A customer portrait analysis method based on multi-dimensional logistics data

Abstract

The present invention discloses a customer portrait analysis method based on logistics multi-dimensional data, which belongs to the field of data analysis technology and includes the following steps: S1, obtaining the purchase records of each customer, wherein the purchase records include product information and logistics information; S2, determining the popular product data according to the co-occurrence matrix corresponding to the product information of all customers; S3, determining the popular area data according to the logistics information of all customers; S4, uploading the popular product data and the popular area data as the customer portrait analysis results to a multi-source heterogeneous fusion database. The present invention determines the product direction of interest and the popular purchase area of ​​users in the platform, which is convenient for grasping the user's purchasing tendency and habits, and is conducive to the platform to understand the user's preferences and adjust the strategy in time.

Description

Technical Field

[0001] The present invention belongs to the technical field of data analysis, and specifically relates to a customer portrait analysis method based on multi-dimensional logistics data. Background Art

[0002] In the field of e-commerce, current consumers are accustomed to online shopping on major platforms, and customers' demand for accurate platform recommendations is increasing. The platform needs to accurately grasp and efficiently respond to customer needs, so it is necessary to conduct a profile analysis of platform users so that the platform can meet the purchasing needs of most customers. However, the existing analysis methods for customer preferences are not accurate enough and cannot combine product information and logistics information. Summary of the invention

[0003] In order to solve the above problems, the present invention proposes a customer portrait analysis method based on multi-dimensional logistics data.

[0004] The technical solution of the present invention is: a customer portrait analysis method based on logistics multi-dimensional data comprises the following steps:

[0005] S1. Obtain the purchase records of each customer, where the purchase records include product information and logistics information;

[0006] S2. Determine popular product data based on the co-occurrence matrix corresponding to the product information of all customers;

[0007] S3. Determine the data of popular regions based on the logistics information of all customers;

[0008] S4. Upload the popular product data and popular region data as customer portrait analysis results to the multi-source heterogeneous fusion database.

[0009] The information that shopping platforms need to store is diverse, covering various forms of heterogeneous data, so it needs to be uploaded to a multi-source heterogeneous fusion database for storage. A multi-source heterogeneous fusion database is a special data warehouse or data management system that can integrate heterogeneous data from multiple different data sources.

[0010] Furthermore, in S1, the product information includes the product name, and the logistics information includes the product delivery address.

[0011] Furthermore, S2 includes the following sub-steps:

[0012] S21, removing stop words from each product information, treating all product information as product text, and extracting a co-occurrence matrix of the product text;

[0013] S22, determining the chasing weight of adjacent product information in the product text according to the co-occurrence matrix of the product text;

[0014] S23, determining a follow-up interval for each product information according to the chasing weights of adjacent product information in the product text;

[0015] S24. Determine popular product data according to the co-occurrence matrix and the follow-up interval of each product information.

[0016] The beneficial effect of the above further scheme is: in the present invention, the product text contains all product names, and the product names are various and the amount of information is huge. It is necessary to analyze each product information in the product text one by one, so the chasing weight between adjacent product information is calculated; according to the chasing weight between the current product information and the previous product information and the next product information, the following interval of the current product information is determined, and the important product information, that is, the popular product data, is determined by the following interval.

[0017] The co-occurrence matrix is ​​used to represent the co-occurrence relationship between words or phrases in a document, and can be used to analyze the correlation and semantic relationship between words.

[0018] Furthermore, in S22, the pursuit weight S of the i-th product information and the i+1-th product information in the product text is i,i+1 The calculation formula is: ; In the formula, Represents the word vector of the first word in the i-th product information, Represents the word vector of the first word in the i+1th product information, Represents the word vector of the Nth word in the i+1th product information, Represents the word vector of the mth word in the i-th product information, represents the word vector of the Mth word in the i-th product information, M represents the number of words contained in the i-th product information, N represents the number of words contained in the i+1-th product information, d(·) represents the Euclidean distance function of the vector, and c0 represents an element vector randomly selected from the co-occurrence matrix.

[0019] The denominator calculates the Euclidean distance between the word vector of each word in the previous product information and the word vector of the word in the next product information one by one.

[0020] Furthermore, in S23, the upper limit L1 of the follow-up interval of the i-th product information is calculated as follows: Where S i,i+1 represents the pursuit weight of the i-th product information and the i+1-th product information, S i-1,i represents the pursuit weight of the i-1th product information and the i-th product information, and max(·) represents the maximum value function;

[0021] The calculation formula for the lower limit L2 of the follow-up interval of the i-th product information is: ; Where min(·) represents the minimum value function.

[0022] Further, S24 includes the following sub-steps:

[0023] S241, extracting the global representation value of the product text;

[0024] S242, multiplying the global representation value by the maximum singular value of the co-occurrence matrix to obtain a global hot value;

[0025] S243: The commodity information whose following interval contains the global hot value is used as hot commodity data.

