CN116433339B

CN116433339B - Order data processing method, device, equipment and storage medium

Info

Publication number: CN116433339B
Application number: CN202310705913.2A
Authority: CN
Inventors: 周志胜
Original assignee: All House Premium Technology Shenzhen Co ltd
Current assignee: All House Premium Technology Shenzhen Co ltd
Priority date: 2023-06-15
Filing date: 2023-06-15
Publication date: 2023-08-18
Anticipated expiration: 2043-06-15
Also published as: CN116433339A

Abstract

The invention relates to the field of data processing, and discloses a method, a device, equipment and a storage medium for processing order data, which are used for improving the accuracy of order data processing. The method comprises the following steps: acquiring historical household order data of a target user based on a household service system, and classifying the order data to obtain a normally completed order and an abnormally unfinished order; extracting order features of a normally completed order and creating a first feature distribution space; performing order feature analysis on the abnormal unfinished order and creating a second feature distribution space; performing cross interest analysis and feature search on the first feature distribution space to obtain a third feature distribution space; according to the second characteristic distribution space, characteristic distribution screening and eliminating are carried out on the third characteristic distribution space, and a target characteristic distribution space is generated; and generating a target pushing list according to the target feature distribution space, and pushing home services to the target pushing list through the home service system.

Description

Order data processing method, device, equipment and storage medium

Technical Field

The present invention relates to the field of data processing, and in particular, to a method, an apparatus, a device, and a storage medium for processing order data.

Background

In recent years, more and more home electronics providers need better personalized home services for customers, and more in-depth knowledge of customer preferences and behaviors. In this embodiment, with the continuous development of machine learning algorithms and data processing technologies, research on a method for processing household order data is a popular field.

The deficiencies in the prior art include the following: the quality of the household order data may not be entirely accurate or misleading, and therefore it is desirable to ensure data quality and accuracy. Current data analysis methods may not be able to handle the diversity and high dimensionality of household order data, requiring more advanced algorithms and techniques to handle such data. Some existing recommendation systems may lack a degree of personalization or accuracy, requiring a more accurate and intelligent recommendation system to meet customer needs, i.e., low accuracy of existing solutions.

Disclosure of Invention

The invention provides a method, a device, equipment and a storage medium for processing order data, which are used for improving the accuracy of order data processing.

The first aspect of the present invention provides a method for processing order data, where the method for processing order data includes:

Acquiring historical household order data corresponding to a target user based on a preset household service system, and classifying the historical household order data to obtain a normally completed order and an abnormally unfinished order;

extracting order features of the normally completed order to obtain a plurality of first order feature labels and a plurality of first home attribute feature labels, and creating a first feature distribution space according to the plurality of first order feature labels and the plurality of first home attribute feature labels;

performing order feature analysis on the abnormal unfinished order to obtain a plurality of second order feature labels and a plurality of second home attribute feature labels, and creating a second feature distribution space according to the plurality of second order feature labels and the plurality of second home attribute feature labels;

performing cross interest analysis and feature search on the first feature distribution space to obtain a third feature distribution space;

according to the second characteristic distribution space, characteristic distribution screening and eliminating are carried out on the third characteristic distribution space, and a target characteristic distribution space is generated;

and generating a target pushing list of the target user according to the target feature distribution space, and pushing home service to the target pushing list through the home service system.

With reference to the first aspect, in a first implementation manner of the first aspect of the present invention, the acquiring, by using a preset-based home service system, historical home order data corresponding to a target user, and classifying the historical home order data to obtain a normally completed order and an abnormally unfinished order includes:

acquiring user information of a target user, and inquiring orders in a target time period from a preset home service system according to the user information to obtain historical home order data;

performing data deduplication and data format standardization processing on the historical household order data to obtain standard household order data;

and carrying out order data classification on the standard household order data according to preset order node information to obtain a normally completed order and an abnormally unfinished order.

With reference to the first aspect, in a second implementation manner of the first aspect of the present invention, the extracting order features of the normally completed order to obtain a plurality of first order feature labels and a plurality of first home attribute feature labels, and creating a first feature distribution space according to the plurality of first order feature labels and the plurality of first home attribute feature labels includes:

Extracting order features of the normally completed order to obtain a plurality of first order feature labels, wherein the plurality of first order feature labels comprise: purchase frequency, order amount, order type, and order time;

extracting home attribute characteristics of the normally completed order to obtain a plurality of first home attribute characteristic labels, wherein the plurality of first home attribute characteristic labels comprise: household product type, color and material;

performing feature label clustering on the plurality of first order feature labels to obtain a plurality of first feature clustering points, and performing feature label clustering on the plurality of first home attribute feature labels to obtain a plurality of second feature clustering points;

and performing feature distribution space mapping on the plurality of first feature cluster points and the plurality of second feature cluster points to generate a first feature distribution space.

With reference to the first aspect, in a third implementation manner of the first aspect of the present invention, performing order feature analysis on the abnormal unfinished order to obtain a plurality of second order feature labels and a plurality of second home attribute feature labels, and creating a second feature distribution space according to the plurality of second order feature labels and the plurality of second home attribute feature labels, where the method includes:

Extracting order features of the abnormal unfinished order to obtain a plurality of second order feature labels, wherein the plurality of second order feature labels comprise: order transportation state, order cancellation condition and order reminding content;

extracting home attribute features of the abnormal unfinished order to obtain a plurality of second home attribute feature labels, wherein the plurality of second home attribute feature labels comprise: home product size, quality, and return time;

performing feature label clustering on the plurality of second order feature labels to obtain a plurality of third feature clustering points, and performing feature label clustering on the plurality of second home attribute feature labels to obtain a plurality of fourth feature clustering points;

and performing feature distribution space mapping on the plurality of third feature cluster points and the plurality of fourth feature cluster points to generate a second feature distribution space.

With reference to the first aspect, in a fourth implementation manner of the first aspect of the present invention, the performing cross-interest analysis and feature search on the first feature distribution space to obtain a third feature distribution space includes:

respectively constructing association rules between each first order feature tag and household product types to obtain a plurality of product association rules;

Determining a corresponding plurality of first feature points based on the first feature distribution space, wherein the plurality of first feature points comprises: the order volume is maximum, the profit is highest, the sales amount is highest, and the recommended access volume is the most;

according to the product association rules, cross interest analysis is carried out on each first feature point, and cross association data of each first feature point is obtained;

and generating a third feature distribution space according to the first feature distribution space and the cross-correlation data of each first feature point.

With reference to the first aspect, in a fifth implementation manner of the first aspect of the present invention, the performing feature distribution screening and rejection on the third feature distribution space according to the second feature distribution space, to generate a target feature distribution space includes:

identifying overlapping points and difference points of the second characteristic distribution space and the third characteristic distribution space to obtain a target overlapping point and a target difference point;

according to the target overlapping points, overlapping points are removed from the third feature distribution space, and an initial feature distribution space is obtained;

searching adjacent points for target difference points in the initial feature distribution space to obtain first-level adjacent points;

After traversing the target difference points in sequence, traversing the second-level adjacent points of the first-level adjacent points in sequence until adjacent points which do not meet the preset requirement in the initial feature distribution space are not traversed, and obtaining an original feature distribution space;

and carrying out feature point distribution integration on the original feature distribution space to obtain a target feature distribution space.

