Method and device for putting into operation, electronic equipment and storage medium

Abstract

Embodiments of the present application provide a method and device for putting forward a conversion-boosting activity, electronic equipment and a storage medium, and belong to the technical field of data analysis. Embodiments of the present application filter key user features and key strategy features based on historical user data and historical strategy data. A user strategy binary tree is constructed based on the key user features and the key strategy features, a comprehensive score of each leaf node of the user strategy binary tree is calculated, and the leaf node with the smallest comprehensive score is determined as an optimal leaf node. User-related nodes and strategy-related nodes on a link from a root node of the user strategy binary tree to the optimal leaf node are obtained, a target user is determined according to the user-related nodes, and a conversion-boosting activity is determined according to the strategy-related nodes. The conversion-boosting activity is put forward to a user end of the target user. The corresponding key features can be accurately obtained, and the accuracy and speed of finding the target user and specifying the activity strategy are greatly improved.

Description

Technical Field

[0001] This application relates to the field of data analysis technology, and in particular to a method, apparatus, electronic device, and storage medium for launching conversion promotion activities. Background Technology

[0002] To facilitate data management, bank users with assets exceeding 10,000 yuan can be categorized as "10,000-yuan households," while those whose assets just surpass 10,000 yuan this month can be called "new 10,000-yuan households." Banks need to target these high-potential "new 10,000-yuan households" with promotional activities to increase their numbers. This requires developing specific strategies for these activities, such as issuing coupons, reward points, and gifts. Currently, selecting target users and developing strategies relies almost entirely on the experience of bank staff, lacking data support, making it difficult to guarantee the effectiveness of these promotional activities. Summary of the Invention

[0003] To address the aforementioned technical problems, embodiments of this application provide a method, apparatus, electronic device, and storage medium for launching conversion promotion activities.

[0004] In a first aspect, embodiments of this application provide a method for launching conversion promotion activities, the method comprising:

[0005] The initial user characteristics and initial strategy characteristics are filtered to obtain the key user characteristics and key strategy characteristics;

[0006] Construct a decision tree model based on the key user characteristics and the key strategy characteristics;

[0007] The decision tree model is trained based on historical user data and historical policy data to obtain a user policy binary tree. Each leaf node of the user policy binary tree consists of positive and negative samples.

[0008] Calculate the comprehensive score of each leaf node based on the positive and negative samples, and determine the leaf node with the smallest comprehensive score as the optimal leaf node;

[0009] Obtain the link from the root node of the user policy binary tree to the optimal leaf node, obtain the user-related nodes on the link, and determine the target user based on the user-related nodes;

[0010] Obtain the strategy-related nodes on the link, and determine conversion promotion activities based on the strategy-related nodes;

[0011] The conversion promotion campaign is delivered to the user's client of the target user.

[0012] In one embodiment, the filtering of initial user characteristics and initial policy characteristics includes:

[0013] An initial prediction model is constructed, and the initial prediction model is trained based on the historical user data and the historical strategy data to obtain a new 10,000-yuan household prediction model; wherein, the historical user data is the user characteristics corresponding to the historical new 10,000-yuan households, and the historical strategy data is the activity strategy that was launched for the historical new 10,000-yuan households.

[0014] The initial user characteristics and the initial strategy characteristics are filtered using the new ten-thousand-yuan household prediction model.

[0015] In one embodiment, calculating the comprehensive score of each leaf node based on the positive and negative samples includes:

[0016] Obtain the number of positive samples and the number of negative samples for each leaf node, calculate the quotient of the number of negative samples and the number of positive samples, and use the quotient as the first score;

[0017] Obtain the positive sample probability and negative sample probability of each leaf node, calculate the product of the positive sample probability and the negative sample probability, and use the product as the second score;

[0018] The first score and the second score are summed to obtain the comprehensive score.

[0019] In one embodiment, obtaining user-related nodes on the link includes:

[0020] Obtain the non-leaf nodes on the link other than the root node and the optimal leaf node, and determine the node content of the non-leaf nodes;

[0021] If the node content includes the key user characteristics, then the non-leaf node is determined to be the user-related node.

