User behavior analysis method, device, medium and electronic device

Abstract

The present invention relates to the field of data processing and provides a user behavior analysis method, device, medium and equipment. The user behavior analysis method includes: obtaining multiple groups of influencing factors and user behaviors corresponding to each group of influencing factors according to a specific dimension, wherein each group of influencing factors includes at least one non-digital factor, and the user behaviors include at least a first behavior; obtaining the information entropy of the user's first behavior according to the multiple groups of influencing factors and the user behaviors corresponding to each group of influencing factors; obtaining the conditional entropy of the user's first behavior under the condition that the target influencing factor is known according to the multiple groups of influencing factors and the user behaviors corresponding to each group of influencing factors; obtaining the mutual information between the target influencing factor and the user's first behavior according to the information entropy and the conditional entropy; and obtaining the degree of correlation between the target influencing factor and the first behavior according to the mutual information. The user behavior analysis method can objectively and accurately analyze user behavior.

Description

Technical Field

[0001] The present invention belongs to the field of data processing, and in particular relates to a user behavior analysis method, device, medium and electronic equipment. Background Art

[0002] In data analysis, the analysis of user behavior often involves many non-digital factors in addition to some digitizable factors. Taking user purchasing behavior as an example, the factors that prompt users to make purchases are complex and diverse. They include both digitizable factors, such as the price and sales volume of the product, and non-digitizable factors, such as the delivery method of the product and user reviews. Therefore, when reviewing and analyzing user purchasing behavior, it is necessary to consider both digitizable and non-digital factors. However, most existing technologies analyze digitizable factors based on various dimensions, but cannot objectively analyze non-digital factors. Summary of the Invention

[0003] In view of the above-mentioned shortcomings of the prior art, the object of the present invention is to provide a user behavior analysis method, device, medium and electronic device to solve the problem that the prior art cannot objectively analyze non-digital influencing factors.

[0004] To achieve the above-mentioned purpose and other related purposes, the first aspect of the present invention provides a user behavior analysis method, including: obtaining multiple groups of influencing factors and user behaviors corresponding to each group of influencing factors according to a specific dimension, wherein each group of influencing factors includes at least one non-digital factor, and the user behaviors corresponding to at least one group of influencing factors include a first behavior; obtaining the information entropy of the user performing the first behavior according to the multiple groups of influencing factors and the user behaviors corresponding to each group of influencing factors; obtaining the conditional entropy of the user performing the first behavior under the condition that the target influencing factor is known according to the multiple groups of influencing factors and the user behaviors corresponding to each group of influencing factors; obtaining the mutual information between the target influencing factor and the user performing the first behavior according to the information entropy and the conditional entropy; and obtaining the degree of correlation between the target influencing factor and the first behavior according to the mutual information.

[0005] In an embodiment of the first aspect, obtaining the degree of correlation between the target influencing factor and the first behavior based on the mutual information includes: determining whether the time interval for the user to perform the first behavior meets a preset condition; when the time interval for the user to perform the first behavior meets the preset condition, obtaining the degree of correlation between the target influencing factor and the first behavior based on the mutual information; when the time interval for the user to perform the first behavior does not meet the preset condition, correcting the mutual information based on the historical behavior of the user and its similar users, and obtaining the degree of correlation between the target influencing factor and the first behavior based on the corrected mutual information.

[0006] In an embodiment of the first aspect, the method for correcting the mutual information based on the historical behavior of the user and similar users includes: obtaining the time when the user last performed the first behavior under the condition that the target influencing factor is included as a reference time; obtaining the average period during which the user's similar users perform the first behavior as a reference period; and correcting the mutual information based on the reference time and the reference period.

[0007] In an embodiment of the first aspect, a method for correcting the mutual information based on the reference time and the reference period includes: processing the reference time, the reference period, and the mutual information using a first correction formula to correct the mutual information, where the first correction formula is:

[0008]

[0009] Where I(U;X) represents the mutual information, t(X) last represents the reference time, T refill(P) represents the reference period, N is such that 0≤t(X) last -N×T refill(P) <T refill(P) The largest natural number that holds true.

[0010] In an embodiment of the first aspect, a method for correcting the mutual information based on the reference time and the reference period includes: processing the reference time, the reference period, and the mutual information using a second correction formula to correct the mutual information, where the second correction formula is:

[0011]

[0012] Where I(U;X) represents the mutual information, t(X) last represents the reference time, T refill(P') represents the reference period.

[0013] In an embodiment of the first aspect, the number of the target influencing factors is at least two. After obtaining the degree of correlation between the target influencing factors and the first behavior, the user behavior analysis method further includes: sorting each target influencing factor according to the degree of correlation between each target influencing factor and the first behavior.