With reference to the first aspect, in a sixth implementation manner of the first aspect of the present invention, the generating, according to the target feature distribution space, a target push list of the target user, and performing, by the home service system, home service push on the target push list includes:

acquiring a plurality of second feature points in the target feature distribution space, and calculating the association weight coefficient of the plurality of second feature points and the target user to obtain the association weight coefficient of each second feature point;

sorting the associated products of the target user according to the associated weight coefficient of each second feature point to obtain a target pushing list of the target user;

and matching a target pushing mode corresponding to the target pushing list based on the home service system, and pushing home service according to the target pushing mode and the target pushing list.

A second aspect of the present invention provides an order data processing apparatus, including:

the acquisition module is used for acquiring historical household order data corresponding to a target user based on a preset household service system, and classifying the historical household order data to obtain a normally completed order and an abnormally unfinished order;

the first creating module is used for extracting order features of the normally completed order to obtain a plurality of first order feature labels and a plurality of first home attribute feature labels, and creating a first feature distribution space according to the plurality of first order feature labels and the plurality of first home attribute feature labels;

the second creating module is used for carrying out order feature analysis on the abnormal unfinished order to obtain a plurality of second order feature labels and a plurality of second home attribute feature labels, and creating a second feature distribution space according to the plurality of second order feature labels and the plurality of second home attribute feature labels;

the analysis module is used for carrying out cross interesting analysis and feature search on the first feature distribution space to obtain a third feature distribution space;

The screening module is used for screening and eliminating the characteristic distribution of the third characteristic distribution space according to the second characteristic distribution space to generate a target characteristic distribution space;

and the pushing module is used for generating a target pushing list of the target user according to the target feature distribution space and pushing home services to the target pushing list through the home service system.

A third aspect of the present invention provides an order data processing apparatus, comprising: a memory and at least one processor, the memory having instructions stored therein; the at least one processor invokes the instructions in the memory to cause the order data processing apparatus to perform the order data processing method described above.

A fourth aspect of the present invention provides a computer readable storage medium having instructions stored therein which, when run on a computer, cause the computer to perform the method of processing order data as described above.

In the technical scheme provided by the invention, order feature extraction is carried out on a normally completed order, and a first feature distribution space is created; performing order feature analysis on the abnormal unfinished order and creating a second feature distribution space; performing cross interest analysis and feature search on the first feature distribution space to obtain a third feature distribution space; according to the second characteristic distribution space, characteristic distribution screening and eliminating are carried out on the third characteristic distribution space, and a target characteristic distribution space is generated; according to the invention, the target feature distribution space is predicted more accurately through a machine learning algorithm and cross interest analysis according to the historical order data of the clients, the client unit price and purchase frequency of the clients can be improved through cross sales and personalized recommendation, the profit rate is indirectly improved, the clients can be better interacted and communicated through the recommendation and the pushing list, more close client relations are established, and the accuracy of order data processing is further improved.

Drawings

FIG. 1 is a diagram illustrating an embodiment of a method for processing order data according to an embodiment of the present invention;

FIG. 2 is a flow chart of creating a first feature distribution space in an embodiment of the invention;

FIG. 3 is a flow chart of cross-interest analysis and feature search in an embodiment of the invention;

FIG. 4 is a flow chart of feature distribution screening and culling in an embodiment of the present invention;

FIG. 5 is a schematic diagram of an embodiment of an apparatus for processing order data according to an embodiment of the present invention;

FIG. 6 is a schematic diagram of an embodiment of an order data processing apparatus according to an embodiment of the present invention.

Detailed Description

The embodiment of the invention provides a method, a device, equipment and a storage medium for processing order data, which are used for improving the accuracy of order data processing. The terms "first," "second," "third," "fourth" and the like in the description and in the claims and in the above drawings, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments described herein may be implemented in other sequences than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus.

For ease of understanding, a specific flow of an embodiment of the present invention is described below, referring to fig. 1, and one embodiment of a method for processing order data in an embodiment of the present invention includes:

s101, acquiring historical household order data corresponding to a target user based on a preset household service system, and classifying the historical household order data to obtain a normally completed order and an abnormally unfinished order;

it will be appreciated that the execution subject of the present invention may be an order data processing device, which is a terminal or a server, and is not limited herein. The embodiment of the invention is described by taking a server as an execution main body as an example.

Specifically, the server acquires user information of a target user, and queries orders in a target time period from a preset home service system according to the user information to obtain historical home order data. And the server performs data deduplication and data format standardization processing on the historical household order data to obtain standard household order data. And the server classifies the order data of the standard household order data according to the preset order node information to obtain a normally completed order and an abnormally unfinished order. For example, the server first obtains user information of the target user, e.g., the user ID of the target user is 123456. The server inquires order data during 2022, 1/3, 4/31 from a preset home service system according to the user ID of the target user. The home service system returns all order data of the target user in the time period, including order numbers, order placing time, commodity information, prices and the like. And the server performs data deduplication and data format standardization processing on the acquired historical household order data. By repeating the operations, the server eliminates duplicate order data, ensuring that each order is calculated only once. In this embodiment, through the standardized processing of the data format, the server unifies the fields, units and naming modes of the order data, so that the order data has the characteristics of consistency and easy processing. Further, according to the preset order node information, the server classifies the order data of the standard household order data. The server classifies orders into normally completed orders and abnormally unfinished orders according to the status of nodes such as order confirmation, shipment, signing and receiving. A normally completed order refers to an order that has records on each node and has completed, while an abnormally incomplete order refers to an order that lacks records or has an abnormal status on a certain node. In this embodiment, the server obtains historical home order data corresponding to the target user based on a preset home service system, and classifies the historical home order data to obtain the results of normally completed orders and abnormally incomplete orders. The household service system can be helped to better know the order situation of the user by the classification, timely find out abnormal situations, take corresponding measures to solve the problems, and improve user experience and service quality.

S102, extracting order features of a normally completed order to obtain a plurality of first order feature labels and a plurality of first home attribute feature labels, and creating a first feature distribution space according to the plurality of first order feature labels and the plurality of first home attribute feature labels;

