Customer tendency analysis system, method and electronic device combined with follow-up status

By combining the customer tendency analysis system of the follow-up status, we collect and reduce the dimensionality follow-up data, conduct positive and liking analysis, and calculate the purchasing tendency rate in traditional methods, we solve the problem of low accuracy in customer tendency distribution in traditional methods. The problem improves the accuracy and comprehensiveness of the analysis.

CN118446730BActive Publication Date: 2024-11-05SHENZHEN DAGUANJIA SOFTWARE & TECH SERVICE CO LTD
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202410798214.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-06-19
Publication Date
2024-11-05
Estimated Expiration
2044-06-19

AI Technical Summary

Technical Problem

Traditional customer tendency analysis methods rely on single and simple data sources, resulting in low accuracy in customer tendency distribution and inability to fully capture the complex tendencies in the customer follow-up process.

Method used

Using a customer tendency analysis system that combines the follow-up status, collects follow-up data, performs dimensionality reduction processing, obtains reliable follow-up frequency and content sets, performs positive and favorite fit, and combines analysis of the change of favorite migration and purchase. Prone rate.

Benefits of technology

It improves the reliability and comprehensiveness of data source of customer tendency analysis, takes into account the shift changes to customer tendencies at different stages, and improves the accuracy of the analysis.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118446730B_ABST
    Figure CN118446730B_ABST
Patent Text Reader

Abstract

The present invention relates to a customer tendency analysis system, method and electronic device combined with a follow-up status, including: collecting follow-up frequency data and follow-up content data in the process of following up target customers; collecting historical follow-up data within a historical time and performing dimension reduction to obtain a dimension reduction follow-up frequency set and a dimension reduction follow-up content set; performing correlation dimension reduction on the dimension reduction follow-up frequency set and the dimension reduction follow-up content set according to the current follow-up product, obtaining a reliable follow-up frequency set and a reliable follow-up content set, fitting to obtain a positive degree analyzer and a like degree analyzer, respectively performing positive degree analysis and like degree analysis on the follow-up frequency data and the follow-up content data, and obtaining positive degree and like degree; performing like degree migration change analysis according to similar matching follow-up content to obtain a like correction coefficient, correcting the like degree, performing customer tendency analysis in combination with the positive degree, and obtaining a customer tendency analysis result. The present invention achieves the technical effect of improving the accuracy and comprehensiveness of customer tendency analysis.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of customer systems, and in particular to a customer tendency analysis system, method and electronic equipment combined with a follow-up status. Background Art

[0002] In modern marketing and customer relationship management, analyzing customer tendencies is the key to developing effective strategies. However, traditional customer tendency analysis methods usually rely on a single and simple analysis model, which may lead to inaccurate and incomplete results. Traditional methods often rely on limited data sources, such as simple purchase history or single customer feedback, which cannot fully capture the complex tendencies in the customer follow-up process.

[0003] Therefore, the existing technology has the problem that the analysis data source is too simple and the analysis method is relatively single, resulting in the problem of low accuracy of customer preference distribution. Summary of the invention

[0004] Aiming at the problem that the data source of customer tendency analysis in the prior art is too simple and the analysis method is relatively single, resulting in low accuracy of customer tendency distribution, the present invention proposes a customer tendency analysis system combined with follow-up status.

[0005] The present invention provides a customer tendency analysis system combined with follow-up status, comprising: collecting follow-up data in the process of following up target customers, wherein the follow-up data includes follow-up frequency data and follow-up content data;

[0006] Collecting a historical follow-up data set within a historical period, performing dimensionality reduction of the follow-up frequency on the historical follow-up data set to obtain a dimensionality-reduced follow-up frequency set, and performing dimensionality reduction of the follow-up content on the historical follow-up data set to obtain a dimensionality-reduced follow-up content set;

[0007] According to the current follow-up product, the dimension-reduced follow-up frequency set and the dimension-reduced follow-up content set are associated and dimension-reduced to obtain a reliable follow-up frequency set and a reliable follow-up content set;

[0008] According to the reliable follow-up frequency set and the reliable follow-up content set, a positivity fitting and a likeability fitting are performed to obtain a positivity analyzer and a likeability analyzer, and positivity analysis and likeability analysis are performed on the follow-up frequency data and the follow-up content data respectively to obtain positivity and likeability;

[0009] Similar matching follow-up contents are indexed in the reliable follow-up content set according to the follow-up content data, and a preference migration change analysis is performed based on the matching follow-up contents to obtain a preference correction coefficient, the preference is corrected to obtain a corrected preference, and a customer tendency analysis is performed in combination with the positivity to obtain a purchase tendency rate as a customer tendency analysis result.

[0010] In a second aspect, the present invention provides a method for analyzing customer tendency in combination with follow-up status, comprising: collecting follow-up data in the process of following up target customers, wherein the follow-up data comprises follow-up frequency data and follow-up content data;

[0011] Collecting a historical follow-up data set within a historical period, performing dimensionality reduction of the follow-up frequency on the historical follow-up data set to obtain a dimensionality-reduced follow-up frequency set, and performing dimensionality reduction of the follow-up content on the historical follow-up data set to obtain a dimensionality-reduced follow-up content set;

[0012] According to the current follow-up product, the dimension-reduced follow-up frequency set and the dimension-reduced follow-up content set are associated and dimension-reduced to obtain a reliable follow-up frequency set and a reliable follow-up content set;

[0013] According to the reliable follow-up frequency set and the reliable follow-up content set, a positivity fitting and a likeability fitting are performed to obtain a positivity analyzer and a likeability analyzer, and positivity analysis and likeability analysis are performed on the follow-up frequency data and the follow-up content data respectively to obtain positivity and likeability;

[0014] Similar matching follow-up contents are indexed in the reliable follow-up content set according to the follow-up content data, and a preference migration change analysis is performed based on the matching follow-up contents to obtain a preference correction coefficient, the preference is corrected to obtain a corrected preference, and a customer tendency analysis is performed in combination with the positivity to obtain a purchase tendency rate as a customer tendency analysis result.

[0015] In a third aspect, the present invention provides an electronic device comprising: a memory for storing a computer software program; and a processor for reading and executing the computer software program, thereby implementing the customer tendency analysis method combined with follow-up status of the second aspect.

[0016] By providing the above invention content, the beneficial effects of the present invention are: by collecting follow-up data of follow-up customers, specifically including follow-up frequency and follow-up content, and then collecting historical follow-up data sets, dimensionality reduction of follow-up frequency and dimensionality reduction of follow-up content are performed, data features are retained and data value is guaranteed, and then reliable reliable follow-up frequency sets and reliable follow-up content sets are extracted by further dimensionality reduction, and a positivity analyzer and a liking analyzer for analyzing customer positivity and liking for products are fitted, and positivity and liking are obtained by analysis, and further the liking is corrected according to the migration change of liking of matching follow-up content, and the purchase propensity rate is obtained in combination with the positivity analysis as the customer propensity analysis result, thereby improving the reliability and comprehensiveness of the customer propensity analysis data source, and considering the offset changes of the same data affecting customer propensity at different stages, thereby improving the comprehensiveness and accuracy of customer propensity analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of the structure of the customer tendency analysis system combined with the follow-up status provided by the present invention;

[0018] Figure 2 A schematic diagram of the flow of the customer tendency analysis method combined with the follow-up status provided by the present invention;

[0019] Figure 3 This is a functional block diagram of the electronic device provided by the present invention.