[0014] In an embodiment of the first aspect, obtaining the information entropy of the user performing the first behavior based on the multiple groups of influencing factors and the user behaviors corresponding to each group of influencing factors includes: identifying missing data and abnormal data in the influencing factors; preprocessing the missing data and abnormal data in the influencing factors; and processing the influencing factors and the user behaviors corresponding to each group of influencing factors after preprocessing to obtain the information entropy of the user performing the first behavior.

[0015] The second aspect of the present invention provides a user behavior analysis device, including a data acquisition module for acquiring multiple groups of influencing factors and user behaviors corresponding to each group of influencing factors according to a specific dimension, wherein each group of influencing factors includes at least one non-digital factor, and the user behaviors corresponding to at least one group of influencing factors include a first behavior; an information entropy acquisition module, connected to the data acquisition module, for acquiring the information entropy of the user performing the first behavior based on the multiple groups of influencing factors and the user behaviors corresponding to each group of influencing factors; a conditional entropy acquisition module, connected to the data acquisition module, for acquiring the conditional entropy of the user performing the first behavior under the condition that the target influencing factor is known based on the multiple groups of influencing factors and the user behaviors corresponding to each group of influencing factors; a mutual information acquisition module, connected to the information entropy acquisition module and the conditional entropy acquisition module, for acquiring the mutual information between the target influencing factor and the user performing the first behavior based on the information entropy and the conditional entropy; and a behavior analysis module, connected to the mutual information acquisition module, for acquiring the degree of correlation between the target influencing factor and the first behavior based on the mutual information.

[0016] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the user behavior analysis method described in any one of the first aspects of the present invention.

[0017] A fourth aspect of the present invention provides an electronic device, comprising: a memory storing a computer program; a processor communicatively connected to the memory, and executing the user behavior analysis described in any one of the first aspects of the present invention when calling the computer program.

[0018] As described above, the user behavior analysis method described in one or more embodiments of the present invention has the following beneficial effects:

[0019] The user behavior analysis method can measure the amount of information generated by non-digital factors that affect user purchasing behavior in the form of entropy, and express the degree of correlation between non-digital factors and user behavior in numerical form through information entropy, conditional entropy, and mutual information. This process is completely unaffected by user subjective factors, and thus the user behavior analysis method can objectively and accurately analyze user behavior.

[0020] In addition, the user behavior analysis method can also modify the mutual information based on the historical behavior of the user and its similar users. By superimposing a time decay term on the mutual information, the accuracy and objectivity of the analysis results can be further improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 Shown is a flowchart of a specific embodiment of the user behavior analysis method of the present invention.

[0022] Figure 2 Shown is a flowchart of modifying mutual information in a specific embodiment of the user behavior analysis method of the present invention.

[0023] Figure 3 Shown is a flow chart of another specific embodiment of the user behavior analysis method of the present invention.

[0024] Figure 4 Shown is a structural diagram of the user behavior analysis device according to a specific embodiment of the present invention.

[0025] Figure 5 Shown is a schematic structural diagram of the electronic device according to a specific embodiment of the present invention.

[0026] Component number description

[0027] 1 User behavior analysis device

[0028] 11 Data Acquisition Module

[0029] 12 Information entropy acquisition module

[0030] 13 Conditional Entropy Acquisition Module

[0031] 14 Mutual Information Acquisition Module

[0032] 15 Behavior Analysis Module

[0033] 500 Electronic Equipment

[0034] 510 Memory

[0035] 520 processor

[0036] 530 Display

[0037] Steps S11 to S15

[0038] Steps S21 to S23

[0039] Steps S31 to S37 DETAILED DESCRIPTION

[0040] The following describes the embodiments of the present invention through specific examples. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, unless there is a conflict, the following embodiments and the features in the embodiments can be combined with each other. In addition, some exemplary embodiments of the present invention are described as devices represented by block diagrams and processes or methods represented by flowcharts. Although the flowcharts describe the operating process of the present invention as sequential processing, many of the operations therein can be performed in parallel, concurrently or simultaneously. In addition, the order of the operations can be rearranged. The process of the present invention can be terminated when its operations are completed, but it can also include additional steps not shown in the flowcharts. The process of the present invention can correspond to methods, functions, procedures, subroutines, subprograms, etc.

[0041] It should be noted that the diagrams provided in the following embodiments are merely schematic illustrations of the basic concept of the present invention. The diagrams only show components relevant to the present invention and are not drawn according to the number, shape, and size of components in actual implementation. In actual implementation, the type, quantity, and proportion of each component may be arbitrarily varied, and the component layout may also be more complex. Furthermore, in this document, relational terms such as "first," "second," and the like are used solely to distinguish one entity or operation from another, and do not necessarily require or imply any actual relationship or order between these entities or operations.