in particular, a server feature distribution space refers to a space formed by a set of feature interactions, in which each point represents a sample and each dimension represents a feature. Feature distribution space is commonly used in data analysis and machine learning, and can be used to determine classification, clustering, relevance, etc. of data points. For example, for home electronics providers, the server constructs a feature distribution space using features such as order amount, order product type, purchase time, customer region, etc., to determine information such as customer preference, purchase habit, and region distribution. In this feature distribution space, each data point represents an order, and each dimension represents a feature, such as order amount, order product type, purchase time, customer area, etc. By analyzing the feature distribution of these orders, regularity, trends, and preferences between orders can be discovered to better serve and recommend products to customers. The server extracts a plurality of first order feature labels from the normally completed order data, including purchase frequency, order amount, order type, and order time. By analyzing the order data, the server calculates the purchase frequency (e.g., number of purchases per month), the amount of the order, the type of order (online or phone order), and the time of the order (time period of order, date of order, etc.). The server extracts a plurality of first home attribute feature tags from the order data which is normally completed, wherein the first home attribute feature tags comprise types, colors and materials of home products. By analyzing commodity information in the orders, the server acquires the types (such as furniture, household appliances and the like), colors and materials (such as wood, metal and the like) of household products related to each order. And the server clusters the characteristic labels of the plurality of first order characteristic labels and classifies the similar characteristic labels into the same category. For example, clustering of purchase frequency, order amount, and order type may result in different clusters points, representing different clusters of order features. And similarly, carrying out feature tag clustering on the plurality of first home attribute feature tags, and classifying the similar feature tags into the same category. For example, the type, color and material of the household product are clustered, so that different clustering points can be obtained, and different household attribute characteristic clusters are represented. Further, a plurality of first feature cluster points and a plurality of second feature cluster points are mapped to a feature distribution space, creating a first feature distribution space. In the feature distribution space, each feature cluster point represents a cluster category, and the relation between different features is represented by the position and the relative position of the feature cluster point. For example, the server maps the order feature cluster points and the home attribute feature cluster points to two-dimensional or three-dimensional coordinates, thereby forming a first feature distribution space. For example, assume a home electronics platform, wherein the order data includes information such as purchase frequency, order amount, order type, and order time. The server acquires historical household order data of the target user from the household service system. According to the user information and time range of the target user, the server queries the system and obtains order data of the target user in the past year. The server performs data deduplication and format standardization processing on the historical order data, and ensures the accuracy and consistency of the data. Further, the server obtains standard household order data, and subsequent analysis and processing are convenient. And the server extracts order features. For the purchase frequency, the server calculates the number of purchases of the target user in the past year, and finds that the target user purchases the home product 3 times per month on average. And for the order amount, the server counts the order amount of the target user, and finds that the average order amount of the target user is 500 yuan. In addition, the server observes that the order type of the target user is mainly online. While in terms of order time, the server analyzes the target user's time of placement and finds that the target user is prone to placement on weekends, particularly Saturday afternoon. In addition to the order feature, the server also extracts home attribute features. And the server extracts characteristic labels such as the type, the color, the material and the like of the household products purchased by the target user by analyzing commodity information in the order. For example, the target user purchases different types of household products such as furniture, light fixtures, and decorations. In terms of color, the target user tends to purchase a product of the white and gray series. In terms of materials, the target user purchases household products made of different materials such as wood, metal, cloth and the like. The server then performs a cluster analysis on these feature tags. For order feature labels, the server clusters the purchase frequency, order amount, and order type, thereby identifying different user groups. For the home attribute feature labels, the server clusters the types, colors and materials of home products so as to know the preferences of users for different features. Further, the first feature cluster point and the second feature cluster point are mapped to a feature distribution space. For example, in a two-dimensional space, a first feature cluster point may be represented as (x 1, y 1), and a second feature cluster point may be represented as (x 2, y 2). By mapping, the server forms an image in the feature distribution space, which image is composed of the cluster points. The first characteristic clustering point obtained by the server is (3, 8) representing the clustering points of high-frequency purchase, high-amount order and online ordering, and the second characteristic clustering point is (2, 5) representing the clustering points of purchasing furniture, white and wooden materials. In the feature distribution space, the positions and the relative positions of the two clustering points can reflect the relationship between the high-frequency purchase order and the purchase of furniture, white and wooden materials. By creating the first feature distribution space, the server more intuitively knows the association between the order feature and the home attribute. This helps the server identify different types of user groups and their purchasing preferences, thereby optimizing marketing strategies, personalized recommendations, and customized services. In this embodiment, by extracting order features of a normally completed order, obtaining a plurality of first order feature tags and a plurality of first home attribute feature tags, and creating a first feature distribution space according to the feature tags, the server can better understand and analyze purchasing behavior and preferences of a user, thereby providing more accurate personalized services and decision support for a home service system.

S103, carrying out order feature analysis on the abnormal unfinished order to obtain a plurality of second order feature labels and a plurality of second home attribute feature labels, and creating a second feature distribution space according to the plurality of second order feature labels and the plurality of second home attribute feature labels;

it should be noted that, the server obtains abnormal unfinished order data of the target user from a preset home service system. And according to the user information and the screening conditions, the server queries the system and acquires abnormal unfinished order data in the target time period. The server performs order feature extraction on the abnormal incomplete order data. For example, the server extracts the shipping status of the order to see if the current delivery status of the order is in transit, delayed, or otherwise. The server extracts the cancellation condition of the order and judges whether the order is cancelled or in a pending state. In addition, the server extracts the reminder content in the order to see if there is important information that requires attention or processing by the user. In this embodiment, the server performs home attribute feature extraction. And for abnormal unfinished orders, the server extracts characteristic labels such as the size, the quality and the return time of household products. These features may help the server analyze problems in abnormal orders, such as size mismatch, quality problems, or return time delays, etc. And the server performs cluster analysis on the second order feature labels and the second home attribute feature labels. Through clustering, the server classifies orders and household attributes with similar characteristics into the same category to better understand and analyze commonalities and disparities of abnormal orders. For example, the server clusters shipping status, cancellation, and reminder content to identify different types of abnormal orders. Further, the server maps these third feature cluster points and fourth feature cluster points in the feature distribution space, thereby creating a second feature distribution space. This feature distribution space may be used to visualize the feature relationships of the abnormal order, helping the server to better understand and analyze the feature distribution of the abnormal order. For example, assume that a server obtains abnormal incomplete order data of a target user from a home service system. Through order feature extraction, the server finds that the transportation state of some orders is delayed, other orders are cancelled, and other orders contain return reminding contents. In this embodiment, through home attribute feature extraction, the server finds that some orders relate to the problem of size mismatch, other orders relate to the quality problem, and some orders have a return time delay. The server performs cluster analysis on the second order feature labels, and finds that a delayed transportation category, a cancelled order category and the like exist. In this embodiment, cluster analysis is performed on the second home attribute feature tag, and the server may find a size mismatch class and a quality problem class. Further, the server creates a second feature distribution space by mapping these third feature cluster points and fourth feature cluster points in the feature distribution space. In this space, the server clearly sees the relationships between the different types of outstanding orders and their home property characteristics. For example, the server may observe that orders of the delayed shipment category are concentrated in a particular area, while orders of the cancel order category are distributed in another area. Orders of the size mismatch category and the quality problem category may also exhibit different aggregation patterns in the feature distribution space in this embodiment. Such a second feature distribution space provides a valuable visualization tool for the server to further analyze and understand anomalous outstanding orders. By observing and comparing the distribution of different feature labels, the server identifies potential problems and abnormal patterns, thereby taking corresponding measures to improve the order flow, solve the problems, and improve the user satisfaction and business efficiency. In summary, by performing order feature analysis on an abnormally unfinished order and creating a second feature distribution space, the server is able to more fully understand the relationship between order features and home attributes and gain insight therefrom to support subsequent business decisions and improvement measure formulation.