[0020] In the accompanying drawings, the components represented by the reference numerals are listed as follows:

[0021] Follow-up data collection module 1000, historical follow-up data dimensionality reduction module 2000, correlation dimensionality reduction module 3000, preliminary analysis module 4000, tendency analysis module 5000, electronic device 500, memory 510, processor 520, computer program 511. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0023] In the description of this application, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, "plurality" means two or more, unless otherwise clearly and specifically defined.

[0024] In the description of the present application, the term "for example" is used to mean "used as an example, illustration or description". Any embodiment described as "for example" in the present application is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any technician in the field to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid unnecessary details to obscure the description of the present invention. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in the present application.

[0025] Example 1

[0026] like Figure 1 As shown, an embodiment of the present invention provides a customer tendency analysis system combined with follow-up status, including: a follow-up data collection module 1000, a historical follow-up data dimension reduction module 2000, an associated dimension reduction module 3000, a preliminary analysis module 4000 and a tendency analysis module 5000.

[0027] In the embodiment of the present application, the follow-up data collection module 1000 is used to collect follow-up data in the process of following up target customers, wherein the follow-up data includes follow-up frequency data and follow-up content data.

[0028] In the embodiment of the present application, the target customer is the customer currently being followed up, such as a corporate customer or an individual customer, and the follow-up content is the follow-up of the target customer on a certain product, such as introducing the customer and answering the customer's questions, etc. The follow-up product is, for example, a physical product or service.

[0029] Collect follow-up data during the process of following up target customers, specifically collect follow-up frequency data and follow-up content data. Follow-up frequency data is, for example, the ratio of the number of interactions between staff and customers during the follow-up process to the entire follow-up time. Follow-up content data is, for example, the specific content of each follow-up communication with customers, such as product introductions, questions, concerns or special concerns raised by customers about the product, and customer decision-making progress, etc., as the data basis for subsequent analysis of customer tendencies.

[0030] In the embodiment of the present application, “collecting follow-up data in the process of following up target customers” includes:

[0031] Collecting the communication frequency of the target customer during the process of following up the target customer as follow-up frequency data;

[0032] The communication content of the target customer during the process of following up the target customer is collected, and key content is extracted to be used as follow-up content data.

[0033] In an embodiment of the present application, the follow-up data collected during the process of following up the target customers can be obtained by recording the follow-up activity record data such as phone calls, emails, meetings, etc. for each follow-up with the customer through a customer relationship management system (CRM). The follow-up data can also be collected through other methods.

[0034] Based on the collected follow-up data, the number and time of communication with target customers can be calculated as follow-up frequency data. Follow-up frequency data can reflect the customer's enthusiasm for the product. For example, the greater the follow-up frequency, the higher the enthusiasm.

[0035] Based on the collected follow-up data, the communication content of the target customers during the follow-up process can be obtained, such as product introduction, customer feedback on the product, questions, doubts or special concerns raised by the customer about the product, and customer decision-making progress and other information.

[0036] Furthermore, the key content of the communication content is extracted, such as using natural language processing (NLP) technology to analyze the follow-up content, converting the speech text, and then performing word segmentation, deleting stop words, and stem extraction to obtain the key content and then obtain the follow-up content data. The follow-up content data can reflect the customer's preference for the product.

[0037] For example, in a follow-up communication, the target customer feedback is "the quality of the product is very good, but the price is a bit expensive", then you can process and extract the key content of "quality", "good", "price" and "expensive" as the follow-up content data for this communication.

[0038] In this way, the follow-up frequency data and follow-up content data are obtained as key data for subsequent customer tendency analysis.

[0039] The historical follow-up data dimension reduction module 2000 is used to collect historical follow-up data sets within historical time, perform dimension reduction of follow-up frequency on the historical follow-up data sets to obtain a dimension reduced follow-up frequency set, and perform dimension reduction of follow-up content on the historical follow-up data sets to obtain a dimension reduced follow-up content set.

[0040] In the embodiment of the present application, a historical follow-up data set is collected within a historical period to serve as a data source for the current customer tendency analysis based on the follow-up data. For example, the historical follow-up data set includes complete follow-up data of multiple customer follow-ups conducted by the enterprise within the historical event, specifically including the follow-up frequency, follow-up content, and record data of whether the customer finally purchased the product for each customer follow-up. Optionally, the historical follow-up data of multiple products may be included.

[0041] The historical follow-up data set may include a lot of data with lower value. In order to improve the quality of the data source, the historical follow-up data set is subjected to dimensionality reduction of follow-up frequency and follow-up content to retain high-quality follow-up frequency data and follow-up content data. At the same time, the original data features should be retained to improve the analysis quality.

[0042] Exemplarily, the data may be reduced in dimension using principal component analysis (PCA), t-SNE or other dimensionality reduction methods.

[0043] In the embodiment of the present application, “collecting a historical follow-up data set within a historical period, performing dimensionality reduction of the follow-up frequency on the historical follow-up data set, obtaining a dimensionality-reduced follow-up frequency set, and performing dimensionality reduction of the follow-up content on the historical follow-up data set” includes:

[0044] Extracting and classifying the follow-up frequency and the follow-up content of the historical follow-up data set to obtain a historical follow-up frequency distribution and a historical follow-up content distribution, wherein the historical follow-up frequency distribution includes a plurality of different historical follow-up frequencies and the number of occurrences of the plurality of historical follow-up frequencies;

[0045] Performing dimension reduction of the follow-up frequency on the historical follow-up frequency distribution to obtain a dimension-reduced follow-up frequency set, wherein the dimension reduction is performed to improve the feature retention of the historical follow-up frequency distribution and the number of occurrences of the dimension-reduced follow-up frequency;

[0046] The dimension of the follow-up content is reduced for the historical follow-up content distribution to obtain a dimension-reduced follow-up content set.

[0047] In an embodiment of the present application, all historical follow-up data within a historical period can be collected based on the enterprise's CRM system, automated follow-up tools, and voice transcription system, and the data can be further sorted and cleaned to remove duplicate and invalid data, such as historical follow-up data with missing follow-up content, and finally a historical follow-up data set can be obtained to ensure the comprehensiveness of the data.

[0048] The follow-up frequency and follow-up content of the complete historical follow-up data of each follow-up in the historical follow-up data set are extracted and classified. Specifically, the same follow-up frequency is classified into a category of follow-up frequency, and the same follow-up content is classified into a category of follow-up content. In this way, all multiple different historical follow-up frequencies and multiple different historical follow-up contents are obtained.

[0049] Furthermore, according to the multiple different historical follow-up frequencies and the multiple different historical follow-up contents, the number of occurrences of the multiple different historical follow-up frequencies and the multiple different historical follow-up contents in the historical follow-up data set is counted to construct the historical follow-up frequency distribution and the historical follow-up content distribution. The historical follow-up frequency distribution includes multiple different historical follow-up frequencies and the number of occurrences of each historical follow-up frequency, and the historical follow-up content distribution includes multiple different historical follow-up contents and the number of occurrences of the multiple historical follow-up contents.