[0022] In one embodiment, determining the target user based on the user-related nodes includes:

[0023] Obtain the node content of the user-related nodes, and locate the target user based on the node content.

[0024] In one embodiment, obtaining policy-related nodes on the link includes:

[0025] If the node content includes the key strategy features, then the non-leaf node is determined to be a strategy-related node.

[0026] In one embodiment, determining conversion-promoting activities based on the strategy-related nodes includes:

[0027] Obtain the node content of the strategy-related nodes, find the activity strategy based on the node content, and combine the activity strategies into the conversion promotion activity.

[0028] Secondly, embodiments of this application provide a device for launching conversion promotion activities, the device comprising:

[0029] The filtering module is used to filter the initial user features and initial strategy features to obtain key user features and key strategy features;

[0030] The construction module is used to construct a decision tree model based on the key user features and the key strategy features;

[0031] The training module is used to train the decision tree model based on historical user data and historical policy data to obtain a user policy binary tree, wherein each leaf node of the user policy binary tree consists of positive samples and negative samples.

[0032] The scoring module is used to calculate the comprehensive score of each leaf node based on the positive samples and the negative samples, and to determine the leaf node with the smallest comprehensive score as the optimal leaf node.

[0033] The acquisition module is used to acquire the link from the root node of the user policy binary tree to the optimal leaf node, acquire the user-related nodes on the link, and determine the target user based on the user-related nodes.

[0034] The determination module is used to acquire strategy-related nodes on the link and determine conversion promotion activities based on the strategy-related nodes;

[0035] The delivery module is used to deliver the conversion promotion activity to the user's client of the target user.

[0036] Thirdly, embodiments of this application provide an electronic device, including a memory and a processor, wherein the memory is used to store a computer program, and the computer program executes the method for launching conversion promotion activities provided in the first aspect when the processor is running.

[0037] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when run on a processor, executes the method for launching conversion promotion activities provided in the first aspect.

[0038] The conversion promotion method provided in this application involves screening initial user features and initial strategy features to obtain key user features and key strategy features; constructing a decision tree model based on the key user features and key strategy features; training the decision tree model based on historical user data and historical strategy data to obtain a user strategy binary tree, where each leaf node of the user strategy binary tree consists of positive and negative samples; calculating the comprehensive score of each leaf node based on the positive and negative samples, and determining the leaf node with the smallest comprehensive score as the optimal leaf node; obtaining the link from the root node of the user strategy binary tree to the optimal leaf node, obtaining user-related nodes on the link, and determining the target user based on the user-related nodes; obtaining strategy-related nodes on the link, and determining the conversion promotion activity based on the strategy-related nodes; and delivering the conversion promotion activity to the user's end. This embodiment of the application, by screening key features obtained from multi-dimensional historical data, has the characteristics of accuracy and real-time performance, while reducing data noise and model complexity. By constructing a decision tree based on key features and obtaining the link where the optimal leaf node is located based on the decision tree, the corresponding key features can be accurately obtained, which greatly improves the accuracy and speed of finding target users and specifying activity strategies. Attached Figure Description

[0039] To more clearly illustrate the technical solutions of this application, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of this application and should not be considered as a limitation on the scope of protection of this application. In the various drawings, similar components are numbered similarly.

[0040] Figure 1 A flowchart illustrating a method for launching conversion promotion activities according to an embodiment of this application is shown.

[0041] Figure 2 This illustration shows a structural diagram of a user policy binary tree provided in an embodiment of this application;

[0042] Figure 3 This illustration shows another structural diagram of the user policy binary tree provided in an embodiment of this application;

[0043] Figure 4 A schematic diagram of the delivery device for promoting conversion activities provided in an embodiment of this application is shown;

[0044] Figure 5 A schematic diagram of the structure of an electronic device provided in an embodiment of this application is shown. Detailed Implementation

[0045] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0046] The components of the embodiments of this application described and illustrated in the accompanying drawings can be arranged and designed in a variety of different configurations. Therefore, the following detailed description of the embodiments of this application provided in the drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0047] In the following, the terms “comprising,” “having,” and their cognates, which may be used in various embodiments of this application, are intended only to indicate a particular feature, number, step, operation, element, component, or combination thereof, and should not be construed as excluding, firstly, the presence of one or more other features, numbers, steps, operations, elements, components, or combinations thereof, or adding the possibility of one or more features, numbers, steps, operations, elements, components, or combinations thereof.