S104, performing cross interest analysis and feature search on the first feature distribution space to obtain a third feature distribution space;

specifically, the server builds association rules for the relation between each first order feature tag and the household product type. Association rule analysis is a method for discovering the correlation and dependency between features. By analyzing the data between the order feature labels and the household product types, the server determines association rules between them, such as the relationship rules between the frequency of purchase and the household product types. These product association rules may help the server better understand the links between order features and home attributes. The server determines a corresponding plurality of first feature points based on the first feature distribution space. The first feature point is a data point having a significant feature in the feature distribution space, such as a point with the largest order volume, highest profit, highest sales, or the largest recommended visit. By analyzing and calculating the first feature distribution space, the server determines feature points that represent key features in the order data. And the server performs cross interest analysis on each first characteristic point according to the product association rule. Cross-interest analysis is a method for studying the association and interaction between two or more features. The server will apply the feature labels in the product association rules and analyze each of the first feature points to determine their relationship and importance to other features. For example, for a feature point with the greatest order volume, the server may analyze other features associated with the feature point, such as purchase frequency, order amount, etc., to learn cross-correlation data therebetween. Further, the server generates a third feature distribution space based on the first feature distribution space and the cross-correlation data for each of the first feature points. The third feature distribution space is generated on the basis of cross analysis and feature searching, which provides a more comprehensive view of feature correlations. In this space, the server clearly sees the relationship between the first feature point and other features associated with it. By observing the third feature distribution space, the server obtains more in-depth insight into order features, home attributes, and associations between them, supporting more accurate business decisions and strategic planning. For example, assume that the server determines a feature point with the largest order quantity in the first feature distribution space, and constructs an association rule with the household product type. One of the association rules is: when the purchase frequency is high, the order amount is large, and the order type is a household ornament, the probability that the household product type is a wall hanging decorative picture is high. And the server performs cross interest analysis on the feature point with the largest order quantity. The server analyzes the feature points in combination with feature tags in the association rules, such as purchase frequency, order amount, and order type. Through analysis, the purchase frequency in the orders corresponding to the feature points is high, the amount of the orders is large, and the types of the orders are home decorations. Further, the server generates a third feature distribution space based on the first feature distribution space and the cross-correlation data for each first feature point. In this space, the server sees the relationship between the feature point where the order quantity is largest and the home ornament type and other features. By observing the third feature distribution space, the server finds that there is a significant cross-correlation between the feature point with the largest order volume and the purchase frequency, the large order amount, and the specific household product type. By performing cross-interest analysis and feature searching on the first feature distribution space, the server obtains a third feature distribution space that provides more insight and understanding of the relationship between the order features and the home attributes. Such analysis and visualization tools can help servers discover potential market trends, user preferences, and business opportunities, supporting more accurate product positioning, marketing, and business decisions.

S105, screening and eliminating the characteristic distribution of the third characteristic distribution space according to the second characteristic distribution space to generate a target characteristic distribution space;

specifically, the server identifies overlapping points and difference points of the second feature distribution space and the third feature distribution space. By comparing the two feature distribution spaces, the server finds the overlap point and the difference point between them. The overlapping points represent data points having similar feature distributions in the two feature distribution spaces, while the difference points represent data points having different feature distributions in the two feature distribution spaces. And the server eliminates the overlapping points of the third feature distribution space based on the target overlapping points. The server obtains an initial feature distribution space by removing the target overlap point from the third feature distribution space. This step aims at removing data points in the third feature distribution space that overlap higher with the second feature distribution space in order to more accurately capture the distribution of the target feature. And the server searches adjacent points of the target difference points in the initial feature distribution space to obtain first-level adjacent points. Adjacent points refer to data points that have an adjacent relationship in the feature space to the target differential point. The server further expands the feature distribution space of the server by searching for adjacent points and finds other feature points associated with the target discrepancy point. After traversing the target differential points and searching the first-level adjacent points, the server continues traversing the second-level adjacent points of the first-level adjacent points until no adjacent points meeting the preset requirements exist in the initial feature distribution space. This process can help the server capture a wider range of feature points and further enrich the feature distribution space of the server. Further, the server performs feature point distribution integration on the original feature distribution space to obtain a target feature distribution space. By integrating and summarizing the feature points which are screened and searched, the server obtains a target feature distribution space which more completely and accurately reflects feature distribution. For example, assume that the second feature distribution space of the server contains two features of order shipping status and order cancellation, and the third feature distribution space contains two features of home product size and return time. The server identifies overlapping points and difference points of the two feature distribution spaces and eliminates overlapping points which overlap with the second feature distribution space more highly. In this example, the server eliminates overlap points from the third feature distribution space that overlap higher with the second feature distribution space. And the server searches adjacent points of the rest target difference points to expand the characteristic distribution space of the server. It is assumed that during the search, the server finds a target difference point representing an order with a larger home product size but a shorter return time. The server searches for a first level neighbor of the point and finds data points adjacent to it and having a similar characteristic distribution, such as an order with a slightly smaller home product size but a longer return time. Continuing to search for secondary neighbors, the server may find other orders with similar feature distributions, such as orders with similar home product size and return time to the target point of difference. Through this process, the server expands the feature distribution space stepwise, thereby obtaining more feature points related to the target difference points. Finally, the server performs feature point distribution integration on the original feature distribution space to obtain a target feature distribution space, wherein the target feature distribution space comprises feature points which are obviously different from the second feature distribution space and the third feature distribution space. In this embodiment, the server may generate the target feature distribution space by identifying the overlapping points and the difference points, eliminating the overlapping points, searching the neighboring points, and integrating the feature points in the second feature distribution space and the third feature distribution space. The space can more accurately reflect the relation between the order feature and the household attribute, and provide valuable information for subsequent analysis and decision-making.

S106, generating a target pushing list of the target user according to the target feature distribution space, and pushing home services to the target pushing list through the home service system.

Specifically, the server acquires a plurality of second feature points from the target feature distribution space, wherein the feature points represent order features and home attributes with higher relevance to the target user. For each second feature point, the server calculates an association weight coefficient between the second feature point and the target user. The correlation weight coefficient reflects the importance degree of the second characteristic point to the target user, and the server calculates according to a specific algorithm or evaluation index. For example, the server calculates a similarity score between the feature point and the target user as the association weight coefficient using a similarity-based method. And the server sorts the associated products of the target users according to the associated weight coefficient of each second characteristic point to obtain a target pushing list of the target users. The associated weight coefficient is used as a basis for sorting, and a higher weight coefficient means that the matching degree of the product and the target user is higher, and the product should be prioritized in the push list. Further, the server can rank the most relevant and most potential products in front of the target pushing list, and the pushing effectiveness and individuation degree are improved. Further, based on the home service system, the server matches the target push list with a corresponding target push mode. The target push mode is a predefined push rule or policy, determined according to the preferences, behavior and needs of the target user. After matching the target push list and the target push mode, the server carries out home service push to the target user through various channels and modes according to the functions and capabilities of the specific home service system. For example, customized home services may be provided by sending personalized push information to the user via a mobile application, a short message, an email, an intelligent device, etc. Through the flow, the server can generate a target pushing list of the target user according to the target feature distribution space, and home service pushing is carried out on the list through the home service system. The pushing mode can better meet the demands and interests of users, provide personalized home service experience and improve the satisfaction degree and service effect of the users. For example, assume a case where one second feature point in the target feature distribution space represents that the order transportation state is delayed. By calculating the associated weight coefficient of the feature point and the target user, the server determines the importance of the feature point to the target user. When ordering the associated products of the target user, the server finds that one of the associated products is a home delivery company that provides expedited shipping services. Based on the second feature point's associated weight coefficient with the target user, the server ranks the product in front of the target push list because it has a higher association with the target user's order shipping status delay. And the server matches the target push list with a corresponding target push mode. Assume that the target push mode is personalized push according to the user's preferences and needs. In this embodiment, the server sets the push mode to provide a quick solution and discount offers according to the preference and order situation of the target user. For example, the server pushes a message to the target user, the content being: "because of your order shipping status delay, the server provides your emergency shipping services and an additional 10% discount offer. Clicking on the link looks at the details and confirms your order. By such personalized pushing, the server meets the urgent needs of the user and provides a targeted solution, and in this embodiment, discount offers are also provided as incentives. Further, the server sends the personalized push information to the target user through the home service system. The user can receive the push notification through the mobile application program, click the notification to view the detailed information and confirm the order. In this embodiment, the home service system records feedback and behavior of the user, and is used for subsequent data analysis and optimization of the push strategy. By the above example, it can be seen how to generate the target push list of the target user according to the target feature distribution space, and perform personalized home service push through the home service system. The pushing mode can meet the requirements and preferences of users, can improve the experience and satisfaction of the users, and promotes the provision and sales of home services.