[0050] The dimensionality reduction is performed on the historical follow-up frequency distribution in order to improve the feature retention of the historical follow-up frequency distribution and the number of occurrences of the dimensionality reduction follow-up frequency. The dimensionality reduction of the historical follow-up frequency distribution is performed. Among them, the different multiple historical follow-up frequencies in the historical follow-up frequency distribution can reflect the distribution characteristics of the follow-up frequency in the process of following up customers, and the number of occurrences of each historical follow-up frequency can reflect the data value of each historical follow-up frequency. The more historical follow-up frequencies that appear, the more frequently they appear in the follow-up customers, and the higher the data value is when conducting customer tendency analysis. Therefore, considering the purposes of two directions, the dimensionality reduction of the historical follow-up frequency distribution is performed to obtain the dimensionality reduction follow-up frequency set.

[0051] Based on the same consideration, the dimension of the historical follow-up content distribution is reduced by improving the feature retention of the historical follow-up content distribution and the number of occurrences of the reduced-dimensional follow-up content, and the dimension of the historical follow-up content distribution is reduced to obtain a reduced-dimensional follow-up content set.

[0052] In the embodiment of the present application, “reducing the dimension of the historical follow-up frequency distribution to obtain a reduced-dimensional follow-up frequency set” includes:

[0053] Selecting a basic follow-up frequency in the historical follow-up frequency distribution, calculating the deviation between other historical follow-up frequencies and the basic follow-up frequency and performing distribution probability allocation to obtain a historical frequency probability distribution;

[0054] According to the number of dimensionality reduction, randomly select dimensionality reduction for the historical follow-up frequency distribution to obtain a first dimensionality reduction follow-up frequency set, calculate the deviation between each first dimensionality reduction follow-up frequency in the first dimensionality reduction follow-up frequency set and the basic follow-up frequency and perform distribution probability allocation to obtain a first dimensionality reduction frequency probability distribution;

[0055] Construct a frequency dimension reduction function to optimize the dimension reduction of the historical follow-up frequency distribution, as follows:

[0056] ;

[0057] Among them, FDR is the frequency dimension reduction fitness, and is the weight, SIM is the feature retention of the dimensionality reduction frequency probability distribution and the historical frequency probability distribution, M is the number of dimensionality reduction, is the number of occurrences of the i-th dimension reduction follow-up frequency in the dimension reduction follow-up frequency set, The number of base occurrences for the base follow-up frequency;

[0058] Calculate similarity based on the first dimension reduction frequency probability distribution and the historical frequency probability distribution to obtain a first feature retention degree, and calculate and obtain a first frequency dimension reduction fitness of the first dimension reduction follow-up frequency set according to the frequency dimension reduction function;

[0059] Continue to iteratively reduce the dimension of the historical follow-up frequency distribution until convergence, and obtain a reduced-dimensional follow-up frequency set.

[0060] In the embodiment of the present application, a basic follow-up frequency is selected in the historical follow-up frequency distribution and used as a benchmark, which can help measure the deviation and change of other follow-up frequencies and provide a reference point for subsequent analysis. Exemplarily, the median historical follow-up frequency in the historical follow-up frequency distribution is selected as the basic follow-up frequency, or the average historical follow-up frequency is used as the basic follow-up frequency, or the historical follow-up frequency with the most occurrences is used as the basic follow-up frequency.

[0061] Furthermore, the deviations between other historical follow-up frequencies and the basic follow-up frequency in the historical follow-up frequency distribution are calculated, for example, the difference in follow-up frequencies is calculated, and then the distribution probability is allocated according to the deviation. For example, the maximum deviation is calculated, and the distribution probability corresponding to the maximum deviation is allocated to 0, and the distribution probability of the basic follow-up frequency itself (that is, the deviation is 0) is allocated to 1. Further, probability allocation is performed for multiple other follow-up frequency deviations. For example, when the deviation is half of the maximum deviation, the allocated distribution probability is 0.5.

[0062] In this way, multiple distribution probabilities can be obtained by assigning multiple historical follow-up frequencies in the historical follow-up frequency distribution, thereby obtaining a historical frequency probability distribution.

[0063] According to the number of dimensionality reduction, randomly select dimensionality reduction from multiple historical follow-up frequencies in the historical follow-up frequency distribution to obtain the first dimensionality reduction follow-up frequency set after dimensionality reduction. The number of dimensionality reduction is the number of data after dimensionality reduction, for example, 1000, and can also be set according to computing power and needs.

[0064] Furthermore, based on the same method, the deviation between each first dimension reduction follow-up frequency and the basic follow-up frequency in the first dimension reduction follow-up frequency set is calculated and the distribution probability is assigned to obtain the first dimension reduction frequency probability distribution. By calculating the frequency deviation and distribution probability after dimension reduction, it is helpful to evaluate the dimension reduction effect and provide data support for subsequent optimization.

[0065] Construct a frequency dimension reduction function to optimize the dimension reduction of the historical follow-up frequency distribution, as follows:

[0066] ;

[0067] Among them, FDR is the frequency dimensionality reduction fitness, which can reflect the quality of the dimensionality reduction result. The larger the frequency dimensionality reduction fitness, the better the dimensionality reduction effect. and is the weight, and The sum of is 1, which can be set according to the optimization purpose. SIM is the feature retention degree of the dimensionality reduction frequency probability distribution and the historical frequency probability distribution, specifically the similarity between the dimensionality reduction frequency probability distribution and the historical frequency probability distribution. M is the number of dimensionality reduction. is the number of occurrences of the i-th dimension reduction follow-up frequency in the dimension reduction follow-up frequency set, The basic number of occurrences for the basic follow-up frequency.

[0068] The similarity is calculated based on the first dimensionality reduction frequency probability distribution and the historical frequency probability distribution. The similarity calculation can be performed using methods in the prior art, such as calculating the KL divergence or JS divergence or Hellinger distance of the first dimensionality reduction frequency probability distribution and the historical frequency probability distribution to obtain the similarity of the two frequency probability distributions.

[0069] The similarity between the first dimensionality reduction frequency probability distribution and the historical frequency probability distribution is taken as the first feature retention degree, and then the first frequency dimensionality reduction fitness of the first dimensionality reduction follow-up frequency set is calculated based on the frequency dimensionality reduction function according to the number of occurrences of each first dimensionality reduction follow-up frequency in the historical follow-up frequency distribution and the basic number of occurrences of the basic follow-up frequency.

[0070] Furthermore, the historical follow-up frequency distribution is randomly selected for dimensionality reduction according to the number of dimensionality reduction, a second dimensionality reduction follow-up frequency set is obtained, and the dimensionality reduction fitness of the second frequency is calculated.

[0071] Continue to iterate the dimensionality reduction of the historical follow-up frequency distribution until convergence, for example, the number of iterative dimensionality reductions that reach convergence is 50 times, complete the dimensionality reduction optimization, and obtain the final dimensionality reduction result with the maximum frequency dimensionality reduction fitness, that is, the final dimensionality reduction follow-up frequency set.

[0072] The embodiment of the present application constructs a frequency dimensionality reduction function and uses the dimensionality reduction concept of t-SNE to reduce the dimensionality of the historical follow-up frequency distribution, retaining the data characteristics of the historical follow-up frequency and at the same time trying to improve the data value of the reduced dimensionality follow-up frequency as a data source for subsequent customer tendency analysis, thereby improving the accuracy and comprehensiveness of the analysis.