[0042] When reviewing and analyzing the user's purchasing behavior, both digitizable factors and non-digital factors must be considered. However, most of the existing technologies analyze digitizable factors based on various dimensions, but cannot objectively analyze non-digital factors. At least in response to the above problems, the present invention provides a user behavior analysis method, which can measure the amount of information generated by non-digital factors that affect user purchasing behavior in the form of entropy, and express the degree of correlation between non-digital factors and user purchasing behavior in numerical form through information entropy, conditional entropy, and mutual information. This process is completely unaffected by the user's subjective factors, and thus the user behavior analysis method can objectively and accurately analyze user behavior.

[0043] Hereinafter, specific embodiments of the present invention will be described with reference to the accompanying drawings through exemplary embodiments.

[0044] In one embodiment of the present invention, the user behavior analysis method is used to quantitatively analyze the degree of correlation between non-digital factors and user behavior. It should be noted that in some other embodiments, the user behavior analysis method can also be used to quantitatively analyze the degree of correlation between digital factors and user behavior. In some other embodiments, the user behavior analysis method can also be used to quantitatively analyze the degree of correlation between digital factors and non-digital factors and user behavior at the same time.

[0045] Figure 1 The flowchart of the user behavior analysis method in this embodiment is shown. Figure 1 As shown, the user behavior analysis method in this embodiment includes the following steps S11 to S15.

[0046] Step S11: Acquire multiple groups of influencing factors and the user behaviors corresponding to each group of influencing factors based on specific dimensions. Each group of influencing factors includes at least one non-digital factor, and the user behaviors corresponding to at least one group of influencing factors include the first behavior. Non-digital factors refer to influencing factors that cannot be described numerically, such as product reviews, delivery methods, etc. The specific dimensions refer to dimensions related to the user behaviors and can be set based on actual needs or experience, or obtained through statistical methods. The specific methods are not limited in this invention. Each group of influencing factors corresponds to a user behavior, and each influencing factor in each group corresponds to a dimension. It should be noted that the user behaviors corresponding to different groups of influencing factors can be the same or different. For example, if the specific dimensions include promotional activities, product sales, and product reviews, then one group of influencing factors may be: a promotional activity of a discount for purchases above a certain limit, product sales exceeding 1,000, and product reviews of excellent; the user behavior corresponding to this group of influencing factors is the user's purchase behavior. Another group of influencing factors may be: a promotional activity of a gift for purchases above a certain limit, product sales exceeding 20,000, and product reviews of good; the user behavior corresponding to this group of influencing factors is the user's non-purchase behavior.

[0047] Step S12: Obtain information entropy of the user performing the first behavior based on the multiple groups of influencing factors and the user behaviors corresponding to each group of influencing factors. The information entropy represents the uncertainty of the user performing the first behavior. A larger information entropy indicates a higher uncertainty of the user performing the first behavior, while a smaller information entropy indicates a lower uncertainty of the user performing the first behavior.

[0048] Step S13: Obtaining the conditional entropy of the user's first behavior under the condition that the target influencing factor is known, based on the multiple groups of influencing factors and the user behaviors corresponding to each group of influencing factors. Specifically, for any target influencing factor X, the conditional entropy is used to represent the uncertainty of the user's first behavior under the condition that the target influencing factor X is known. The smaller the conditional entropy, the greater the influence of the target influencing factor X on the user's first behavior. Conversely, the larger the conditional entropy, the smaller the influence of the target influencing factor X on the user's first behavior.

[0049] Step S14: Obtain the mutual information between the target influencing factor and the first behavior of the user based on the information entropy and the conditional entropy. For example, for the first behavior and any target influencing factor X, the mutual information I(U;X) between the target influencing factor X and the first behavior of the user can be expressed as: I(U;X)=H(U)-H(U|X), where H(U) represents the information entropy of the first behavior of the user, and H(U|X) represents the conditional entropy of the first behavior of the user under the condition that the target influencing factor X is known.

[0050] Step S15: Obtain the degree of association between the target influencing factor and the first behavior based on the mutual information. Specifically, if the mutual information between the target influencing factor and the first behavior is closer to 1, the higher the degree of association between the target influencing factor and the first behavior, and the greater the influence of the target influencing factor on the user's performance of the first behavior. Conversely, if the mutual information between the target influencing factor and the first behavior is closer to 0, the lower the degree of association between the target influencing factor and the first behavior, and the smaller the influence of the target influencing factor on the user's performance of the first behavior.

[0051] It should be understood that the above-mentioned labels S11 to S15 are only used to identify different steps, rather than to limit the order of these steps. In specific applications, the execution order of the above-mentioned steps can be determined according to actual needs. For example, step S12 can be executed first and then step S13, or step S13 can be executed first and then step S12, or step S12 and step S13 can be executed at the same time. The present invention does not impose any restrictions on this.