In the embodiment of the invention, order feature extraction is carried out on a normally completed order, and a first feature distribution space is created; performing order feature analysis on the abnormal unfinished order and creating a second feature distribution space; performing cross interest analysis and feature search on the first feature distribution space to obtain a third feature distribution space; according to the second characteristic distribution space, characteristic distribution screening and eliminating are carried out on the third characteristic distribution space, and a target characteristic distribution space is generated; according to the invention, the target feature distribution space is predicted more accurately through a machine learning algorithm and cross interest analysis according to the historical order data of the clients, the client unit price and purchase frequency of the clients can be improved through cross sales and personalized recommendation, the profit rate is indirectly improved, the clients can be better interacted and communicated through the recommendation and the pushing list, more close client relations are established, and the accuracy of order data processing is further improved.

In a specific embodiment, the process of executing step S101 may specifically include the following steps:

(1) Acquiring user information of a target user, and inquiring orders in a target time period from a preset home service system according to the user information to obtain historical home order data;

(2) Performing data deduplication and data format standardization processing on the historical household order data to obtain standard household order data;

(3) And classifying order data of the standard household order data according to the preset order node information to obtain a normally completed order and an abnormally unfinished order.

Specifically, the server obtains user information of the target user, including user ID, name, contact information, and the like. This information may be collected during user registration or login. For example, user ID 123456, surname xx, contact xx@example.com. The server queries from a preset home service system by utilizing the user information of the target user. The server obtains the corresponding order data according to the target time period (e.g., 2023, 1, to 2023, 3, 31). And the server acquires all household order data of the target user in the time period through a query interface or database query language provided by the system. After the historical household order data are obtained, the server performs duplication removal and data format standardization processing on the data so as to ensure the accuracy and consistency of the data. The deduplication operation may be implemented by comparing order numbers or other unique identifiers, ensuring that each order appears only once. The data format standardization process comprises unifying date format, normalized product name and price format, etc., so that the data is easy to process and analyze. And according to the preset order node information, the server classifies the order data of the standard household order data. The order node information includes order status, payment status, delivery status, etc. The server classifies the order data into two categories, normal completed orders and abnormal unfinished orders, based on the information. For example, completed orders may be those orders whose status is delivered, paid, and not cancelled; while outstanding orders may be those orders whose status is pending, unpaid or cancelled. In this embodiment, the server obtains user information of the target user and queries the target of the historical home order data. The server obtains accurate and standard-compliant order data from the system and classifies it as normal completed orders and abnormal unfinished orders. These order data and classification results provide the basis for subsequent analysis and decision making, such as analysis of user behavior, identification of order trends, and monitoring and handling of anomalies.

In a specific embodiment, as shown in fig. 2, the process of executing step S102 may specifically include the following steps:

s201, extracting order features of a normally completed order to obtain a plurality of first order feature labels, wherein the plurality of first order feature labels comprise: purchase frequency, order amount, order type, and order time;

s202, extracting home attribute features of a normally completed order to obtain a plurality of first home attribute feature labels, wherein the plurality of first home attribute feature labels comprise: household product type, color and material;

s203, performing feature label clustering on the plurality of first order feature labels to obtain a plurality of first feature clustering points, and performing feature label clustering on the plurality of first home attribute feature labels to obtain a plurality of second feature clustering points;

s204, performing feature distribution space mapping on the first feature cluster points and the second feature cluster points to generate a first feature distribution space.

Specifically, the server performs order feature extraction on the order which is normally completed, so as to obtain a plurality of first order feature labels. These feature tags include purchase frequency, order amount, order type, order time, etc. The purchase frequency represents the number of times the user places an order in a period of time, the amount of the order represents the total amount of each order, the type of order represents the category of the order (e.g., purchase, return, change, etc.), and the time of the order represents the creation time or delivery time of the order. By extracting these features, the server knows the user's buying habit, distribution of the amount of the order, preference of the type of order, distribution of the time of the order, and so on. Aiming at the order which is normally completed, the server performs home attribute feature extraction to obtain a plurality of first home attribute feature tags. These feature labels include household product type, color, material, etc. The household product type represents the type of the purchased product in the order, the color represents the color attribute of the product, and the material represents the material attribute of the product. By extracting these features, the server knows the type preferences of household products purchased by the user, the color preferences, and the preferences for products of different materials. And then, the server performs feature label clustering on the plurality of first order feature labels to obtain a plurality of first feature clustering points. Cluster analysis may help the server categorize similar order features to better understand the relevance and interaction between order features. For example, purchase frequency and order amount may be aggregated in one type of cluster point, while order type and order time may be aggregated in another type of cluster point. Similarly, the server performs feature tag clustering on the plurality of first home attribute feature tags to obtain a plurality of second feature clustering points. These cluster points can help the server to learn about correlations and co-occurrence relationships between household attributes. For example, household product types and colors may be aggregated in one cluster point, while textures may be aggregated in another cluster point. Further, the server performs feature distribution space mapping on the plurality of first feature cluster points and the plurality of second feature cluster points, thereby generating a first feature distribution space. The feature distribution space may help the server visualize and understand the distribution and interrelationships between different features. For example, the server draws a scatter plot of purchase frequency and order amount in the feature distribution space to observe correlations and trends between them. Through such analysis, the server finds that customers who purchase more frequently tend to have a higher tendency for the order amount, while customers who purchase less frequently may exhibit a more decentralized distribution of order amounts. In addition, the server draws the characteristics of the type, the color, the material and the like of the household products in the characteristic distribution space so as to observe aggregation and distribution conditions among the characteristics. For example, assume that the server obtains a feature distribution space in which the purchase frequency and the order amount are two axes. The server draws some data points in this space and finds that points with higher purchase frequency tend to be in the higher order amount area, while points with lower purchase frequency are distributed in the lower order amount area. This implies that there may be a positive correlation between the frequency of purchases and the amount of orders, i.e. customers with high frequency of purchases are more inclined to place orders of high amounts. In addition, the server observes the distribution condition of the characteristics such as the type, the color and the material of the household products in the characteristic distribution space. Through cluster analysis of these features, the server finds that certain specific types of household products, color preferences, or texture preferences may form unique aggregate regions in the feature distribution space. For example, the server may observe that sofa, red, and woody form an aggregated region in the feature distribution space, which means that there is some correlation between these features. By generating the first feature distribution space, the server is able to more fully understand the relationship between the order features and the home attributes and obtain valuable information therefrom.