[0048] Furthermore, the terms "first," "second," and "third" are used only to distinguish descriptions and should not be interpreted as indicating or implying relative importance.

[0049] Unless otherwise specified, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of this application pertain. Terms (such as those defined in commonly used dictionaries) shall be interpreted as having the same meaning as in their contextual meaning in the relevant technical field and shall not be construed as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of this application.

[0050] Example 1

[0051] In the fintech sector, banks need to manage users' assets. Users with assets exceeding 10,000 yuan are called "10,000-yuan households," while users whose assets just exceeded 10,000 yuan this month are called "new 10,000-yuan households."

[0052] To precisely target and convert users with the potential to become new millionaires, user segmentation is necessary. Current technology first identifies the target audience based on user attribute tags, then develops specific campaign strategies using tags related to coupons, red envelopes, points, interest, cashback, gifts, and memberships. However, because this strategy generation method relies entirely on experience and lacks corresponding data support, the effectiveness of the selected campaigns is often difficult to guarantee.

[0053] Based on this, the present disclosure provides a method for launching conversion promotion activities based on a tree model.

[0054] For details, see Figure 1 Methods for launching conversion promotion activities include:

[0055] Step S110: Filter the initial user features and initial strategy features to obtain key user features and key strategy features;

[0056] First, it is necessary to construct initial user characteristics and initial strategy characteristics. Specifically, these can be organized into multi-dimensional initial user characteristics and initial strategy characteristics. Initial user characteristics include various user attributes such as gender, age, and historical behavior; initial strategy characteristics include activity types such as coupons, red envelopes, gifts, and memberships, as well as the strength of the activity discounts, such as coupon amounts.

[0057] In one embodiment, an initial prediction model is constructed, and the initial prediction model is trained based on the historical user data and the historical strategy data to obtain a new ten-thousand-yuan household prediction model; wherein, the historical user data are the user characteristics corresponding to historical new ten-thousand-yuan households, and the historical strategy data are the activity strategies deployed to the historical new ten-thousand-yuan households; the new ten-thousand-yuan household prediction model is used to filter the initial user characteristics and the initial strategy characteristics.

[0058] The initial prediction model is built based on the XGBoost algorithm. The model features are user features and policy features. The training samples are historical data of users participating in promotional activities over a period of time. The model label is whether a user becomes a new high-net-worth individual after being targeted by the promotion. The trained XGBoost model can directly obtain importance scores for user features and policy features. User features and policy features with high scores are selected as the key user features and key policy features for the next step of building the decision tree.

[0059] Step S120: Construct a decision tree model based on the key user features and the key strategy features; First, construct a decision tree based on the key user features and the key strategy features so that the decision tree can be trained in the next step.

[0060] Step S130: Train the decision tree model based on historical user data and historical policy data to obtain a user policy binary tree, where each leaf node of the user policy binary tree consists of positive and negative samples; for example... Figure 2 As shown, Figure 2This is a schematic diagram of a user policy binary tree provided in an embodiment of this application. The bottom layer of the binary tree consists of leaf nodes. Each non-leaf node represents the value of a user or policy feature. gini represents the Gini coefficient of the node, pos represents the number of positive samples in the node, and neg represents the number of negative samples in the node.

[0061] The training decision tree model uses user and policy features as its features, and the training samples are historical data of users participating in promotional activities over a period of time. The model label indicates whether a user becomes a new "ten-thousand-yuan household" after being targeted by the promotion. After training, a binary tree is obtained, named the "User-Policy Binary Tree." This binary tree has the following characteristics: each non-leaf node corresponds to a value judgment of a user or policy feature on a certain attribute, and this value judgment corresponds to a selection rule for the user or policy on a certain dimension. Each leaf node consists of positive and negative samples, where positive samples represent becoming a new "ten-thousand-yuan household."