[0073] In the above content, the dimensionality reduction of the historical follow-up frequency distribution is taken as an example to illustrate the principle of dimensionality reduction. The dimensionality reduction steps of the historical follow-up content distribution are the same as those of the historical follow-up frequency distribution, but the data content is different. Among them, the historical follow-up content with the largest number of occurrences can be selected as the basic follow-up content. The historical follow-up content can be analyzed by natural language processing (NLP), latent semantic analysis (LSA) and other methods, and then the deviation from the theme in the basic follow-up content is analyzed, and the distribution probability is allocated for dimensionality reduction. The specific content of dimensionality reduction will not be repeated in detail.

[0074] The associated dimension reduction module 3000 is used to perform associated dimension reduction on the reduced dimension follow-up frequency set and the reduced dimension follow-up content set according to the current follow-up product, so as to obtain a reliable follow-up frequency set and a reliable follow-up content set.

[0075] In an embodiment of the present application, after the initial dimensionality reduction to obtain the dimensionality reduction follow-up frequency set and the dimensionality reduction follow-up content set, the dimensionality reduction follow-up frequency set and the dimensionality reduction follow-up content set may include the follow-up frequency and follow-up content when following up on multiple products. There are differences in the follow-up data of different products. In order to further improve the data quality, the dimensionality reduction follow-up frequency set and the dimensionality reduction follow-up content set are associated and reduced in dimension according to the current follow-up products of the target customers to obtain the follow-up frequency and follow-up content of similar or identical products to improve the data quality.

[0076] In the embodiment of the present application, “according to the current follow-up product, the dimension reduction follow-up frequency set and the dimension reduction follow-up content set are associated and dimension reduced to obtain a reliable follow-up frequency set and a reliable follow-up content set” includes:

[0077] Collect product feature information of the follow-up products currently following up the target customers;

[0078] Extracting multiple historical product feature information of multiple follow-up products corresponding to the dimensionality reduction follow-up frequency and the dimensionality reduction follow-up content in the dimensionality reduction follow-up frequency set and the dimensionality reduction follow-up content set;

[0079] Analyze and obtain the plurality of historical product feature information and the plurality of product-related parameters of the product feature information;

[0080] According to multiple product association parameters, the dimensionality reduction follow-up frequency and dimensionality reduction follow-up content with the largest association quantity threshold of product association parameters are selected from the dimensionality reduction follow-up frequency set and the dimensionality reduction follow-up content set respectively, and a reliable follow-up frequency set and a reliable follow-up content set are obtained by output.

[0081] In an embodiment of the present application, product feature information of the follow-up products of the current follow-up target customers is collected, for example, data such as the category, function, brand, specification, price, etc. of the follow-up products are collected as product feature information to distinguish different products.

[0082] Extract multiple historical product feature information of multiple historical follow-up products corresponding to the historical follow-up data corresponding to the dimensionality reduction follow-up frequency and the dimensionality reduction follow-up content in the dimensionality reduction follow-up frequency set and the dimensionality reduction follow-up content set.

[0083] Furthermore, based on multiple historical product feature information and current product feature information, product association analysis is performed, for example, the similarity between the product feature information and multiple historical product feature information is analyzed and calculated. For example, Euclidean distance, cosine similarity, Jaccard similarity coefficient and other methods can be used for analysis and calculation, and the similarity is used as a product association parameter to obtain multiple product association parameters.

[0084] According to multiple product association parameters, historical follow-up products with the largest association quantity threshold of product association parameters are selected as multiple associated follow-up products. The association quantity threshold is set according to the amount of data required for the final model fitting training, for example, 500, and can also be set by yourself.

[0085] Select the dimensionality reduction follow-up frequencies and dimensionality reduction follow-up contents in the follow-up data corresponding to multiple related follow-up products in the dimensionality reduction follow-up frequency set and the dimensionality reduction follow-up content set respectively as the dimensionality reduction follow-up frequencies and dimensionality reduction follow-up contents of products similar to the current follow-up product, and output a reliable follow-up frequency set and a reliable follow-up content set.

[0086] The embodiment of the present application performs associated dimensionality reduction on the dimensionality reduction follow-up frequency set and the dimensionality reduction follow-up content set according to product feature information to obtain the follow-up frequency and follow-up content of similar or identical products, and conducts subsequent customer tendency analysis, thereby further improving the quality of the analysis data source.

[0087] In an embodiment of the present application, the preliminary analysis module 4000 is used to perform positivity fitting and likeability fitting according to the reliable follow-up frequency set and the reliable follow-up content set, obtain a positivity analyzer and a likeability analyzer, and perform positivity analysis and likeability analysis on the follow-up frequency data and the follow-up content data, respectively, to obtain positivity and likeability.

[0088] In the embodiment of the present application, the reliable follow-up frequency set and the reliable follow-up content set are the data sources for customer tendency analysis. The reliable follow-up frequency set and the reliable follow-up content set are used to fit and train a positivity analyzer and a likeability analyzer for performing positivity analysis and likeability analysis, respectively. The positivity analysis and likeability analysis are performed on the follow-up frequency data and follow-up content data currently collected based on the target customers to obtain the positivity and likeability.

[0089] Among them, the positivity and likeability can be identified based on the customer's purchase probability when different follow-up frequencies and follow-up contents appear, and then the positivity analyzer and the likeability analyzer can be fitted and trained.

[0090] In the embodiment of the present application, “according to the reliable follow-up frequency set and the reliable follow-up content set, performing positivity fitting and likeability fitting, obtaining a positivity analyzer and a likeability analyzer, and performing positivity analysis and likeability analysis on the follow-up frequency data and the follow-up content data, respectively” includes:

[0091] Extract the number of occurrences of each reliable follow-up frequency in the reliable follow-up frequency set in the historical follow-up frequency distribution, and the number of purchases when each reliable follow-up frequency appears, and calculate to obtain a sample positivity set;

[0092] Extracting the number of occurrences of each reliable follow-up content in the reliable follow-up content set in the historical follow-up content distribution and the number of purchases when each reliable follow-up content appears, and calculating to obtain a sample preference set;

[0093] Using the reliable follow-up frequency set and the sample positivity set as fitting data, performing fitting training, and obtaining a positivity analyzer;

[0094] Using the reliable follow-up content set and the sample likeability set to perform fitting training to obtain a likeability analyzer;

[0095] The positivity analyzer and the likeability analyzer are respectively used to perform positivity analysis and likeability analysis on the follow-up frequency data and the follow-up content data to obtain positivity and likeability.

[0096] In an embodiment of the present application, the number of occurrences of each reliable follow-up frequency in the reliable follow-up frequency set in the historical follow-up frequency distribution is extracted, that is, the total number of occurrences of each reliable follow-up frequency in the historical customer follow-ups, and the number of purchases when each reliable follow-up frequency appears, and the ratio of the number of purchases to the number of occurrences is calculated as the positivity. In this way, multiple positivity of multiple reliable follow-up frequencies in the reliable follow-up frequency set is calculated to obtain a sample positivity set.

[0097] The number of occurrences of each reliable follow-up content in the reliable follow-up content set in the historical follow-up content distribution and the number of purchases when each reliable follow-up content appears are extracted, and the ratio of the number of purchases and the number of occurrences when each reliable follow-up content appears is calculated to obtain a sample preference set.

[0098] Using reliable follow-up frequency sets and sample positivity sets as fitting data, and using appropriate machine learning algorithms (such as linear regression, logistic regression, support vector machine, etc.), we conduct fitting training for positivity analysis to predict and analyze the customer's positivity towards the product and obtain a positivity analyzer.

