evaluation, while KL divergence evaluates the overall distribution. This application creatively combines the silhouette coefficient and KL divergence to evaluate clustering performance, thus considering both inter-class and overall distribution, resulting in more accurate evaluation results.

[0058] Furthermore, this application embodiment provides a specific execution step for step S03, namely, for step S03, as follows: Figure 4 As shown, it includes the following: S031: Calculate the product of the first weight and the silhouette coefficient, and calculate the product of the second weight and the relative entropy; S032: The sum of the two products is taken as the group separation degree of the user group.

[0059] Specifically, for example, if the first weight is km1 and the second weight is km2, the contour coefficient can be calculated using the contour coefficient formula to obtain the contour coefficient of each user. Then, the contour coefficients of all users are averaged to obtain the contour coefficient of the user group. For example, l1 is the contour coefficient of the user group and l2 is the relative entropy of the user group. Thus, the group separation degree is: km1xl1+km2xl2. This application will not elaborate on this.

[0060] After determining the separation degree, this application can separate users from the user group according to preset rules. Specifically, it can gradually separate from the edge to the center, first removing users at the edge of the user group's defined space, then recalculating the group separation degree. When the recalculated group separation degree is less than the set value, the separation stops, thereby continuously stripping away "unfit" users from the edge to the center.

[0061] Of course, other user separation methods can also be used in this application, and this application does not limit them. For example, the contour coefficient and KL divergence between each user and other users can be used as measurement factors to form the similarity between each user and the user group, and then "non-group" users can be separated according to the similarity.

[0062] After obtaining the group separation degree, the clustering effect of the user group can be evaluated. The higher the group separation degree, the worse the clustering effect, that is, more users are classified into the wrong user group. Based on the model space, this application can use distance as the partitioning index to separate "non-social" users, specifically including: S04: For each separated user, based on its user data and the user data of at least one user in each user group, obtain the feature distance between the user and each user group. S05: Based on the characteristic distance between the separated users and each user group, the separated users are reclassified into one of the user groups.

[0063] In this embodiment, the feature distance between the user and each user group is combined to re-add the separated users to the new user group from the perspective of the model space. Then, the group separation degree of the new user group can be verified. Through continuous iteration, user groups with good clustering effect can be obtained.

[0064] The specific layering is explained in detail below. Furthermore, since the model space of the value model defines coordinate axes, it can be divided into partitions based on specific spatial contexts, such as... Figure 6 As shown, step S1 includes: S11: For each feature dimension, find the coordinate range of all user coordinates on the corresponding coordinate axis; S12: Divide the coordinate range into multiple partitions according to the set number of layers; S13: For each feature dimension, group at least one user group within the same partition into the same interval, so as to divide all user groups into multiple intervals on each single feature dimension.

[0065] For example, for each feature dimension, assuming the number of layers is set to four, the coordinate axis corresponding to the coordinate range of the user group can be divided into four equal partitions. Then, at least one user group in the same partition can be grouped into the same interval. At this time, the interval dividing line is equivalent to each equal division point.

[0066] As can be seen from the above embodiments, this application combines user groups and user value models to divide each user group into different intervals, and then obtains the dividing line of each interval. Since each coordinate in the model space of the user value model corresponds to a value assessment score, after determining each interval dividing line, the assessment score of each user group on each single feature dimension can be determined by combining the coordinates of each interval dividing line. Then, by summing the assessment scores of the three feature dimensions, the high-value user groups can be determined. Although the interaction data of a single user is relatively small in this application, the data of a single analysis object is increased accordingly after dividing the user groups and using the data of the user groups as the unit. According to the coordinates of the interval dividing line, the value assessment score of each interval is determined, which is equivalent to equating the interaction data of car company users in one interval with a user in other high-frequency interaction industries. This solves the problem of insufficient data volume in the current car company industry due to low-frequency interaction. It can provide business logic for giving corresponding value assessment scores to each user group, and can push high-value customers to car companies or match low-value users with business logic for potential value mining, so that different resources and directions can be tilted towards each user.