[0062] Step S140: Calculate the comprehensive score of each leaf node based on the positive and negative samples, and determine the leaf node with the smallest comprehensive score as the optimal leaf node; for example... Figure 3 As shown, Figure 3 This is a schematic diagram of the link structure where the optimal leaf node of the user policy binary tree provided in this application embodiment is located.

[0063] The larger the positive sample size and the higher the proportion, the better it meets the requirements. Therefore, it is necessary to select the leaf node with the largest positive sample size and the highest proportion.

[0064] In one embodiment, calculating the comprehensive score of each leaf node based on the positive and negative samples includes:

[0065] Obtain the number of positive samples and the number of negative samples for each leaf node, calculate the quotient of the number of negative samples and the number of positive samples, and use the quotient as the first score; obtain the positive sample probability and the negative sample probability for each leaf node, calculate the product of the positive sample probability and the negative sample probability, and use the product as the second score; sum the first score and the second score to obtain the comprehensive score.

[0066] The smaller the first score, the greater the proportion of positive samples. The second score is the Gini coefficient of each leaf node, where the Gini coefficient equals the probability of a positive sample multiplied by the probability of a negative sample. A smaller Gini coefficient indicates a more homogeneous classification for that leaf node. Therefore, the optimal leaf node can be selected using the following criteria: positive samples should be significantly greater than negative samples, meaning the first score should be as small as possible; the Gini coefficient should be as small as possible, as a smaller Gini coefficient indicates a more homogeneous sample.

[0067] like Figure 3As shown, leaf node 1 meets the requirements of a small Gini coefficient and a much larger number of positive samples than negative samples, therefore it can be defined as the optimal leaf node. The attribute values ​​of user characteristics and policies on the link nodes from the root node to leaf node 1 are the selection criteria for the optimal user and activity policy. Based on these values, we can generate target users and activity policies respectively.

[0068] At the same time, the length of the conversion link can be controlled by controlling the depth of the decision tree, which can also control the complexity of the activity strategy to a certain extent, making it easier to adjust the activity strategy in a flexible manner.

[0069] Step S150: Obtain the link from the root node of the user policy binary tree to the optimal leaf node, obtain the user-related nodes on the link, and determine the target user based on the user-related nodes.

[0070] In one embodiment, obtaining user-related nodes on the link includes:

[0071] Obtain the non-leaf nodes on the link other than the root node and the optimal leaf node, and determine the node content of the non-leaf nodes; if the node content includes the key user features, then determine the non-leaf node as the user-related node.

[0072] In one embodiment, determining the target user based on the user-related nodes includes: obtaining the node content of the user-related nodes and searching for the target user based on the node content.

[0073] Step S160: Obtain the strategy-related nodes on the link, and determine the conversion promotion activities based on the strategy-related nodes;

[0074] In one embodiment, obtaining the policy-related node on the link includes: if the node content includes the key policy feature, then determining the non-leaf node as the policy-related node.

[0075] In one embodiment, determining the conversion promotion activity based on the strategy-related nodes includes: obtaining the node content of the strategy-related nodes, finding the activity strategy based on the node content, and combining the activity strategy into the conversion promotion activity.

[0076] Step S170: Deploy the conversion promotion activity to the user terminal of the target user.

[0077] Once the target users have been identified, selected conversion-boosting campaigns can be launched on their user interfaces.

[0078] It is important to note that the focus of this application's embodiments is on the process of selecting the target audience from the entire population based on decision trees and selecting conversion-driving activities from the activity strategy, rather than the content of the campaign or user attribute tags. This method has good scalability and can also be applied to other activities or other information delivery based on audience segmentation.

[0079] The conversion promotion campaign method provided in this embodiment selects key user and key strategy features based on historical user and strategy data. It features accuracy and real-time performance while reducing data noise and model complexity. A user strategy binary tree is constructed based on the key user and strategy features. The comprehensive score of each leaf node in the user strategy binary tree is calculated, and the leaf node with the lowest comprehensive score is determined as the optimal leaf node. User-related nodes and strategy-related nodes on the link from the root node of the user strategy binary tree to the optimal leaf node are obtained. Target users are determined based on user-related nodes, and conversion promotion campaigns are determined based on strategy-related nodes. The conversion promotion campaign is then delivered to the user's end. This method can accurately obtain the corresponding key features, significantly improving the accuracy and speed of finding target users and specifying campaign strategies.