[0052] Optionally, the user behavior may also include a second behavior. The user behavior analysis method can also objectively analyze the degree of correlation between the target influencing factor and the first behavior and the second behavior at the same time. In some embodiments, the first behavior may be a purchase behavior of a target product, and the second behavior may be a non-purchase behavior of a target product. The target products include but are not limited to medicines, but the present invention is not limited to this. When analyzing the user's purchase behavior or non-purchase behavior, the dimensions that affect the user behavior include but are not limited to: advertising creativity, promotional activities, product sales, product praise, after-sales guarantee, delivery method and / or delivery time.

[0053] In addition, it should be understood that the types of user behaviors are not limited in this embodiment. In addition to the first behavior and the second behavior, the user behavior may also include a third behavior, a fourth behavior, etc. Regardless of the types of user behaviors, the user behavior analysis method described in this embodiment can objectively and accurately analyze the degree of correlation between the target influencing factors and each user behavior.

[0054] Optionally, in this embodiment, a large amount of user data and behavior may be pre-collected as a data set, and dimensions related to the first behavior may be obtained from this data set using a data mining algorithm. Specifically, data mining refers to the process of algorithmically searching for information hidden within a large amount of data. In this embodiment, data mining may be implemented using various methods, such as statistics, online analytical processing, intelligence retrieval, machine learning, expert systems, and pattern recognition, thereby obtaining dimensions related to the first behavior as the specific dimensions.

[0055] Optionally, obtaining the information entropy of the user performing the first behavior based on the multiple groups of influencing factors and the user behaviors corresponding to each group of influencing factors includes: identifying missing data and abnormal data in the influencing factors; preprocessing the missing data and abnormal data in the influencing factors; processing the influencing factors and the user behaviors corresponding to each group of influencing factors after preprocessing to obtain the information entropy of the user performing the first behavior. Specifically, since a large amount of user data and behaviors can be collected in step S11, problems such as missing data and data anomalies are inevitable during the collection process, which will affect the analysis results in step S15. To address this issue, this embodiment can eliminate the impact of problems such as missing data and data anomalies on the analysis results by preprocessing the data collected in step S11, thereby improving the accuracy of behavioral analysis.

[0056] Optionally, the preprocessing of missing data in this embodiment may include missing data discarding, missing data filling, etc. Missing data discarding refers to directly discarding all influencing factors of the group containing the missing data. Missing data filling refers to obtaining an estimated value of the missing data by analyzing existing data and replacing the missing data with the estimated value. The missing data filling methods in this embodiment may include mean interpolation, similar mean interpolation, modeling prediction, high-dimensional mapping, multiple interpolation, maximum likelihood estimation, compressed sensing, and matrix completion.

[0057] Optionally, in this embodiment, the preprocessing of abnormal data may include abnormal data discarding, abnormal data replacement, etc. Abnormal data discarding and abnormal data replacement are similar to the implementation methods of missing data discarding and missing data replacement, and are not described in detail here.

[0058] It should be noted that the above preprocessing methods for missing data and abnormal data are only two optional schemes for implementing preprocessing in this embodiment, but the present invention is not limited to this. In specific applications, data standardization, data normalization, data encoding and other methods can also be used to preprocess the data obtained in step S11.

[0059] Optionally, in this embodiment, step S12 may use the following formula to obtain the information entropy H(U): Among them, p(u i ) represents user u i The probability of the first behavior occurring can be obtained by counting the number of times the first behavior occurs by users in the data set, where n represents the number of users in the data set.

[0060] Optionally, in this embodiment, step S13 may use the following formula to obtain the conditional entropy H(U|X): Where X={x1,x2,...,x m}, P(u i ,x j ) represents user u i The joint probability of the first behavior and the target influencing factor X occurring can be obtained by counting the number of times the user occurs the first behavior when the target influencing factor X in the data set is known, where m is the number of dimensions included in the target influencing factor X.

[0061] In one embodiment of the present invention, the step of obtaining the degree of association between the target influencing factor and the first behavior based on the mutual information includes: determining whether the time interval between the user's performance of the first behavior satisfies a preset condition; when the time interval between the user's performance of the first behavior satisfies the preset condition, obtaining the degree of association between the target influencing factor and the first behavior based on the mutual information; when the time interval between the user's performance of the first behavior does not satisfy the preset condition, correcting the mutual information based on the historical behavior of the user and similar users, and obtaining the degree of association between the target influencing factor and the first behavior based on the corrected mutual information. Specifically, for many products, such as medicines and food, users often repurchase the products some time after purchasing them, i.e., they engage in repeat purchases. Therefore, whether a user will perform the first behavior is also related to the historical behavior of the user and similar users. Therefore, the user analysis method described in this embodiment may further include the step of correcting the mutual information based on the historical behavior of the user and similar users to improve the accuracy of the obtained mutual information. The similar users may be all members of the same level as the user on the platform. For example, if the user is a level 1 user, the similar users are all level 1 users on the platform.