[0067] Meanwhile, this application integrates machine learning algorithms and user value models, combines them with the automotive industry scenario, and reasonably modifies the original traditional model. It uses algorithms and business to supplement the calculation range of the original model, which can more comprehensively meet the business's needs for evaluating user value, and further promote the marketing efficiency of refined operations and the utilization rate of marketing resources.

[0068] Corresponding to the business operation method described in the above embodiments, this application provides a business operation licensing device. For ease of explanation, only the parts related to this application are shown.

[0069] Figure 7 An embodiment of this application illustrates a business operation apparatus, comprising: Module 1 acquires a user feature dataset, which includes user feature groups of N users. Each user feature group includes K feature dimensions, which include the number of times the user logs in on the advertising platform where the target object is located, the user's login duration, and the user's payment depth for the target object. User segmentation module 2 divides the N users into M user groups based on the user feature dataset; The dividing line determination module 3 determines at least one dividing line on any feature dimension. The at least one dividing line is used to divide the M user groups into at least two feature intervals on any feature dimension. Each feature interval corresponds to a value assessment score. Evaluation module 4 determines the comprehensive evaluation score for each user group based on the value evaluation score of each user group in the feature interval of each feature dimension. The business operation module 5 determines the business strategy for each user group based on the comprehensive evaluation score of each user group, and performs business operations on the M user groups according to the business strategy.

[0070] By combining the three modules mentioned above, during business operations, user groups and user value models are used to divide each user group into different intervals, resulting in interval dividing lines. Since each coordinate in the user value model's model space corresponds to a value assessment score, after determining each interval dividing line, the assessment score of each user group on each single feature dimension can be determined by combining the coordinates of each interval dividing line. Then, by summing the assessment scores of the three feature dimensions, high-value user groups can be identified. Although the interaction data of a single user is relatively small in this application, the data is divided into user groups, thus increasing the data for a single analysis object. Determining the value assessment score of each interval based on the coordinates of the interval dividing lines is equivalent to equating the interaction data of a car company user in one interval with that of a user in other high-frequency interaction industries. This solves the problem of insufficient data volume in the current car company industry due to low-frequency interactions. It can provide business logic for giving corresponding value assessment scores to each user group, push high-value customers to car companies, or match low-value users with business logic for potential value mining, thereby allowing different resources and directions to be allocated to each user.

[0071] In an optional embodiment, the evaluation module includes: The value assessment score of each user group in each feature interval on each feature dimension is multiplied by the corresponding weight to obtain the weighted score of each feature dimension. For each user group, the sum of all weighted scores is calculated to obtain a comprehensive evaluation score.

[0072] In an optional embodiment, the business operation module includes: If the overall evaluation score of a user group is lower than the first threshold, the business strategy for the user group is determined to be to push low-end vehicle product introduction advertisements, and low-end vehicle product introduction advertisements are pushed to the user group at appropriate times. If the overall evaluation score of a user group is lower than the second threshold but not lower than the first threshold, the business strategy for the user group is determined to be to push information about car company test drive activities, and the information about car company test drive activities is pushed to the user group at an appropriate time. If the overall evaluation score of a user group is lower than the third threshold but not lower than the second threshold, the business strategy for the user group is determined to be to push the latest promotional activities and high-end car information, and the latest promotional activities and high-end car information will be pushed to the user group at appropriate times.

[0073] Based on the same inventive concept, in optional embodiments, it further includes: The contour coefficient and relative entropy determination module determines the contour coefficient and relative entropy of each user group based on the data of all users in each user group. The separation degree determination module determines the group separation degree of each user group based on the profile coefficient and relative entropy of each user group; The user separation module separates at least one user from each user group based on the group separation degree of each user group.

[0074] Based on the same inventive concept, in optional embodiments, it further includes: The feature distance determination module, for each separated user, obtains the feature distance between the user and each user group based on the user data of the user and the user data of at least one user in each user group; The classification module reclassifies the separated users into one of the user groups based on the feature distance between the separated users and each user group.

[0075] Based on the same inventive concept, in optional embodiments, it further includes: The clustering module clusters users based on the user value model and user interaction data to obtain multiple user groups.