[0080] Example 2

[0081] Furthermore, embodiments of this disclosure provide a delivery device for promoting conversion activities.

[0082] Specifically, such as Figure 4 As shown, the delivery device 400 for promoting conversion activities includes:

[0083] The filtering module 410 is used to filter the initial user features and initial strategy features to obtain key user features and key strategy features;

[0084] Module 420 is used to construct a decision tree model based on the key user features and the key strategy features;

[0085] Training module 430 is used to train the decision tree model based on historical user data and historical policy data to obtain a user policy binary tree, wherein each leaf node of the user policy binary tree consists of positive samples and negative samples.

[0086] The scoring module 440 is used to calculate the comprehensive score of each leaf node based on the positive samples and the negative samples, and to determine the leaf node with the smallest comprehensive score as the optimal leaf node.

[0087] The acquisition module 450 is used to acquire the link from the root node of the user policy binary tree to the optimal leaf node, acquire the user-related nodes on the link, and determine the target user based on the user-related nodes.

[0088] The determination module 460 is used to obtain policy-related nodes on the link and determine conversion promotion activities based on the policy-related nodes.

[0089] The delivery module 470 is used to deliver the conversion promotion activity to the user terminal of the target user.

[0090] The filtering module 410 is also used for:

[0091] An initial prediction model is constructed, and the initial prediction model is trained based on the historical user data and the historical strategy data to obtain a new 10,000-yuan household prediction model; wherein, the historical user data is the user characteristics corresponding to the historical new 10,000-yuan households, and the historical strategy data is the activity strategy that was launched for the historical new 10,000-yuan households.

[0092] The initial user characteristics and the initial strategy characteristics are filtered using the new ten-thousand-yuan household prediction model.

[0093] The scoring module 440 is also used for:

[0094] Obtain the number of positive samples and the number of negative samples for each leaf node, calculate the quotient of the number of negative samples and the number of positive samples, and use the quotient as the first score;

[0095] Obtain the positive sample probability and negative sample probability of each leaf node, calculate the product of the positive sample probability and the negative sample probability, and use the product as the second score;

[0096] The first score and the second score are summed to obtain the comprehensive score.

[0097] The acquisition module 450 is further configured to:

[0098] Obtain the non-leaf nodes on the link other than the root node and the optimal leaf node, and determine the node content of the non-leaf nodes;

[0099] If the node content includes the key user characteristics, then the non-leaf node is determined to be the user-related node.

[0100] The acquisition module 450 is further configured to:

[0101] Obtain the node content of the user-related nodes, and locate the target user based on the node content.

[0102] The determining module 460 is further configured to:

[0103] If the node content includes the key strategy features, then the non-leaf node is determined to be a strategy-related node.

[0104] The determining module 460 is further configured to:

[0105] Obtain the node content of the strategy-related nodes, find the activity strategy based on the node content, and combine the activity strategies into the conversion promotion activity.

[0106] The conversion promotion activity delivery device 400 provided in this embodiment can implement the conversion promotion activity delivery method provided in Embodiment 1. To avoid repetition, it will not be described again here.

[0107] The conversion promotion campaign delivery device provided in this embodiment filters key user features and key strategy features based on historical user data and historical strategy data. It features accuracy and real-time performance while reducing data noise and model complexity. A user strategy binary tree is constructed based on the key user features and key strategy features. The comprehensive score of each leaf node in the user strategy binary tree is calculated, and the leaf node with the smallest comprehensive score is determined as the optimal leaf node. User-related nodes and strategy-related nodes on the link from the root node of the user strategy binary tree to the optimal leaf node are obtained. Target users are determined based on user-related nodes, and conversion promotion activities are determined based on strategy-related nodes. The conversion promotion activities are then delivered to the user terminals of the target users. This device can accurately obtain the corresponding key features, significantly improving the accuracy and speed of finding target users and specifying activity strategies.