In a specific embodiment, the process of executing step S103 may specifically include the following steps:

(1) Extracting order features of the abnormal unfinished order to obtain a plurality of second order feature labels, wherein the plurality of second order feature labels comprise: order transportation state, order cancellation condition and order reminding content;

(2) Extracting home attribute features of the abnormal unfinished order to obtain a plurality of second home attribute feature labels, wherein the plurality of second home attribute feature labels comprise: home product size, quality, and return time;

(3) Performing feature label clustering on the plurality of second order feature labels to obtain a plurality of third feature clustering points, and performing feature label clustering on the plurality of second home attribute feature labels to obtain a plurality of fourth feature clustering points;

(4) And performing feature distribution space mapping on the plurality of third feature cluster points and the plurality of fourth feature cluster points to generate a second feature distribution space.

Specifically, for an abnormally incomplete order, the server extracts a plurality of second order feature labels therefrom. These feature tags include the shipping status of the order, i.e., the current logistics of the order, such as being dispatched, shipped or to be shipped, etc.; the cancellation of an order, i.e., whether the order is cancelled or paused; and the reminding content of the order, namely the reminding information related to the order, such as overdue reminding or replenishment reminding. And the server extracts home attribute characteristics of the abnormal unfinished orders, and the server obtains a plurality of second home attribute characteristic labels. The characteristic labels comprise the sizes of household products, namely the length, width, height and other size information of the products; the quality of household products, namely the weight or the quality of materials of the products and the like; and return time, i.e., the time the customer applies for a return. The server clusters the feature labels of the plurality of second order feature labels to find features that are similar or related to each other. Through cluster analysis, the server gathers orders with similar transportation states, cancellation situations or reminding contents together to form a plurality of third feature cluster points. In this embodiment, feature label clustering is performed on the plurality of second home attribute feature labels, so that a plurality of fourth feature clustering points can be obtained, where home products have features such as similar size, quality or return time. Further, the server maps the plurality of third feature cluster points and the plurality of fourth feature cluster points into a feature distribution space, thereby generating a second feature distribution space. In this feature distribution space, the server observes the distance, distribution and relative position between the different cluster points, as well as their relationship to other features. Such feature distribution space may help the server better understand the feature distribution of abnormally incomplete orders, revealing potential associations and laws between order features and home attributes. For example, assume that the server extracts order shipment status, order cancellation status, and order reminder content from the abnormal incomplete order as second order feature tags, and home product size, quality, and return time as second home attribute feature tags. After feature label clustering, the server obtains two third feature clustering points: one cluster point represents a collection of orders with similar shipping status and cancellation status and another cluster point represents a collection of orders with similar order alert content. In this embodiment, the server obtains two fourth feature cluster points: one cluster point represents a set of orders with similar household product sizes and the other cluster point represents a set of orders with similar household product quality and return time. These third feature cluster point and fourth feature cluster point are mapped into the feature distribution space, and the server observes their distribution in space. For example, the server finds orders with similar shipping status and cancellation conditions clustered together in the feature distribution space to form a tight cluster. Likewise, orders with similar order alert content may form another cluster. In addition, orders of similar household product sizes may form one aggregation point, while orders of similar quality and return time may form another aggregation point. By generating the second feature distribution space, the server more intuitively understands the feature distribution of the abnormally unfinished order. The server observes the associations between different features, such as the correlation between order shipping status and order cancellation, and the degree of association between home product size and quality. Such information may provide important references for subsequent analysis and decision making, such as optimizing logistics management, improving order processing flow, or optimizing product size and quality control.

In a specific embodiment, as shown in fig. 3, the process of executing step S104 may specifically include the following steps:

s301, respectively constructing association rules between each first order feature tag and household product types to obtain a plurality of product association rules;

s302, determining a plurality of corresponding first feature points based on a first feature distribution space, wherein the plurality of first feature points comprise: the order volume is maximum, the profit is highest, the sales amount is highest, and the recommended access volume is the most;

s303, performing cross interest analysis on each first feature point according to a plurality of product association rules to obtain cross association data of each first feature point;

s304, generating a third feature distribution space according to the first feature distribution space and the cross-correlation data of each first feature point.

Specifically, the server will establish association rules between the feature tag and the household product type for each first order. This may be accomplished by a data mining algorithm, such as an association rule mining algorithm (e.g., apriori algorithm). The server will analyze the order data and the household product type data to find the correlation between them. For example, the server may get the following product association rules: if the order type is "furniture," the household product type may be "sofa" or "bed. These association rules can reveal the relationship between the order features and the type of household product. The server will determine a plurality of first feature points based on the first feature distribution space. These feature points represent a collection of orders placed at a particular feature value. The server selects some important feature points such as maximum order volume, maximum profit, maximum sales and maximum recommended access volume. For example, the server determines a feature point with the largest order quantity, which represents the set of orders with the largest order quantity under that feature. The server will perform cross-interest analysis on each first feature point using the product association rules to obtain cross-association data. This may help the server to learn the relationship between each feature point and other features. For example, for a feature point where the order quantity is maximum, the server analyzes the association relationship between it and other features (such as order amount, order type, etc.). This will provide cross-correlation data about the different features. Further, the server will combine the first feature distribution space and the cross-correlation data for each first feature point to generate a third feature distribution space. This feature distribution space will reflect the integrated relationship between the order features and the household product types. The server presents a third feature distribution space through a visualization technique, such as drawing a scatter plot or thermodynamic diagram, to demonstrate the degree of association and distribution among the different features. In this embodiment, the server constructs a product association rule, determines a first feature point, acquires cross-association data of the first feature point, and finally generates a third feature distribution space. This will help the server to better understand the relationship between the order features and the household product types, providing an important basis for subsequent analysis and decision making. For example, assume that the server is analyzing order data of a home electronics platform and the purchase frequency is a first order feature tag, and the home product type is a first home attribute feature tag. The server discovers the following two product association rules through an association rule mining algorithm:

Rule 1: if the purchase frequency is high (frequency greater than 5), the household product type may be a sofa or a bed.

Rule 2 if the purchase frequency is medium frequency (frequency between 3 and 5), the household product type may be a table or chair.