[0062] Optionally, Figure 2 The flowchart of this embodiment is to modify the mutual information according to the historical behavior of the user and similar users. Figure 2 As shown, in this embodiment, the process of correcting the mutual information according to the historical behavior of the user and its similar users includes the following steps S21 to S23.

[0063] Step S21: Obtain the time when the user last performed the first behavior under the condition that the target influencing factor is included as a reference time.

[0064] Step S22: obtaining an average period during which similar users of the user perform the first behavior as a reference period. The average period may be obtained by statistical methods or the like.

[0065] Step S23: Modify the mutual information according to the reference time and the reference period.

[0066] As can be seen from the above description, in this embodiment, the mutual information can be modified based on the attributes of the target product and its repurchase cycle, where the repurchase cycle refers to the time interval between a user's first purchase and their next purchase. This approach allows for a more objective and accurate analysis of the correlation between non-digital factors and user purchasing behavior.

[0067] Optionally, in some embodiments, the preset condition includes a first preset condition. The first preset condition is, for example, that the time intervals between the user's multiple occurrences of the first behavior are the same or approximately the same, and the approximately same is, for example, that the difference between the two time intervals is less than ten days, twenty days, or thirty days. For example, if the first behavior is the purchase of chronic disease medications, the chronic disease medications are, for example, therapeutic drugs for diseases such as diabetes, hypertension, and heart disease. Since users need to purchase such drugs periodically, the purchase behavior of such drugs is approximately a periodic behavior. At this time, the first behavior satisfies the first preset condition, and its corresponding time decay term shows an obvious periodic law. Based on this, in this embodiment, the mutual information can be corrected by superimposing a periodic time decay term on the mutual information.

[0068] Furthermore, for any user, when the first behavior satisfies the first preset condition, the mutual information may be corrected using the following formula in this embodiment:

[0069]

[0070] Wherein, I(U;X) represents the mutual information, represents the modified mutual information, t(X) last represents the time when the user last performed the first behavior under the condition of including the target influencing factor, that is, the reference time, T refill(P) represents the average period during which the same type of users perform the first behavior, i.e., the reference period, t(X) last With T refill(P) The unit is, for example, hours, days, weeks, etc., and N is such that 0≤t(X) last -N×T refill(P) <T refill(P) The maximum natural number that holds true. In addition, the average period T contained in the denominator of the above formula refill(P) By correcting the mutual information in the above manner, the corrected mutual information can more objectively and accurately reflect the degree of correlation between the target influencing factor and the first behavior of the user.

[0071] Optionally, in some other embodiments, the preset condition includes a second preset condition. The second preset condition is, for example, that the time interval between the user's multiple occurrences of the first behavior is uncertain. For example, if the first behavior is the purchase of non-chronic disease medications, such as cold medicines, since users usually only need to purchase such products once or a small number of times, the time at which the user will make the next purchase after a purchase is uncertain. Therefore, the time decay term corresponding to such behavior does not follow the periodic law but follows the exponential decay law. Such behavior can be regarded as meeting the second preset condition. Based on this, in this embodiment, an exponential decay term can be superimposed on the mutual information to achieve correction of the mutual information.

[0072] Furthermore, for any user, when the first behavior satisfies the second preset condition, the mutual information may be corrected using the following formula in this embodiment:

[0073]

[0074] Wherein, I(U;X) represents the mutual information, represents the modified mutual information, t(X) last represents the time from the last time the user performed the first behavior under the condition of including the target influencing factor, that is, the reference time, T refill(P') represents the average period during which the same type of users perform the first behavior, i.e., the reference period, t(X) last With T refill(P The unit of ') is, for example, hour, day, week, etc. By adopting the above method to correct the mutual information, the corrected mutual information can more objectively and accurately reflect the correlation degree between the target influencing factor and the first behavior.

[0075] According to the above description, when the first behavior is the user's purchase behavior of the target product, the user behavior analysis method described in this embodiment can measure the amount of information generated by the non-digital factors that affect the user's purchase behavior in the form of entropy, and express the degree of correlation between the non-digital factors and the user's purchase behavior in numerical form through information entropy, conditional entropy, and mutual information. This process is completely unaffected by the user's subjective factors, so the user behavior analysis method can objectively and accurately analyze user behavior. In addition, the user behavior analysis method described in this embodiment can also correct the mutual information according to the attributes of the target product and its repurchase cycle, and further improve the accuracy and objectivity of the analysis results by superimposing a time decay term on the mutual information.