[0108] Example 3

[0109] Furthermore, this disclosure provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the computer program, when run on the processor, executes the conversion promotion activity delivery method provided in Embodiment 1.

[0110] For details, see Figure 5 The electronic device 500 includes: a transceiver 501, a bus interface and processor 502, and a memory 503. Figure 5 In this context, the bus architecture can include any number of interconnected buses and bridges, specifically linking various circuits together, represented by one or more processors (processor 502) and memory (memory 503). The bus architecture can also link together various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. The bus interface provides an interface. The transceiver 501 can be multiple elements, including transmitters and receivers, providing a unit for communicating with various other devices over a transmission medium. The processor 502 is responsible for managing the bus architecture and general processing, and the memory 503 can store data used by the processor 502 during operation.

[0111] The electronic device 500 provided in this embodiment of the invention can implement the method for launching conversion promotion activities provided in Embodiment 1. To avoid repetition, it will not be described again here.

[0112] The electronic device provided in this embodiment filters key user features and key policy features based on historical user data and historical policy data, featuring accuracy and real-time performance while reducing data noise and model complexity. A user policy binary tree is constructed based on the key user features and key policy features. The comprehensive score of each leaf node in the user policy binary tree is calculated, and the leaf node with the smallest comprehensive score is determined as the optimal leaf node. User-related nodes and policy-related nodes on the link from the root node of the user policy binary tree to the optimal leaf node are obtained. Target users are determined based on user-related nodes, and conversion promotion activities are determined based on policy-related nodes. The conversion promotion activities are then delivered to the user terminals of the target users. This allows for precise acquisition of corresponding key features, significantly improving the accuracy and speed of finding target users and specifying activity strategies.

[0113] Example 4

[0114] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the method for launching conversion promotion activities provided in Embodiment 1.

[0115] In this embodiment, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0116] The computer-readable storage medium provided in this embodiment can implement the method for launching conversion promotion activities provided in Embodiment 1. To avoid repetition, it will not be described again here.

[0117] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal that includes that element.

[0118] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0119] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A method for launching conversion promotion activities, characterized in that, The method comprises: screening initial user features and initial strategy features to obtain key user features and key strategy features; the initial user features are various attributes of a user, including gender, age, and historical behavior; the initial strategy features are activity types; constructing a decision tree model according to the key user features and the key strategy features; training the decision tree model based on historical user data and historical strategy data to obtain a user strategy binary tree, the lowest layer of the user strategy binary tree being a leaf node, each non-leaf node representing a value of a user or strategy feature, and each leaf node of the user strategy binary tree being composed of positive samples and negative samples; calculating a comprehensive score of each leaf node according to the positive samples and the negative samples, and determining a leaf node with the smallest comprehensive score as an optimal leaf node; obtaining a link from a root node to the optimal leaf node of the user strategy binary tree, obtaining user-related nodes on the link, and determining a target user according to the user-related nodes; obtaining strategy-related nodes on the link, and determining a conversion-promoting activity according to the strategy-related nodes; delivering the conversion-promoting activity to a user terminal of the target user; the step of determining a conversion-promoting activity according to the strategy-related nodes comprises: obtaining node content of the strategy-related nodes, searching for activity strategies according to the node content, and combining the activity strategies into the conversion-promoting activity; the step of screening initial user features and initial strategy features comprises: constructing an initial estimation model, training the initial estimation model based on the historical user data and the historical strategy data to obtain a new ten-thousand-yuan customer estimation model; wherein the historical user data are user features corresponding to historical new ten-thousand-yuan customers, and the historical strategy data are activity strategies to which the historical new ten-thousand-yuan customers are delivered; screening the initial user features and the initial strategy features using the new ten-thousand-yuan customer estimation model; the initial estimation model is an estimation model constructed based on an xgboost algorithm, wherein model features are user features and strategy features, training samples are historical data of users participating in activity delivery in the past period of time, and model labels are whether a user becomes a new ten-thousand-yuan customer after being delivered with an activity; the step of calculating a comprehensive score of each leaf node according to the positive samples and the negative samples comprises: obtaining a number of positive samples and a number of negative samples of each leaf node, calculating a quotient value of the number of negative samples and the number of positive samples, and taking the quotient value as a first score; obtaining a positive sample probability and a negative sample probability of each leaf node, calculating a product of the positive sample probability and the negative sample probability, and taking the product as a second score; summing the first score and the second score to obtain the comprehensive score.