[0099] Based on the same training fitting method, a reliable follow-up content set and a sample likeness set are used as fitting training data, and fitting training is performed to obtain a likeability analyzer.

[0100] In this way, the positivity analyzer and the likeability analyzer are respectively used to perform positivity analysis and likeability analysis on the follow-up frequency data and the follow-up content data of the current follow-up target customer, and the positivity and likeability are obtained as the analysis results of the current target customer.

[0101] In an embodiment of the present application, the tendency analysis module 5000 is used to index similar matching follow-up content in the reliable follow-up content set according to the follow-up content data, and perform a preference migration change analysis based on the matching follow-up content to obtain a preference correction coefficient, correct the preference to obtain a corrected preference, and perform customer tendency analysis in combination with the positivity to obtain a purchase tendency rate as a customer tendency analysis result.

[0102] In the embodiment of the present application, the same follow-up frequency in different periods can more accurately and repeatedly reflect the customer's enthusiasm for the product, but when the customer is at different stages of product purchase quantity, the same follow-up content may reflect the customer's preference for the product. For example, when the customer just starts to buy the product, even if the customer's evaluation of the product in the follow-up content is average, there is a certain purchase probability. However, after the customer purchases a part of the product, due to the product inventory, the customer's evaluation of the product in the follow-up content is average, and the purchase rate is low and the preference is small. Therefore, it is necessary to correct the preference according to the customer's product purchase quantity stage.

[0103] Among them, similar matching follow-up contents are indexed in the reliable follow-up content set according to the follow-up content data, and the migration change analysis of customers' preference at different product purchase quantity stages is carried out based on the similar matching follow-up contents to obtain the preference correction coefficient, correct the preference, and obtain the corrected preference.

[0104] Based on the corrected modified liking degree combined with the aforementioned positivity, customer tendency analysis is performed to predict the probability of obtaining target customers to purchase follow-up products, that is, the purchase tendency rate, as the customer tendency analysis result.

[0105] In the embodiment of the present application, “indexing similar matching follow-up content in the reliable follow-up content set according to the follow-up content data, and performing a likeability migration change analysis based on the matching follow-up content to obtain a likeability correction coefficient” includes:

[0106] Analyzing the similarity between each reliable follow-up content in the reliable follow-up content set and the follow-up content data, and outputting the reliable follow-up content with a similarity greater than a threshold as a plurality of matching follow-up contents;

[0107] According to the multiple matching follow-up contents, the purchase rate of each matching follow-up content appearing in different purchase volume stages is indexed in the historical follow-up data set to obtain multiple likeability correction tables;

[0108] Extracting the current purchase volume stage of the target customer, indexing in the plurality of likeability correction tables, obtaining a plurality of matching likeability correction coefficients, and calculating the likeability correction coefficients;

[0109] The likeability correction coefficient is used to perform correction calculation on the likeability to obtain a corrected likeability.

[0110] In an embodiment of the present application, the similarity between each reliable follow-up content in the reliable follow-up content set and the current follow-up content data is analyzed. For example, the meaning of each reliable follow-up content and the current follow-up content data can be understood based on a natural language analysis method, and then the meaning similarity is analyzed. Optionally, a text similarity algorithm (such as cosine similarity, Jaccard similarity coefficient, TF-IDF, etc.) can also be used to calculate the similarity between the follow-up contents. Reliable follow-up contents with similarities greater than a similarity threshold are output as multiple matching follow-up contents. The similarity threshold can be set according to the similarity distribution, for example, the smallest similarity within the first 30% of similarities is used as the similarity threshold.

[0111] According to multiple matching follow-up contents, the purchase rate of each matching follow-up content appearing in different purchase volume stages is indexed in the historical follow-up data set. For example, the purchase rate of a matching follow-up content appearing in multiple stages such as the 0-10 stage and the 10-20 stage of the customer's product purchase volume is extracted, and the change ratio of the purchase rate in different stages of the customer's product purchase volume is further calculated as the likeability correction coefficient. In this way, multiple likeability correction tables for multiple matching follow-up contents can be constructed, and each likeability correction table includes a mapping relationship between multiple purchase volume stages and multiple likeability correction coefficients.

[0112] Based on the enterprise's customer system, etc., extract the target customer's product purchase quantity, obtain the target customer's current purchase volume stage, index in multiple like correction tables, obtain the corresponding multiple matching like correction coefficients, and further calculate the mean to obtain the like correction coefficient.

[0113] Optionally, based on the similarities between the multiple matching follow-up contents and the follow-up content data, a weighted calculation may be performed on multiple matching like correction coefficients to obtain a final like correction coefficient.

[0114] The likeability correction coefficient is used to correct the likeability obtained in the above analysis to obtain the corrected likeability. This takes into account the deviation of the follow-up content data in reflecting the customer's likeability of the product at different stages of the customer's product purchase volume, and improves the accuracy of the final customer tendency analysis.

[0115] In the embodiment of the present application, “performing customer propensity analysis to obtain a purchase propensity rate” includes:

[0116] Collect sample positivity set, sample modified liking set, and sample purchase tendency rate set;

[0117] Using the sample positivity set, the sample modified liking set and the sample purchase tendency rate set to train a purchase tendency analyzer;

[0118] Based on the purchase tendency analyzer, the purchase tendency analysis is performed on the positivity and the modified preference to obtain a purchase tendency rate.

[0119] In the embodiment of the present application, based on the historical follow-up data set and processing method in the aforementioned content, processing can also be performed based on the corresponding reliable follow-up frequency set and reliable follow-up content set, and each follow-up data is processed to collect the sample positivity set and the sample corrected liking set, and collect the purchase probability of the user purchasing the product when different sample positivity and sample corrected liking appear, and obtain the sample purchase propensity rate set.

[0120] The sample positivity set, sample modified liking set and sample purchase propensity rate set are used as training data, and an appropriate machine learning model, such as a feedforward neural network, is selected to train a purchase propensity analyzer for predicting purchase probability based on positivity and modified liking. During the training process, the model parameters of the purchase propensity analyzer are continuously adjusted to improve the accuracy until the convergence requirements are met, such as an accuracy rate of 93%.

[0121] The trained purchase tendency analyzer is used to perform purchase tendency analysis on the current positivity and modified preference, and a purchase tendency rate is obtained as a customer tendency analysis result of the customer tendency analysis.

[0122] Based on the above specific implementation methods disclosed, the beneficial effects of the present invention are: by collecting follow-up data of customers, specifically including follow-up frequency and follow-up content, and then collecting historical follow-up data sets, dimensionality reduction of follow-up frequency and dimensionality reduction of follow-up content are performed, data features are retained and data value is guaranteed, and then reliable reliable follow-up frequency sets and reliable follow-up content sets are extracted by further dimensionality reduction, and a positivity analyzer and a liking analyzer for analyzing customer positivity and liking for products are fitted, and positivity and liking are obtained by analysis, and the liking is further corrected according to the migration change of liking of matching follow-up content, and the purchase propensity rate is obtained in combination with the positivity analysis as the customer propensity analysis result, thereby improving the reliability and comprehensiveness of the customer propensity analysis data source, and considering the offset changes of the same data affecting customer propensity at different stages, thereby improving the comprehensiveness and accuracy of customer propensity analysis.