[0076] In one embodiment of the present invention, there are multiple target influencing factors. After obtaining the degree of correlation between the target influencing factors and the first behavior, the user behavior analysis method in this embodiment may further include: sorting each target influencing factor according to the degree of correlation between each target influencing factor and the first behavior.

[0077] Optionally, after ranking each of the influencing factors, the user behavior analysis method in this embodiment may further include: using a display to display the ranking results of the influencing factors so that the user can intuitively and quickly understand the degree of correlation between each of the target influencing factors and the first behavior.

[0078] Optionally, after sorting each of the influencing factors, the user behavior analysis method in this embodiment may further include: grouping users according to the degree of correlation between the target influencing factors and the first behavior, and sorting each of the target influencing factors in each user group. It should be noted that when grouping users, a user may belong to only one group or multiple groups. For example, if the first behavior is the purchase behavior of a target product, and the target influencing factors include promotional activities, product sales, and product praise, then users can be divided into users whose purchasing behavior is affected by promotional activities, users whose purchasing behavior is affected by factors affecting product sales, and users whose purchasing behavior is affected by product praise based on the degree of correlation between the target influencing factors and the first behavior.

[0079] In one embodiment of the present invention, the first behavior is the user's purchase behavior of the target product. In this embodiment, the user behavior analysis method may further include: generating a marketing plan for a certain user or a certain type of user based on the degree of correlation between each of the target influencing factors and the user's purchase behavior. For example, for a certain user, if the analysis result obtained in step S15 is: the mutual information between the two target influencing factors of the promotion activity of full discount and after-sales guarantee and the user's purchase behavior is close to 1, and the mutual information between the target influencing factor of 24-hour delivery and the user's purchase behavior is close to 0, then the user can be marketed by adopting the two aspects of full discount and improving the quality of after-sales guarantee.

[0080] Figure 3 Shown is a flow chart of a user behavior analysis method in a preferred embodiment of the present invention, in which the first behavior is the user's purchase behavior of the target product.

[0081] It should be noted that this preferred example is only used to assist in explaining the user behavior analysis method, and is not used to limit the scope of protection of the present invention. The content contained therein is not necessary for the implementation of the present invention, that is, in some embodiments, not all steps of the preferred embodiment may be included.

[0082] like Figure 3 As shown, the user behavior analysis method in this preferred embodiment includes the following steps S31 to S37.

[0083] Step S31: Obtaining a dimension Y that influences user purchases. Dimension Y may include, for example, advertising creativity, promotional activities, product sales, product reviews, after-sales service, delivery method, and delivery time, but the present invention is not limited thereto. In specific applications, dimension Y may be obtained based on received user instructions or through statistical and correlation analysis, and the present invention does not limit the specific implementation method.

[0084] In step S32, the data on users' purchase behaviors and non-purchase behaviors are grouped according to the dimension Y, product attributes, and users, thereby obtaining records of user purchase behaviors and the dimension Y. The records can be identified in the form of a record table, as shown in Table 1 below, but the present invention is not limited to this.

[0085] Table 1 User purchase behavior record

[0086]

[0087] Step S33: Calculate the information entropy of the user's purchase behavior. In this embodiment, the information entropy can be calculated using the following formula: Among them, p(u i ) represents user u i The probability of a purchase, where n is the total number of users.

[0088] Step S34, calculate the uncertainty of the user's purchase under the influence of the known dimension X, that is, the conditional entropy. In this embodiment, the conditional entropy can be calculated using the following formula: Among them, P(u i ,x j ) represents user u i The joint probability of purchasing behavior and influencing factor X, where m is the number of dimensions contained in influencing factor X. The lower the value of the conditional entropy, the greater the influence of dimension X on the user's purchasing behavior.

[0089] Step S35, calculate the mutual information between the user's purchase behavior and the known dimension X. In this embodiment, the mutual information can be calculated, for example, using the following formula: I(U;X)=H(U)-H(U|X). Among them, the mutual information I(U;X) is an indicator used to measure the correlation between the known dimension X and the user's purchase behavior. When the known dimension X is completely correlated with the user's purchase behavior, the mutual information I(U;X) is 1, and when the known dimension X is completely unrelated to the user's purchase behavior, the mutual information I(U;X) is 0. Based on this, it can be seen that the closer the mutual information I(U;X) is to 1, the greater the influence of the known dimension X on the user's purchase behavior.

[0090] Step S36, the mutual information I(U; X) is corrected in the time dimension. Specifically, based on the mutual information I(U; X), the mutual information value of each dimension of each user and the user's purchasing behavior can be obtained, and the mutual information value can calculate the correlation between the dimension and the user's purchasing behavior from the perspective of statistical analysis. However, this correlation is not only related to statistical information such as frequency, but also to time. Therefore, in step S36, the mutual information I(U; X) can be corrected in the time dimension. For example, in step S36, an attenuation term of superimposed time can be added to the mutual information I(U; X).