[0076] Based on the same inventive concept, in an optional embodiment, the contour coefficient and relative entropy determination module includes: The first calculation module calculates the product of the first weight and the silhouette coefficient, and calculates the product of the second weight and the relative entropy; The separation degree calculation module uses the sum of the two products as the group separation degree of the user group.

[0077] Based on the same inventive concept, in an optional embodiment, the clustering module includes: The model input unit inputs user interaction data into the user value model to obtain the three-dimensional coordinates of each user in the model space, wherein each feature dimension coordinate axis of the model space represents an interaction feature index of a feature dimension. The clustering unit clusters the users based on their three-dimensional coordinates in the model space to obtain the multiple user groups.

[0078] Based on the same inventive concept, in an optional embodiment, each user group corresponds to a value assessment score on each feature dimension coordinate of the model space of the user value model, and the segmentation module includes: The sorting unit sorts each user group according to its corresponding value assessment score for each feature dimension, thus obtaining a user group sequence. The sequence splitting unit splits the user group sequence into a corresponding number of user group subsequences according to a set number of layers and sequence positions; The dividing line determination unit finds the coordinate range of the first and last two user groups in each user group subsequence, and uses the feature coordinates of these two user groups as two interval dividing lines. The interval division unit divides all user groups into multiple intervals based on all interval dividing lines, according to each single feature dimension.

[0079] Based on the same inventive concept, in an optional embodiment, the value assessment score increases or decreases along the coordinate axis on the corresponding feature dimension coordinate. The segmentation module includes: The coordinate lookup unit searches for the coordinate range of all user coordinates on the corresponding coordinate axis for each feature dimension. The partitioning unit divides the coordinate range into multiple partitions according to a set number of layers, and all user groups are divided into multiple intervals on each single feature dimension.

[0080] Based on the same inventive concept, in an optional embodiment, the three feature dimensions in the model space are activity level, payment depth, and interaction depth, respectively.

[0081] Figure 8 This application provides a schematic diagram of the structure of a service operation terminal device 400. The terminal device 400 includes: at least one processor 401 (…). Figure 8 (Only one is shown in the image) a processor, a memory 402, and a computer program 403 stored in the memory 402 and executable on the at least one processor 401. When the processor 401 executes the computer program 403, it implements the steps in the above-described embodiment of the auxiliary treatment method for achieving river rainfall water quality standards.

[0082] The terminal device 400 may be a desktop computer, laptop, handheld computer, or cloud server, etc. This terminal device may include, but is not limited to, a processor 401 and a memory 402. Those skilled in the art will understand that... Figure 8 This is merely an example of terminal device 400 and does not constitute a limitation on terminal device 400. It may include more or fewer components than shown in the figure, or combine certain components, or different components, such as input / output devices, network access devices, etc.

[0083] The processor 401 may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0084] In some embodiments, the memory 402 may be an internal storage unit of the terminal device 400, such as a hard disk or memory of the terminal device 400. In other embodiments, the memory 402 may be an external storage device of the terminal device 400, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the terminal device 400. Furthermore, the memory 402 may include both internal and external storage units of the terminal device 400. The memory 402 is used to store the operating system, applications, bootloader, data, and other programs, such as the program code of the computer program. The memory 402 can also be used to temporarily store data that has been output or will be output.

[0085] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps described in the various method embodiments above.

[0086] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0087] In the above embodiments, the descriptions of each embodiment have different focuses. If a certain embodiment does not describe or record a certain part in detail, please refer to the relevant descriptions of other embodiments.

[0088] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the steps in the above-described method embodiments.

[0089] This application provides a computer program product that, when run on a mobile terminal, enables the mobile terminal to implement the steps described in the above-described method embodiments.

[0090] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying the computer program code to a photographing device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks.