[0123] Example 2

[0124] like Figure 2 As shown, based on the same inventive concept as the customer tendency analysis system combined with the follow-up status provided in Example 1, the embodiment of the present invention provides a customer tendency analysis method combined with the follow-up status, including:

[0125] Collect follow-up data in the process of following up target customers, where the follow-up data includes follow-up frequency data and follow-up content data;

[0126] Collecting a historical follow-up data set within a historical period, performing dimensionality reduction of the follow-up frequency on the historical follow-up data set to obtain a dimensionality-reduced follow-up frequency set, and performing dimensionality reduction of the follow-up content on the historical follow-up data set to obtain a dimensionality-reduced follow-up content set;

[0127] According to the current follow-up product, the dimension-reduced follow-up frequency set and the dimension-reduced follow-up content set are associated and dimension-reduced to obtain a reliable follow-up frequency set and a reliable follow-up content set;

[0128] According to the reliable follow-up frequency set and the reliable follow-up content set, a positivity fitting and a likeability fitting are performed to obtain a positivity analyzer and a likeability analyzer, and positivity analysis and likeability analysis are performed on the follow-up frequency data and the follow-up content data respectively to obtain positivity and likeability;

[0129] Similar matching follow-up contents are indexed in the reliable follow-up content set according to the follow-up content data, and a preference migration change analysis is performed based on the matching follow-up contents to obtain a preference correction coefficient, the preference is corrected to obtain a corrected preference, and a customer tendency analysis is performed in combination with the positivity to obtain a purchase tendency rate as a customer tendency analysis result.

[0130] Furthermore, follow-up data is collected during the process of following up with target customers, including:

[0131] Collecting the communication frequency of the target customer during the process of following up the target customer as follow-up frequency data;

[0132] The communication content of the target customer during the process of following up the target customer is collected, and key content is extracted to be used as follow-up content data.

[0133] Further, collecting a historical follow-up data set within a historical time, performing dimension reduction of follow-up frequency on the historical follow-up data set to obtain a dimension-reduced follow-up frequency set, and performing dimension reduction of follow-up content on the historical follow-up data set, including:

[0134] Extracting and classifying the follow-up frequency and the follow-up content of the historical follow-up data set to obtain a historical follow-up frequency distribution and a historical follow-up content distribution, wherein the historical follow-up frequency distribution includes a plurality of different historical follow-up frequencies and the number of occurrences of the plurality of historical follow-up frequencies;

[0135] Performing dimension reduction of the follow-up frequency on the historical follow-up frequency distribution to obtain a dimension-reduced follow-up frequency set, wherein the dimension reduction is performed to improve the feature retention of the historical follow-up frequency distribution and the number of occurrences of the dimension-reduced follow-up frequency;

[0136] The dimension of the follow-up content is reduced for the historical follow-up content distribution to obtain a dimension-reduced follow-up content set.

[0137] Furthermore, the dimension of the follow-up frequency distribution is reduced to obtain a dimension-reduced follow-up frequency set, including:

[0138] Selecting a basic follow-up frequency in the historical follow-up frequency distribution, calculating the deviation between other historical follow-up frequencies and the basic follow-up frequency and performing distribution probability allocation to obtain a historical frequency probability distribution;

[0139] According to the number of dimensionality reduction, randomly select dimensionality reduction for the historical follow-up frequency distribution to obtain a first dimensionality reduction follow-up frequency set, calculate the deviation between each first dimensionality reduction follow-up frequency in the first dimensionality reduction follow-up frequency set and the basic follow-up frequency and perform distribution probability allocation to obtain a first dimensionality reduction frequency probability distribution;

[0140] Construct a frequency dimension reduction function to optimize the dimension reduction of the historical follow-up frequency distribution, as follows:

[0141] ;

[0142] Among them, FDR is the frequency dimension reduction fitness, and is the weight, SIM is the feature retention of the dimensionality reduction frequency probability distribution and the historical frequency probability distribution, M is the number of dimensionality reduction, is the number of occurrences of the i-th dimension reduction follow-up frequency in the dimension reduction follow-up frequency set, The number of base occurrences for the base follow-up frequency;

[0143] Calculate similarity based on the first dimension reduction frequency probability distribution and the historical frequency probability distribution to obtain a first feature retention degree, and calculate and obtain a first frequency dimension reduction fitness of the first dimension reduction follow-up frequency set according to the frequency dimension reduction function;

[0144] Continue to iteratively reduce the dimension of the historical follow-up frequency distribution until convergence, and obtain a reduced-dimensional follow-up frequency set.

[0145] Furthermore, according to the current follow-up product, the dimension-reduced follow-up frequency set and the dimension-reduced follow-up content set are associated and dimension-reduced to obtain a reliable follow-up frequency set and a reliable follow-up content set, including:

[0146] Collect product feature information of the follow-up products currently following up the target customers;

[0147] Extracting multiple historical product feature information of multiple follow-up products corresponding to the dimensionality reduction follow-up frequency and the dimensionality reduction follow-up content in the dimensionality reduction follow-up frequency set and the dimensionality reduction follow-up content set;

[0148] Analyze and obtain the plurality of historical product feature information and the plurality of product-related parameters of the product feature information;

[0149] According to multiple product association parameters, the dimensionality reduction follow-up frequency and dimensionality reduction follow-up content with the largest association quantity threshold of product association parameters are selected from the dimensionality reduction follow-up frequency set and the dimensionality reduction follow-up content set respectively, and a reliable follow-up frequency set and a reliable follow-up content set are obtained by output.

[0150] Further, according to the reliable follow-up frequency set and the reliable follow-up content set, positivity fitting and likeability fitting are performed to obtain a positivity analyzer and a likeability analyzer, and positivity analysis and likeability analysis are performed on the follow-up frequency data and the follow-up content data respectively, including:

[0151] Extract the number of occurrences of each reliable follow-up frequency in the reliable follow-up frequency set in the historical follow-up frequency distribution, and the number of purchases when each reliable follow-up frequency appears, and calculate to obtain a sample positivity set;

[0152] Extracting the number of occurrences of each reliable follow-up content in the reliable follow-up content set in the historical follow-up content distribution and the number of purchases when each reliable follow-up content appears, and calculating to obtain a sample preference set;

[0153] Using the reliable follow-up frequency set and the sample positivity set as fitting data, performing fitting training, and obtaining a positivity analyzer;

[0154] Using the reliable follow-up content set and the sample likeability set to perform fitting training to obtain a likeability analyzer;

[0155] The positivity analyzer and the likeability analyzer are respectively used to perform positivity analysis and likeability analysis on the follow-up frequency data and the follow-up content data to obtain positivity and likeability.

[0156] Furthermore, similar matching follow-up content is indexed in the reliable follow-up content set according to the follow-up content data, and a preference modification coefficient is obtained by performing a preference migration change analysis based on the matching follow-up content of the same customer, including:

[0157] Analyzing the similarity between each reliable follow-up content in the reliable follow-up content set and the follow-up content data, and outputting the reliable follow-up content with a similarity greater than a threshold as a plurality of matching follow-up contents;

[0158] According to the multiple matching follow-up contents, the purchase rate of each matching follow-up content appearing in different purchase volume stages is indexed in the historical follow-up data set to obtain multiple likeability correction tables;

[0159] Extracting the current purchase volume stage of the target customer, indexing in the plurality of likeability correction tables, obtaining a plurality of matching likeability correction coefficients, and calculating the likeability correction coefficients;

[0160] The likeability correction coefficient is used to perform correction calculation on the likeability to obtain a corrected likeability.