Based on these association rules, the server determines two first feature points: high frequency purchase and medium frequency purchase. The server now performs a cross-point interest analysis on the two feature points to obtain their cross-correlation data. For feature points purchased at high frequencies, the server analyzes the relationship between it and other features. Assuming that the server finds the order for high frequency purchases, the sales of the sofa is the highest, the profit of the bed is the highest, and the product with the highest recommended visit is a chair. These data reflect the correlation between the high frequency purchase feature points and the order amount, profit, and recommended access. For feature points purchased at intermediate frequencies, the server performs a similar analysis. Assuming that the server finds that the sales of the dining table is highest, the profit of the chair is highest, and the product with the highest recommended visit is a sofa in the medium-frequency purchase order. These data reflect the association of medium frequency purchase feature points with order amounts, profits, and recommended access. Further, the server generates a third feature distribution space by combining the purchase frequency as a first feature tag and the household product type as a first household attribute feature tag and cross-correlation data. In this feature distribution space, the server sees the degree of association between the purchase frequency and the order amount, profit, and recommended access amount, and the degree of association between the household product type and the order amount, profit, and recommended access amount. Such visual analysis may help the server discover patterns and trends between features, further guiding the server's decisions and strategies.

In a specific embodiment, as shown in fig. 4, the process of performing step S105 may specifically include the following steps:

s401, identifying overlapping points and difference points of the second feature distribution space and the third feature distribution space to obtain a target overlapping point and a target difference point;

s402, eliminating overlapping points of the third feature distribution space according to the target overlapping points to obtain an initial feature distribution space;

s403, searching adjacent points for target difference points in the initial feature distribution space to obtain first-level adjacent points;

s404, after traversing the target difference points in sequence, traversing the second-level adjacent points of the first-level adjacent points in sequence until no adjacent points meeting the preset requirements exist in the initial feature distribution space, and obtaining an original feature distribution space;

s405, carrying out feature point distribution integration on the original feature distribution space to obtain a target feature distribution space.

Specifically, the server compares the second feature distribution space with the third feature distribution space to find out common features and differences between them. By comparing their feature distributions, overlapping points and difference points can be identified. The overlapping point refers to the feature points existing in the present embodiment in two feature distribution spaces, and the difference point refers to the feature points that appear in only one of the feature distribution spaces. The server will process for the target overlap point. And according to the positions of the overlapped points and the characteristic distribution condition, the server eliminates the points from the third characteristic distribution space to obtain an initial characteristic distribution space. The purpose of this is to reduce the impact of overlapping points on subsequent analysis and processing, making the initial feature distribution space clearer and more accurate. And the server searches adjacent points for the target difference points in the initial characteristic distribution space. The adjacent points refer to other feature points which are closer to the target difference point in the feature space. The server searches for adjacent points of the target differential point according to preset requirements, such as a distance threshold or feature similarity. The first-level adjacent points are feature points directly connected with the target difference points, and the second-level adjacent points are feature points indirectly connected with the target difference points. In the process of traversing the target difference points and the adjacent points, the server screens the adjacent points meeting the preset requirements according to specific requirements, such as relevance or similarity between features, and gradually expands the search range. Such an iterative search process may continue until no satisfactory neighboring points can be found in the initial feature distribution space. The final feature point set constitutes the original feature distribution space. Further, the server integrates the feature point distribution of the original feature distribution space. This step aims to combine similar or related feature points to obtain a more comprehensive and comprehensive target feature distribution space. By integrating similar feature points, redundancy of feature points can be reduced and important feature patterns and trends can be highlighted. Such as the above procedure. Assume that a server is to analyze user purchasing behavior and product attributes on an e-commerce platform. The second feature distribution space may represent the purchase frequency and order amount of the user, while the third feature distribution space may represent the type and color of the product. By comparing and analyzing the two feature distribution spaces, the server finds the overlap point and the difference point. For example, the server finds that certain users have a high frequency of purchase and a high amount of order, and that these users tend to purchase a particular type of product, such as a sofa. These overlapping points represent intersections of high value users and popular products. On the other hand, the point of difference may refer to a user who purchases a low frequency or a low order amount, or a user who purchases other types of products. After culling the overlapping points and searching for adjacent points, the server may find that there are a level of adjacent points near some of the discrepancy points, which may represent a population of users with potential purchase potential or specific preferences. Further searching for secondary neighbors may reveal more associated features or market segments. Finally, by integrating feature points in the original feature distribution space, the server obtains a target feature distribution space, wherein the relationships and distributions between different features are better presented. The feature distribution space can provide important reference basis for personalized recommendation, fine marketing and product planning of the e-commerce platform. In the embodiment, the server realizes the generation of the target feature distribution space by identifying, removing and searching the overlapped points and the difference points and integrating the feature points, so that the relation between the features is better understood and utilized, and data support and guidance are provided for service decision.

In a specific embodiment, the process of executing step S106 may specifically include the following steps:

(1) Acquiring a plurality of second feature points in a target feature distribution space, and calculating association weight coefficients of the plurality of second feature points and a target user to obtain association weight coefficients of each second feature point;

(2) According to the association weight coefficient of each second feature point, ordering the association products of the target user to obtain a target pushing list of the target user;

(3) And matching a target pushing mode corresponding to the target pushing list based on the home service system, and pushing the home service according to the target pushing mode and the target pushing list.

Specifically, an appropriate second feature point, such as a purchase frequency, an order amount, a purchase preference, or the like, is selected from the target feature distribution space. These feature points should be related to the needs and behavior of the target user. The server calculates, for each selected second feature point, an association weight coefficient between it and the target user. This may be achieved by various statistical methods, machine learning algorithms, or rules based on domain knowledge. The association weight coefficient represents the degree of association between the second feature point and the target user, and the higher the value is, the stronger the association between the two is. The server ranks the associated products of the target user based on the associated weight coefficient of each second feature point. The higher the association weight coefficient, the stronger the association degree of the product with the target user, and the product should be preferentially pushed to the target user. Further, the home service system is matched according to products in the target pushing list and a target pushing mode preset in the system. The target push mode defines the policy and manner of pushing and may be based on specific activities, personalized recommendations, promotions, and the like. For example, assuming that the target user is a home fan, the system selects purchase frequency, order amount, and purchase preference as target feature points. According to the purchase record and behavior data of the target user, the system assumes A, B, C three products with high purchase frequency of the target user, and the corresponding association weight coefficients are respectively 0.8, 0.6 and 0.4; the number of products with high order amount is D, E, F, and the corresponding association weight coefficients are 0.7, 0.9 and 0.5 respectively; products with higher purchase preference correlation are G, H, I, and the corresponding association weight coefficients are 0.6, 0.8 and 0.7 respectively. After sorting according to the associated weight coefficients, the order of the target push list may be as follows:

Product E: because the associated weight coefficient of the order amount is highest, the product with high order amount is arranged in front, and the product E obtains a higher associated weight coefficient.

Product B: the associated weight coefficient of the purchase frequency is higher, and the product B is one of the products with high purchase frequency and is therefore ranked second.

Product H: the associated weight coefficient of the purchase preference is higher, and the product H accords with the purchase preference of the target user, so that the product H is ranked in the third position.

According to a preset target pushing mode in the home service system, such as 'new first shot', 'promotion offer', and the like, the system matches the target pushing list and executes a corresponding home service pushing strategy. For example, if the target push mode is "new first," the system will push the most recently marketed household products to target users based on the products purchased frequently in the target push list to satisfy their interests and needs for the new product. Similarly, if the target push mode is "promotional offers," the system will push promotional information and coupons for household products to target users to attract their purchases based on the products in the target push list that have a high order amount. In sum, by acquiring a plurality of second feature points in the target feature distribution space and calculating the association weight coefficient, the association products of the target users are ordered, and home service pushing is performed according to the target pushing mode, so that the individuation degree and pushing effect of the home service can be improved, the requirements of the target users are met, and the user experience is improved.