[0026] Furthermore, in S241, the calculation formula of the global representation value Total of the product text is: ; In the formula, β represents the hyperparameter, C k represents the word vector of the kth word in the product text, K represents the number of words in the product text, C1 represents the word vector of the first word in the product text, and C K It represents the word vector of the Kth word in the product text, max(·) represents the maximum value function, and min(·) represents the minimum value function.

[0027] Furthermore, S3 includes the following sub-steps:

[0028] S31. Complete the product delivery and receiving addresses of each customer to obtain the complete product delivery and receiving addresses;

[0029] S32. Determine the popular region data based on the complete delivery and receipt addresses of all customers’ goods.

[0030] The beneficial effect of the above further scheme is: in the present invention, the customer's commodity delivery address is not necessarily standardized, the administrative area is not necessarily complete, and the street is not necessarily correct. Therefore, the present invention completes the delivery address. After completion, the addresses with the area of ​​"city" and the top half of the addresses with the highest frequency of occurrence are extracted as popular area data.

[0031] The beneficial effects of the present invention are as follows: the present invention performs text processing on the product names of all customers in the platform, determines the connections between adjacent product names, and thereby screens out more popular product names as popular product data; the present invention also analyzes the customer's delivery and receipt addresses to determine popular area data; the popular product data and popular area data can globally integrate the purchase information of all customers in the platform, determine the product interests of users in the platform and popular purchase areas, which is convenient for grasping user purchase tendencies and habits, and is beneficial for the platform to understand user preferences and adjust strategies in a timely manner. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is a flow chart of the customer portrait analysis method based on multi-dimensional logistics data. DETAILED DESCRIPTION

[0033] The embodiments of the present invention will be further described below in conjunction with the accompanying drawings.

[0034] like Figure 1 As shown, the present invention provides a customer portrait analysis method based on logistics multi-dimensional data, comprising the following steps:

[0035] S1. Obtain the purchase records of each customer, where the purchase records include product information and logistics information;

[0036] S2. Determine popular product data based on the co-occurrence matrix corresponding to the product information of all customers;

[0037] S3. Determine the data of popular regions based on the logistics information of all customers;

[0038] S4. Upload the popular product data and popular region data as customer portrait analysis results to the multi-source heterogeneous fusion database.

[0039] The information that shopping platforms need to store is diverse, covering various forms of heterogeneous data, so it needs to be uploaded to a multi-source heterogeneous fusion database for storage. A multi-source heterogeneous fusion database is a special data warehouse or data management system that can integrate heterogeneous data from multiple different data sources.

[0040] In the embodiment of the present invention, in S1, the product information includes the product name, and the logistics information includes the product delivery address.

[0041] In this embodiment of the present invention, S2 includes the following sub-steps:

[0042] S21, removing stop words from each product information, treating all product information as product text, and extracting a co-occurrence matrix of the product text;

[0043] S22, determining the chasing weight of adjacent product information in the product text according to the co-occurrence matrix of the product text;

[0044] S23, determining a follow-up interval for each product information according to the chasing weights of adjacent product information in the product text;

[0045] S24. Determine popular product data according to the co-occurrence matrix and the follow-up interval of each product information.

[0046] In the present invention, the product text contains all product names, and the product names are various and the amount of information is huge. It is necessary to analyze each product information in the product text one by one, so the chasing weight between adjacent product information is calculated; according to the chasing weight between the current product information and the previous product information and the next product information, the following interval of the current product information is determined, and the important product information, that is, the popular product data, is determined by the following interval.

[0047] The co-occurrence matrix is ​​used to represent the co-occurrence relationship between words or phrases in a document, and can be used to analyze the correlation and semantic relationship between words.

[0048] In the embodiment of the present invention, in S22, the pursuit weight S of the i-th product information and the i+1-th product information in the product text is i,i+1 The calculation formula is: ; In the formula, Represents the word vector of the first word in the i-th product information, Represents the word vector of the first word in the i+1th product information, Represents the word vector of the Nth word in the i+1th product information, Represents the word vector of the mth word in the i-th product information, represents the word vector of the Mth word in the i-th product information, M represents the number of words contained in the i-th product information, N represents the number of words contained in the i+1-th product information, d(·) represents the Euclidean distance function of the vector, and c0 represents an element vector randomly selected from the co-occurrence matrix.

[0049] The denominator calculates the Euclidean distance between the word vector of each word in the previous product information and the word vector of the word in the next product information one by one.

[0050] In the embodiment of the present invention, in S23, the upper limit L1 of the following interval of the i-th commodity information is calculated by the formula: Where S i,i+1 represents the pursuit weight of the i-th product information and the i+1-th product information, S i-1,i represents the pursuit weight of the i-1th product information and the i-th product information, and max(·) represents the maximum value function;

[0051] The calculation formula for the lower limit L2 of the follow-up interval of the i-th product information is: ; Where min(·) represents the minimum value function.