[0161] Further, conduct customer propensity analysis to obtain purchase propensity rate, including:

[0162] Collect sample positivity set, sample modified liking set, and sample purchase tendency rate set;

[0163] Using the sample positivity set, the sample modified liking set and the sample purchase tendency rate set to train a purchase tendency analyzer;

[0164] Based on the purchase tendency analyzer, the purchase tendency analysis is performed on the positivity and the modified preference to obtain a purchase tendency rate.

[0165] See also Figure 3 , Figure 3 Schematic diagram of an electronic device provided by an embodiment of the present invention. Figure 3 As shown, an embodiment of the present invention provides an electronic device 500, including a memory 510, a processor 520, and a computer program 511 stored in the memory 510 and executable on the processor 520. When the processor 520 executes the computer program 511, the following steps are implemented: collecting follow-up data in the process of following up target customers, wherein the follow-up data includes follow-up frequency data and follow-up content data; collecting a historical follow-up data set within a historical time, performing dimension reduction of follow-up frequency on the historical follow-up data set to obtain a dimension reduction follow-up frequency set, and performing dimension reduction of follow-up content on the historical follow-up data set to obtain a dimension reduction follow-up content set; and performing dimension reduction of follow-up content on the historical follow-up data set to obtain a dimension reduction follow-up content set according to the current follow-up product. The content set is subjected to correlation dimensionality reduction to obtain a reliable follow-up frequency set and a reliable follow-up content set; according to the reliable follow-up frequency set and the reliable follow-up content set, positivity fitting and likeness fitting are performed to obtain a positivity analyzer and a likeness analyzer, and positivity analysis and likeness analysis are performed on the follow-up frequency data and the follow-up content data respectively to obtain positivity and likeness; according to the follow-up content data, matching follow-up contents with similar indexes are performed in the reliable follow-up content set, and a likeness migration change analysis is performed based on the matching follow-up contents to obtain a likeness correction coefficient, the likeness is corrected to obtain a corrected likeness, and a customer tendency analysis is performed in combination with the positivity to obtain a purchase tendency rate as a customer tendency analysis result.

[0166] It should be noted that in the above embodiments, the description of each embodiment has its own emphasis, and for parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0167] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0168] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0169] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0170] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0171] Although preferred embodiments of the present invention have been described, additional changes and modifications may occur to these embodiments once those skilled in the art understand the basic inventive concepts.

[0172] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention belong to the scope of the present invention and its equivalent technologies, the present invention is also intended to include these changes and variations.

Claims

1. A customer tendency analysis system combined with follow-up status, characterized in that: The system comprises: A follow-up data collection module is used to collect follow-up data in the process of following up target customers, wherein the follow-up data includes follow-up frequency data and follow-up content data; The historical follow-up data dimension reduction module is used to collect the historical follow-up data set within the historical time, perform dimension reduction of the follow-up frequency on the historical follow-up data set to obtain a dimension-reduced follow-up frequency set, and perform dimension reduction of the follow-up content on the historical follow-up data set to obtain a dimension-reduced follow-up content set; An associated dimension reduction module is used to perform associated dimension reduction on the dimension reduction follow-up frequency set and the dimension reduction follow-up content set according to the current follow-up product, so as to obtain a reliable follow-up frequency set and a reliable follow-up content set; A preliminary analysis module, used to perform positivity fitting and likeability fitting according to the reliable follow-up frequency set and the reliable follow-up content set, obtain a positivity analyzer and a likeability analyzer, and perform positivity analysis and likeability analysis on the follow-up frequency data and the follow-up content data, respectively, to obtain positivity and likeability; a tendency analysis module, for indexing similar matching follow-up contents in the reliable follow-up content set according to the follow-up content data, performing a preference migration change analysis based on the matching follow-up contents to obtain a preference correction coefficient, correcting the preference to obtain a corrected preference, and performing customer tendency analysis in combination with the positivity to obtain a purchase tendency rate as a customer tendency analysis result; The process of collecting a historical follow-up data set within a historical period, reducing the dimension of the follow-up frequency of the historical follow-up data set to obtain a reduced-dimensional follow-up frequency set, and reducing the dimension of the follow-up content of the historical follow-up data set includes: Extracting and classifying the follow-up frequency and the follow-up content of the historical follow-up data set to obtain a historical follow-up frequency distribution and a historical follow-up content distribution, wherein the historical follow-up frequency distribution includes a plurality of different historical follow-up frequencies and the number of occurrences of the plurality of historical follow-up frequencies; Performing dimension reduction of the follow-up frequency on the historical follow-up frequency distribution to obtain a dimension-reduced follow-up frequency set, wherein the dimension reduction is performed to improve the feature retention of the historical follow-up frequency distribution and the number of occurrences of the dimension-reduced follow-up frequency; Performing dimensionality reduction on the historical follow-up content distribution to obtain a dimensionality-reduced follow-up content set; According to the reliable follow-up frequency set and the reliable follow-up content set, a positivity fitting and a likeability fitting are performed to obtain a positivity analyzer and a likeability analyzer, and positivity analysis and likeability analysis are performed on the follow-up frequency data and the follow-up content data, respectively, including: Extract the number of occurrences of each reliable follow-up frequency in the reliable follow-up frequency set in the historical follow-up frequency distribution, and the number of purchases when each reliable follow-up frequency appears, and calculate to obtain a sample positivity set; Extracting the number of occurrences of each reliable follow-up content in the reliable follow-up content set in the historical follow-up content distribution and the number of purchases when each reliable follow-up content appears, and calculating to obtain a sample preference set; Using the reliable follow-up frequency set and the sample positivity set as fitting data, performing fitting training, and obtaining a positivity analyzer; Using the reliable follow-up content set and the sample likeability set to perform fitting training to obtain a likeability analyzer; Using the positivity analyzer and the likeability analyzer to perform positivity analysis and likeability analysis on the follow-up frequency data and the follow-up content data, respectively, to obtain positivity and likeability; Among them, indexing similar matching follow-up content in the reliable follow-up content set according to the follow-up content data, and performing a preference degree migration change analysis based on the matching follow-up content to obtain a preference correction coefficient, including: Analyzing the similarity between each reliable follow-up content in the reliable follow-up content set and the follow-up content data, and outputting the reliable follow-up content with a similarity greater than a threshold as a plurality of matching follow-up contents; According to the multiple matching follow-up contents, the purchase rate of each matching follow-up content appearing in different purchase volume stages is indexed in the historical follow-up data set to obtain multiple likeability correction tables; Extracting the current purchase volume stage of the target customer, indexing in the plurality of likeability correction tables, obtaining a plurality of matching likeability correction coefficients, and calculating the likeability correction coefficients; The likeability correction coefficient is used to perform correction calculation on the likeability to obtain a corrected likeability.

2. The system according to claim 1, characterized in that Collect follow-up data during the process of following up with target customers, including: Collecting the communication frequency of the target customer during the process of following up the target customer as follow-up frequency data; The communication content of the target customer during the process of following up the target customer is collected, and key content is extracted to be used as follow-up content data.