[0091] Specifically, in the pharmaceutical field, the attenuation term is related to the attributes of the target product itself. If the target product is a chronic disease medication, the attenuation term changes periodically over time. Within the average repurchase period, the impact of dimension X on user purchasing behavior exhibits a U-shaped relationship. If the target product is a non-chronic disease medication, the attenuation term decreases as time increases.

[0092] Optionally, when the target product is a drug for a chronic disease, the mutual information corrected according to the attenuation term can be expressed as follows: Where t(X) last Indicates the time since the user last purchased the target product, T refill(P) represents the average repurchase cycle of the target product by similar users, and N is such that 0≤t(X) last -N×T refill(P) <T refill(P) The largest natural number that holds true.

[0093] Optionally, when the target product is a non-chronic disease medication, the mutual information corrected according to the attenuation term can be expressed by the following formula:

[0094] Step S37: Obtain the correlation between each dimension and each user's purchase behavior based on the corrected mutual information. Based on this, merchants can adopt corresponding marketing plans for individual users during the marketing process to achieve personalized marketing behavior.

[0095] Based on the above description of the user behavior analysis method, the present invention also provides a user behavior analysis device. Figure 4 In one embodiment of the present invention, the user behavior analysis device 1 includes a data acquisition module 11, an information entropy acquisition module 12, a conditional entropy acquisition module 13, a mutual information acquisition module 14, and a behavior analysis module 15. The user behavior analysis device 1 is used to quantitatively analyze the degree of correlation between non-digital factors and user behavior. It should be noted that Figure 4 The modules in the user behavior analysis device 1 are Figure 1 Steps S11 to S15 in the user behavior analysis method shown correspond to each other one by one, and are not described in detail here to save space.

[0096] In addition, the division of the above modules in this embodiment is only a division of logical functions. In actual implementation, they can be fully or partially integrated into one physical entity, or they can be physically separated. For example, the data acquisition module 11 can be deployed on the user terminal, while the information entropy acquisition module 12, the conditional entropy acquisition module 13, the mutual information acquisition module 14 and the behavior analysis module 15 can be deployed on the server side, and the server side obtains the data collected by the data acquisition module 11 by communicating with the user terminal. For another example, the data acquisition module 11, the information entropy acquisition module 12, the conditional entropy acquisition module 13, the mutual information acquisition module 14 and the behavior analysis module 15 can also all be deployed on the server side. In addition, all modules in the user behavior analysis device 1 can be implemented in the form of software called by processing elements, or in the form of hardware, or some modules can be implemented in the form of software called by processing elements, and some modules can be implemented in the form of hardware.

[0097] Based on the above description of the user behavior analysis method and the user behavior analysis device, another aspect of the present invention further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, which is executed by a processor to implement Figure 1 Specifically, the computer-readable storage medium may be one or more devices for storing data, including read-only memory (ROM), random access memory (RAM), magnetic RAM, kernel memory, magnetic disk storage medium, optical storage medium, flash memory device and / or other machine-readable media for storing information.

[0098] Based on the above description of the user behavior analysis device and the user behavior analysis method, the present invention further provides an electronic device. Figure 5 The structure diagram of the electronic device 500 in one embodiment of the present invention is shown. Figure 5 As shown, the electronic device 500 in this embodiment includes a memory 510 and a processor 520. The memory 510 may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories, on which a computer program is stored. The processor 520 is in communication with the memory 510 and executes the computer program when the computer program is called. Figure 1 The user behavior analysis method shown in the figure. The processor 520 can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0099] Optionally, the electronic device 500 may further include a display 530. The display 530 is communicatively connected to the memory 510 and the processor 520, and is configured to display a GUI interaction interface related to the user behavior analysis method.

[0100] The protection scope of the user behavior analysis method described in the present invention is not limited to the execution order of the steps listed in this embodiment. All solutions implemented by adding, reducing, or replacing steps in the prior art based on the principles of the present invention are included in the protection scope of the present invention.

[0101] The present invention also provides a user behavior analysis device, which can implement the user behavior analysis method described in the present invention. However, the implementation device of the user behavior analysis method described in the present invention includes but is not limited to the structure of the user behavior analysis device listed in this embodiment. All structural deformations and replacements of the existing technology made according to the principles of the present invention are included in the protection scope of the present invention.

[0102] In summary, the user behavior analysis method described in one or more embodiments of the present invention can measure the amount of information generated by non-digital factors that affect user purchasing behavior in the form of entropy, and express the degree of correlation between non-digital factors and user purchasing behavior in numerical form through information entropy, conditional entropy, and mutual information. This process is completely unaffected by user subjective factors, and thus the user behavior analysis method can objectively and accurately analyze user behavior.