[0052] In this embodiment of the present invention, S24 includes the following sub-steps:

[0053] S241, extracting the global representation value of the product text;

[0054] S242, multiplying the global representation value by the maximum singular value of the co-occurrence matrix to obtain a global hot value;

[0055] S243: The commodity information whose following interval contains the global hot value is used as hot commodity data.

[0056] In the embodiment of the present invention, in S241, the calculation formula of the global representation value Total of the commodity text is: ; In the formula, β represents the hyperparameter, C k represents the word vector of the kth word in the product text, K represents the number of words in the product text, C1 represents the word vector of the first word in the product text, and C K It represents the word vector of the Kth word in the product text, max(·) represents the maximum value function, and min(·) represents the minimum value function.

[0057] In this embodiment of the present invention, S3 includes the following sub-steps:

[0058] S31. Complete the product delivery and receiving addresses of each customer to obtain the complete product delivery and receiving addresses;

[0059] S32. Determine the popular region data based on the complete delivery and receipt addresses of all customers’ goods.

[0060] In the present invention, the customer's goods delivery address may not be standardized, the administrative area may not be complete, and the street may not be correct. Therefore, the present invention completes the delivery address. After completion, the addresses with the area of ​​"city" and the top half of the addresses with the highest frequency of occurrence are extracted as popular area data.

[0061] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A customer portrait analysis method based on multi-dimensional logistics data, characterized in that: The following steps are involved: S1. Obtain the purchase records of each customer, where the purchase records include product information and logistics information; S2. Determine popular product data based on the co-occurrence matrix corresponding to the product information of all customers; S3. Determine the data of popular regions based on the logistics information of all customers; S4. Upload the popular product data and popular region data as customer portrait analysis results to the multi-source heterogeneous fusion database; The S2 comprises the following sub-steps: S21, removing stop words from each product information, treating all product information as product text, and extracting a co-occurrence matrix of the product text; S22, determining the chasing weight of adjacent product information in the product text according to the co-occurrence matrix of the product text; S23, determining a follow-up interval for each product information according to the chasing weights of adjacent product information in the product text; S24, determining popular product data according to the co-occurrence matrix and the follow-up interval of each product information; In S22, the pursuit weight S of the i-th product information and the i+1-th product information in the product text is i,i+1 The calculation formula is: ; In the formula, Represents the word vector of the first word in the i-th product information, Represents the word vector of the first word in the i+1th product information, Represents the word vector of the Nth word in the i+1th product information, Represents the word vector of the mth word in the i-th product information, represents the word vector of the Mth word in the i-th product information, M represents the number of words contained in the i-th product information, N represents the number of words contained in the i+1-th product information, d(·) represents the Euclidean distance function of the vector, and c0 represents an element vector randomly selected from the co-occurrence matrix.

2. The customer portrait analysis method based on logistics multi-dimensional data according to claim 1 is characterized in that: In S1, the product information includes the product name, and the logistics information includes the product delivery address.

3. The customer portrait analysis method based on logistics multi-dimensional data according to claim 1 is characterized in that: In S23, the upper limit L1 of the follow-up interval of the i-th product information is calculated as follows: Where S i,i+1 represents the pursuit weight of the i-th product information and the i+1-th product information, S i-1,i represents the pursuit weight of the i-1th product information and the i-th product information, and max(·) represents the maximum value function; The calculation formula for the lower limit L2 of the follow-up interval of the i-th product information is: ; Where min(·) represents the minimum value function.

4. The customer portrait analysis method based on logistics multi-dimensional data according to claim 1 is characterized in that: The S24 comprises the following sub-steps: S241, extracting the global representation value of the product text; S242, multiplying the global representation value by the maximum singular value of the co-occurrence matrix to obtain a global hot value; S243: The commodity information whose following interval contains the global hot value is used as hot commodity data.

5. The customer portrait analysis method based on logistics multi-dimensional data according to claim 4 is characterized in that: In S241, the calculation formula of the global representation value Total of the product text is: ; In the formula, β represents the hyperparameter, C k represents the word vector of the kth word in the product text, K represents the number of words in the product text, C1 represents the word vector of the first word in the product text, and C K It represents the word vector of the Kth word in the product text, max(·) represents the maximum value function, and min(·) represents the minimum value function.

6. The customer portrait analysis method based on logistics multi-dimensional data according to claim 1 is characterized in that: The S3 comprises the following sub-steps: S31. Complete the product delivery and receiving addresses of each customer to obtain the complete product delivery and receiving addresses; S32. Determine the popular region data based on the complete delivery and receipt addresses of all customers’ goods.

Images