2. The method of claim 1, wherein the conversion campaign is launched by, the step of obtaining user-related nodes on the link comprises: obtaining non-leaf nodes on the link except the root node and the optimal leaf node, and determining node content of the non-leaf nodes; if the node content includes the key user features, determining that the non-leaf node is the user-related node.

3. The method of claim 2, wherein the conversion campaign is launched by, the step of determining a target user according to the user-related nodes comprises: Obtaining node content of the user-related node, and searching for the target user according to the node content.

4. The method of claim 2, wherein the conversion campaign is launched by, The obtaining of the policy-related node on the link comprises: If the node content comprises the key policy feature, determining that the non-leaf node is the policy-related node.

5. A device for dispensing a promotional activity, characterized in that The device comprises: The screening module is configured to screen initial user features and initial policy features to obtain key user features and key policy features; the initial user features are various attributes of a user, and the attributes comprise gender, age, and historical behavior; and the initial policy feature is an activity type; The construction module is configured to construct a decision tree model according to the key user features and the key policy features; The training module is configured to train the decision tree model based on historical user data and historical policy data to obtain a user policy binary tree, the lowest layer of the user policy binary tree being a leaf node, each non-leaf node representing a value of a user or a policy feature, and each leaf node of the user policy binary tree being composed of positive samples and negative samples; The scoring module is configured to calculate a comprehensive score of each leaf node according to the positive samples and the negative samples, and determine a leaf node with the smallest comprehensive score as an optimal leaf node; The obtaining module is configured to obtain a link from a root node of the user policy binary tree to the optimal leaf node, obtain a user-related node on the link, and determine a target user according to the user-related node; The determining module is configured to obtain a policy-related node on the link, and determine a conversion-promoting activity according to the policy-related node; The delivery module is configured to deliver the conversion-promoting activity to a user end of the target user. The determining of the conversion-promoting activity according to the policy-related node comprises: Obtaining node content of the policy-related node, searching for an activity policy according to the node content, and combining the activity policy into the conversion-promoting activity; The screening of the initial user features and the initial policy features comprises: Constructing an initial estimation model, training the initial estimation model based on the historical user data and the historical policy data to obtain a new ten-thousand-yuan customer estimation model; wherein the historical user data is user features corresponding to historical new ten-thousand-yuan customers, and the historical policy data is an activity policy delivered to the historical new ten-thousand-yuan customers; Screening the initial user features and the initial policy features by using the new ten-thousand-yuan customer estimation model; The initial estimation model is an estimation model constructed based on an xgboost algorithm, wherein model features are user features and policy features, training samples are historical data of users participating in activity delivery in the past period of time, and a model label is whether a user becomes a new ten-thousand-yuan customer after being delivered with an activity; The calculation of the comprehensive score of each leaf node according to the positive samples and the negative samples comprises: Obtaining a number of positive samples and a number of negative samples of each leaf node, calculating a quotient value of the number of negative samples and the number of positive samples, and taking the quotient value as a first score; Obtaining a positive sample probability and a negative sample probability of each leaf node, calculating a product of the positive sample probability and the negative sample probability, and taking the product as a second score; Summing the first score and the second score results in the composite score.

6. An electronic device, comprising: The computer program product comprises a memory and a processor, the memory stores the computer program, and the computer program performs the method for putting the conversion-boosting activity according to any one of claims 1 to 4 when the processor runs.

7. A computer readable storage medium characterized in that, The computer program product comprises a memory and a processor, the memory stores the computer program, and the computer program performs the method for putting the conversion-boosting activity according to any one of claims 1 to 4 when the processor runs.

Images