[0103] Furthermore, the user behavior analysis method can modify the mutual information based on the attributes of the target product and its repurchase cycle. By superimposing a time decay term on the mutual information, the accuracy and objectivity of the analysis results can be further improved. Therefore, the present invention effectively overcomes the shortcomings of the existing technology and has high industrial application value.

[0104] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the present invention. Anyone skilled in the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by one of ordinary skill in the art without departing from the spirit and technical principles disclosed herein are intended to be covered by the claims of the present invention.

Claims

1. A user behavior analysis method, characterized in that: include: Acquire multiple groups of influencing factors and user behaviors corresponding to each group of influencing factors according to a preset dimension, wherein each group of influencing factors includes at least one non-digital factor, and the user behaviors corresponding to at least one group of influencing factors include the first behavior; Obtaining information entropy of the user's first behavior based on the multiple groups of influencing factors and the user behaviors corresponding to each group of influencing factors; Obtaining, based on the multiple groups of influencing factors and the user behaviors corresponding to each group of influencing factors, a conditional entropy of the user performing the first behavior under the condition that the target influencing factor is known; Obtaining mutual information between the target influencing factor and the first behavior of the user according to the information entropy and the conditional entropy; Determining whether a time interval between the user's first behavior and a preset condition is satisfied; When the time interval between the user performing the first behavior satisfies the preset condition, obtaining the degree of correlation between the target influencing factor and the first behavior according to the mutual information; When the time interval between the user's first behavior and the user's last behavior does not meet the preset condition, obtaining the time when the user last performed the first behavior under the condition that the target influencing factor is included as a reference time; obtaining the average period of the first behavior performed by users of the same type as the user as a reference period; The reference time, the reference period, and the mutual information are processed using a first correction formula to correct the mutual information. The first correction formula is: Where I(U;X) represents the mutual information, t(X) last represents the reference time, T refill(P) represents the reference period, N is such that 0≤t(X) last -N×T refill(P) <T refill(P) The largest natural number that holds true; The degree of correlation between the target influencing factor and the first behavior is obtained according to the corrected mutual information.

2. The user behavior analysis method according to claim 1, characterized in that: A method for correcting the mutual information according to the reference time and the reference period includes: using a second correction formula to process the reference time, the reference period, and the mutual information to correct the mutual information, wherein the second correction formula is: Where I(U;X) represents the mutual information, t(X) last represents the reference time, T refill(P') represents the reference period.

3. The user behavior analysis method according to claim 1, characterized in that: The number of the target influencing factors is at least two. After obtaining the correlation between the target influencing factors and the first behavior, the user behavior analysis method further includes: The target influencing factors are ranked according to the degree of association between the target influencing factors and the first behavior.

4. The user behavior analysis method according to claim 1, characterized in that: The obtaining of information entropy of the user's first behavior according to the multiple groups of influencing factors and the user behaviors corresponding to the groups of influencing factors includes: Identify missing and outlier data in the influencing factors; Preprocessing missing data and abnormal data in the influencing factors; The pre-processed influencing factors and the user behaviors corresponding to each group of influencing factors are processed to obtain information entropy of the user performing the first behavior.

5. A user behavior analysis device for use in the user behavior analysis method according to any one of claims 1 to 4, characterized in that: The user behavior analysis device comprises: a data acquisition module, configured to acquire, based on preset dimensions, multiple groups of influencing factors and user behaviors corresponding to each group of influencing factors, wherein each group of influencing factors includes at least one non-digital factor, and the user behaviors corresponding to at least one group of influencing factors include the first behavior; an information entropy acquisition module, connected to the data acquisition module, for acquiring information entropy of the user's first behavior based on the multiple groups of influencing factors and the user behaviors corresponding to each group of influencing factors; a conditional entropy acquisition module, connected to the data acquisition module, for acquiring, based on the multiple groups of influencing factors and the user behaviors corresponding to each group of influencing factors, the conditional entropy of the user performing the first behavior under the condition that the target influencing factors are known; a mutual information acquisition module, connected to the information entropy acquisition module and the conditional entropy acquisition module, configured to acquire the mutual information between the target influencing factor and the first behavior of the user based on the information entropy and the conditional entropy; A behavior analysis module is connected to the mutual information acquisition module and is used to obtain the degree of correlation between the target influencing factor and the first behavior based on the mutual information.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the user behavior analysis method according to any one of claims 1 to 4 is implemented.

7. An electronic device, characterized in that: The electronic device comprises: a memory storing a computer program; A processor is communicatively connected to the memory and executes the user behavior analysis method according to any one of claims 1 to 4 when calling the computer program.

Images