The method for processing order data in the embodiment of the present invention is described above, and the apparatus for processing order data in the embodiment of the present invention is described below, referring to fig. 5, where an embodiment of the apparatus for processing order data in the embodiment of the present invention includes:

the acquiring module 501 is configured to acquire historical home order data corresponding to a target user based on a preset home service system, and classify the historical home order data to obtain a normally completed order and an abnormally incomplete order;

the first creating module 502 is configured to perform order feature extraction on the normally completed order, obtain a plurality of first order feature labels and a plurality of first home attribute feature labels, and create a first feature distribution space according to the plurality of first order feature labels and the plurality of first home attribute feature labels;

a second creating module 503, configured to perform order feature analysis on the abnormal unfinished order, obtain a plurality of second order feature labels and a plurality of second home attribute feature labels, and create a second feature distribution space according to the plurality of second order feature labels and the plurality of second home attribute feature labels;

An analysis module 504, configured to perform cross interest analysis and feature search on the first feature distribution space, to obtain a third feature distribution space;

the screening module 505 is configured to perform feature distribution screening and rejection on the third feature distribution space according to the second feature distribution space, so as to generate a target feature distribution space;

and the pushing module 506 is configured to generate a target pushing list of the target user according to the target feature distribution space, and perform home service pushing on the target pushing list through the home service system.

Through the cooperative cooperation of the components, extracting order characteristics of a normally completed order and creating a first characteristic distribution space; performing order feature analysis on the abnormal unfinished order and creating a second feature distribution space; performing cross interest analysis and feature search on the first feature distribution space to obtain a third feature distribution space; according to the second characteristic distribution space, characteristic distribution screening and eliminating are carried out on the third characteristic distribution space, and a target characteristic distribution space is generated; according to the invention, the target feature distribution space is predicted more accurately through a machine learning algorithm and cross interest analysis according to the historical order data of the clients, the client unit price and purchase frequency of the clients can be improved through cross sales and personalized recommendation, the profit rate is indirectly improved, the clients can be better interacted and communicated through the recommendation and the pushing list, more close client relations are established, and the accuracy of order data processing is further improved.

The processing device for order data in the embodiment of the present invention is described in detail above in fig. 5 from the point of view of a modularized functional entity, and the processing device for order data in the embodiment of the present invention is described in detail below from the point of view of hardware processing.

Fig. 6 is a schematic structural diagram of an apparatus for processing order data according to an embodiment of the present invention, where the apparatus 600 for processing order data may have a relatively large difference according to a configuration or a performance, and include one or more processors (central processing units, CPU) 610 (e.g., one or more processors) and a memory 620, and one or more storage media 630 (e.g., one or more mass storage devices) storing application programs 633 or data 632. Wherein the memory 620 and the storage medium 630 may be transitory or persistent storage. The program stored in the storage medium 630 includes one or more modules (not shown), each of which includes a series of instruction operations in the processing apparatus 600 for order data. Still further, the processor 610 may be configured to communicate with the storage medium 630 to execute a series of instruction operations in the storage medium 630 on the order data processing device 600.

The order data processing apparatus 600 also includes one or more power supplies 640, one or more wired or wireless network interfaces 650, one or more input/output interfaces 660, and/or one or more operating systems 631, such as Windows Serve, macOS X, unix, linux, freeBSD, and the like. It will be appreciated by those skilled in the art that the configuration of the order data processing apparatus shown in FIG. 6 is not limiting of the order data processing apparatus, including more or fewer components than shown, or a combination of certain components, or a different arrangement of components.

The invention also provides order data processing equipment, which comprises a memory and a processor, wherein the memory stores computer readable instructions, and the computer readable instructions, when executed by the processor, cause the processor to execute the steps of the order data processing method in the above embodiments.

The present invention also provides a computer readable storage medium, which may be a non-volatile computer readable storage medium, or a volatile computer readable storage medium, having stored therein instructions that, when executed on a computer, cause the computer to perform the steps of the order data processing method.

It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, which are not repeated herein.

The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a storage medium, including instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a read-only memory (ROM), a random access memory (randomacceS memory, RAM), a magnetic disk, or an optical disk, or other various media capable of storing program codes.

The above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for processing order data, the method comprising:

performing cross interest analysis and feature search on the first feature distribution space to obtain a third feature distribution space, wherein the method specifically comprises the following steps: respectively constructing association rules between each first order feature tag and household product types to obtain a plurality of product association rules; determining a corresponding plurality of first feature points based on the first feature distribution space, wherein the plurality of first feature points comprises: the order volume is maximum, the profit is highest, the sales amount is highest, and the recommended access volume is the most; according to the product association rules, cross interest analysis is carried out on each first feature point, and cross association data of each first feature point is obtained; generating a third feature distribution space according to the first feature distribution space and the cross-correlation data of each first feature point;

2. The method for processing order data according to claim 1, wherein the acquiring, based on the preset home service system, the historical home order data corresponding to the target user, and classifying the order data to obtain a normally completed order and an abnormally unfinished order, includes:

3. The method for processing order data according to claim 1, wherein said extracting order features of the normally completed order to obtain a plurality of first order feature labels and a plurality of first home attribute feature labels, and creating a first feature distribution space according to the plurality of first order feature labels and the plurality of first home attribute feature labels, comprises:

4. The method for processing order data according to claim 1, wherein the performing order feature analysis on the abnormal unfinished order to obtain a plurality of second order feature labels and a plurality of second home attribute feature labels, and creating a second feature distribution space according to the plurality of second order feature labels and the plurality of second home attribute feature labels comprises:

5. The method for processing order data according to claim 1, wherein the step of performing feature distribution screening and rejection on the third feature distribution space according to the second feature distribution space to generate a target feature distribution space includes:

6. The method for processing order data according to claim 1, wherein the generating a target push list of the target user according to the target feature distribution space and performing home service push on the target push list by the home service system includes:

7. An order data processing apparatus, characterized in that the order data processing apparatus includes:

The analysis module is used for carrying out cross interesting analysis and feature search on the first feature distribution space to obtain a third feature distribution space, and specifically comprises the following steps: respectively constructing association rules between each first order feature tag and household product types to obtain a plurality of product association rules; determining a corresponding plurality of first feature points based on the first feature distribution space, wherein the plurality of first feature points comprises: the order volume is maximum, the profit is highest, the sales amount is highest, and the recommended access volume is the most; according to the product association rules, cross interest analysis is carried out on each first feature point, and cross association data of each first feature point is obtained; generating a third feature distribution space according to the first feature distribution space and the cross-correlation data of each first feature point;

8. An order data processing apparatus, characterized in that the order data processing apparatus comprises: a memory and at least one processor, the memory having instructions stored therein;

the at least one processor invoking the instructions in the memory to cause the order data processing apparatus to perform the order data processing method of any of claims 1-6.

9. A computer readable storage medium having instructions stored thereon, which when executed by a processor, implement a method of processing order data according to any of claims 1-6.