[0091] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0092] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0093] In the embodiments provided in this application, it should be understood that the disclosed apparatus / network devices and methods can be implemented in other ways. For example, the apparatus / network device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0094] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0095] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A business operation method, characterized in that, include: Obtain a user feature dataset, which includes user feature groups of N users. Each user feature group includes K feature dimensions, which include the number of times the user logs in on the advertising platform where the target object is located, the user's login duration, and the user's payment depth for the target object. Based on the user feature dataset, the N users are divided into M user groups; At least one dividing line is determined on any feature dimension, and the at least one dividing line is used to divide the M user groups into at least two feature intervals on any feature dimension, and each feature interval corresponds to a value assessment score. Based on the value assessment score of each user group in the feature interval of each feature dimension, a comprehensive assessment score is determined for each user group. Based on the comprehensive evaluation score of each user group, a business strategy is determined for each user group, and business operations are performed on the M user groups according to the business strategy. Before determining at least one dividing line on any feature dimension, the business operation method further includes: Based on the data of all users in each user group, determine the silhouette coefficient and relative entropy of each user group; The group separation degree of each user group is determined based on the profile coefficient and relative entropy of each user group; Based on the group separation degree of each user group, at least one user is separated from each user group.

2. The business operation method according to claim 1, characterized in that, Based on the value assessment score of each user group within the feature interval of each feature dimension, a comprehensive assessment score is determined for each user group, including: The value assessment score of each user group in each feature interval on each feature dimension is multiplied by the corresponding weight to obtain the weighted score of each feature dimension. For each user group, the sum of all weighted scores is calculated to obtain a comprehensive evaluation score.

3. The business operation method according to claim 1, characterized in that, Based on the comprehensive evaluation score of each user group, a business strategy is determined for each user group, and business operations are executed on the M user groups according to the business strategy, including: If the overall evaluation score of a user group is lower than the first threshold, the business strategy for the user group is determined to be to push low-end vehicle product introduction advertisements, and low-end vehicle product introduction advertisements are pushed to the user group at appropriate times. If the overall evaluation score of a user group is lower than the second threshold but not lower than the first threshold, the business strategy for the user group is determined to be to push information about car company test drive activities, and the information about car company test drive activities is pushed to the user group at an appropriate time. If the overall evaluation score of a user group is lower than the third threshold but not lower than the second threshold, the business strategy for the user group is determined to be to push the latest promotional activities and high-end car information, and the latest promotional activities and high-end car information will be pushed to the user group at appropriate times.

4. The business operation method according to claim 1, characterized in that, After separating at least one user from each user group based on the group separation degree of each user group, the business operation method further includes: For each separated user, based on their user data and the user data of at least one user in each user group, the feature distance between the user and each user group is obtained; Based on the characteristic distance between the separated users and each user group, the separated users are reclassified into one of the user groups.

5. The business operation method according to claim 2, characterized in that, The determination of the group separation degree of each user group based on the silhouette coefficient and relative entropy of each user group includes: Calculate the product of the first weight and the silhouette coefficient, and calculate the product of the second weight and the relative entropy; The sum of the two products is taken as the group separation degree of the user group.

6. The business operation method according to claim 1, characterized in that, The value assessment score increases or decreases along the coordinate axis of the corresponding feature dimension coordinate; Determining at least one dividing line on any feature dimension includes: For each feature dimension, find the coordinate range of all user coordinates on the corresponding coordinate axis; Based on the set number of layers, the coordinate range is divided into multiple partitions, and the dividing line between each pair of adjacent partitions is obtained.

7. A business operation device, characterized in that, include: The acquisition module acquires a user feature dataset, which includes user feature groups of N users. Each user feature group includes K feature dimensions, which include the number of times the user logs in on the advertising platform where the target object is located, the user's login duration, and the user's payment depth for the target object. The user segmentation module divides the N users into M user groups based on the user feature dataset. The dividing line determination module determines at least one dividing line on any feature dimension. The at least one dividing line is used to divide the M user groups into at least two feature intervals on any feature dimension. Each feature interval corresponds to a value assessment score. The evaluation module determines the comprehensive evaluation score for each user group based on the value evaluation score of each user group in the feature interval of each feature dimension. The business operation module determines the business strategy for each user group based on the comprehensive evaluation score of each user group, and executes business operations on the M user groups according to the business strategy. The contour coefficient and relative entropy determination module determines the contour coefficient and relative entropy of each user group based on the data of all users in each user group. The separation degree determination module determines the group separation degree of each user group based on the profile coefficient and relative entropy of each user group; The user separation module separates at least one user from each user group based on the group separation degree of each user group.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images