3. The system according to claim 1, characterized in that Performing dimension reduction of the follow-up frequency on the historical follow-up frequency distribution to obtain a dimension-reduced follow-up frequency set, including: Selecting a basic follow-up frequency in the historical follow-up frequency distribution, calculating the deviation between other historical follow-up frequencies and the basic follow-up frequency and performing distribution probability allocation to obtain a historical frequency probability distribution; According to the number of dimensionality reduction, randomly select dimensionality reduction for the historical follow-up frequency distribution to obtain a first dimensionality reduction follow-up frequency set, calculate the deviation between each first dimensionality reduction follow-up frequency in the first dimensionality reduction follow-up frequency set and the basic follow-up frequency and perform distribution probability allocation to obtain a first dimensionality reduction frequency probability distribution; Construct a frequency dimension reduction function to optimize the dimension reduction of the historical follow-up frequency distribution, as follows: ; Among them, FDR is the frequency dimension reduction fitness, and is the weight, SIM is the feature retention of the dimensionality reduction frequency probability distribution and the historical frequency probability distribution, M is the number of dimensionality reduction, is the number of occurrences of the i-th dimension reduction follow-up frequency in the dimension reduction follow-up frequency set, The number of basic occurrences for basic follow-up frequency; Calculate similarity based on the first dimension reduction frequency probability distribution and the historical frequency probability distribution to obtain a first feature retention degree, and calculate and obtain a first frequency dimension reduction fitness of the first dimension reduction follow-up frequency set according to the frequency dimension reduction function; Continue to iteratively reduce the dimension of the historical follow-up frequency distribution until convergence, and obtain a reduced-dimensional follow-up frequency set.

4. The system according to claim 1, characterized in that According to the current follow-up product, the dimension-reduced follow-up frequency set and the dimension-reduced follow-up content set are associated and dimension-reduced to obtain a reliable follow-up frequency set and a reliable follow-up content set, including: Collect product feature information of the follow-up products currently following up the target customers; Extracting multiple historical product feature information of multiple follow-up products corresponding to the dimensionality reduction follow-up frequency and the dimensionality reduction follow-up content in the dimensionality reduction follow-up frequency set and the dimensionality reduction follow-up content set; Analyze and obtain the plurality of historical product feature information and the plurality of product-related parameters of the product feature information; According to multiple product association parameters, the dimensionality reduction follow-up frequency and dimensionality reduction follow-up content with the largest association quantity threshold of product association parameters are selected from the dimensionality reduction follow-up frequency set and the dimensionality reduction follow-up content set respectively, and a reliable follow-up frequency set and a reliable follow-up content set are obtained by output.

5. The system according to claim 1, characterized in that Conduct customer propensity analysis to obtain purchase propensity rates, including: Collect sample positivity set, sample modified liking set, and sample purchase tendency rate set; Using the sample positivity set, the sample modified liking set and the sample purchase tendency rate set to train a purchase tendency analyzer; Based on the purchase tendency analyzer, the purchase tendency analysis is performed on the positivity and the modified preference to obtain a purchase tendency rate.

6. A customer tendency analysis method combined with follow-up status, characterized in that: The method comprises: Collect follow-up data in the process of following up target customers, where the follow-up data includes follow-up frequency data and follow-up content data; Collecting a historical follow-up data set within a historical period, performing dimensionality reduction of the follow-up frequency on the historical follow-up data set to obtain a dimensionality-reduced follow-up frequency set, and performing dimensionality reduction of the follow-up content on the historical follow-up data set to obtain a dimensionality-reduced follow-up content set; According to the current follow-up product, the dimension-reduced follow-up frequency set and the dimension-reduced follow-up content set are associated and dimension-reduced to obtain a reliable follow-up frequency set and a reliable follow-up content set; According to the reliable follow-up frequency set and the reliable follow-up content set, a positivity fitting and a likeability fitting are performed to obtain a positivity analyzer and a likeability analyzer, and positivity analysis and likeability analysis are performed on the follow-up frequency data and the follow-up content data respectively to obtain positivity and likeability; Indexing similar matching follow-up content in the reliable follow-up content set according to the follow-up content data, and performing a preference migration change analysis based on the matching follow-up content to obtain a preference correction coefficient, correcting the preference to obtain a corrected preference, and performing a customer tendency analysis in combination with the positivity to obtain a purchase tendency rate as a customer tendency analysis result; The process of collecting a historical follow-up data set within a historical period, reducing the dimension of the follow-up frequency of the historical follow-up data set to obtain a reduced-dimensional follow-up frequency set, and reducing the dimension of the follow-up content of the historical follow-up data set includes: Extracting and classifying the follow-up frequency and the follow-up content of the historical follow-up data set to obtain a historical follow-up frequency distribution and a historical follow-up content distribution, wherein the historical follow-up frequency distribution includes a plurality of different historical follow-up frequencies and the number of occurrences of the plurality of historical follow-up frequencies; Performing dimension reduction of the follow-up frequency on the historical follow-up frequency distribution to obtain a dimension-reduced follow-up frequency set, wherein the dimension reduction is performed to improve the feature retention of the historical follow-up frequency distribution and the number of occurrences of the dimension-reduced follow-up frequency; Performing dimensionality reduction on the historical follow-up content distribution to obtain a dimensionality-reduced follow-up content set; According to the reliable follow-up frequency set and the reliable follow-up content set, a positivity fitting and a likeability fitting are performed to obtain a positivity analyzer and a likeability analyzer, and positivity analysis and likeability analysis are performed on the follow-up frequency data and the follow-up content data, respectively, including: Extract the number of occurrences of each reliable follow-up frequency in the reliable follow-up frequency set in the historical follow-up frequency distribution, and the number of purchases when each reliable follow-up frequency appears, and calculate to obtain a sample positivity set; Extracting the number of occurrences of each reliable follow-up content in the reliable follow-up content set in the historical follow-up content distribution and the number of purchases when each reliable follow-up content appears, and calculating to obtain a sample preference set; Using the reliable follow-up frequency set and the sample positivity set as fitting data, performing fitting training, and obtaining a positivity analyzer; Using the reliable follow-up content set and the sample likeability set to perform fitting training to obtain a likeability analyzer; Using the positivity analyzer and the likeability analyzer to perform positivity analysis and likeability analysis on the follow-up frequency data and the follow-up content data, respectively, to obtain positivity and likeability; Among them, indexing similar matching follow-up content in the reliable follow-up content set according to the follow-up content data, and performing a preference degree migration change analysis based on the matching follow-up content to obtain a preference correction coefficient, including: Analyzing the similarity between each reliable follow-up content in the reliable follow-up content set and the follow-up content data, and outputting the reliable follow-up content with a similarity greater than a threshold as a plurality of matching follow-up contents; According to the multiple matching follow-up contents, the purchase rate of each matching follow-up content appearing in different purchase volume stages is indexed in the historical follow-up data set to obtain multiple likeability correction tables; Extracting the current purchase volume stage of the target customer, indexing in the plurality of likeability correction tables, obtaining a plurality of matching likeability correction coefficients, and calculating the likeability correction coefficients; The likeability correction coefficient is used to perform correction calculation on the likeability to obtain a corrected likeability.

7. An electronic device, characterized in that: include: Memory for storing computer software programs; The processor is used to read and execute the computer software program, thereby implementing the customer tendency analysis method combined with follow-up status as described in claim 6.

Citation Information

Patent Citations

  • Forecasting method for customer purchase intention and terminal device

    CN107688967A

  • Customer purchase willingness prediction method, related device and computer storage medium

    CN113902476A

  • Consumption tendency analysis method and device, equipment and storage medium

    CN114219544A

  • Propensity modeling process for customer targeting

    US20230032739A1