A CRM-based franchise store member management system

Abstract

This invention discloses a CRM-based franchise store membership management system, comprising: a cluster set acquisition module for acquiring a set of behavioral clusters; a set sequence construction module for constructing a cluster set sequence; an evolution sequence construction module for constructing an evolution sequence of K behavioral clusters; a current behavior acquisition module for acquiring the current behavior vector; a similar cluster anchoring module for anchoring the most similar behavioral cluster; and an evolution path positioning module for performing differentiated behavior management on target members. After acquiring the current behavior vector of a target member within a target time window, this invention can locate the complete evolution path to which it belongs, thereby resolving the cluster center vectors of each time window in the evolution path into interpretable behavioral features, and thus identifying the typical patterns of the target member. Finally, based on the future behavioral trends revealed by this path, differentiated behavior management is performed on the target member, improving the level of store membership management.

Description

Technical Field

[0001] This invention relates to the field of store membership management, specifically a CRM-based franchise store membership management system. Background Technology

[0002] In CRM-based franchise store membership management practices, operators typically rely on historical consumption data to periodically segment members to identify high-value users, price-sensitive groups, or potential churn customers, and then formulate differentiated strategies such as offering discounts and adjusting contact frequency accordingly. These methods generally use fixed time windows (such as weekly or monthly) as units to independently construct member behavior vectors and perform clustering to generate a set of behavior clusters for the current period.

[0003] However, in the actual operation of franchised stores, a complete business cycle (such as a promotional season, holiday period, or calendar month) often includes multiple phases of operational rhythm, and member behavior evolves dynamically accordingly. For example, some users may participate frequently at the beginning of a promotion, but then gradually become inactive; others may gradually expand from single-category consumption to repurchase multiple categories. Existing solutions, because they only focus on static clustering results within a single window, lack the ability to correlate behavioral clusters from different windows in chronological order, making it difficult to reconstruct the behavioral evolution path of the aforementioned typical groups. As a result, while the system can identify "who is currently in charge," it cannot predict "what will happen in the future," hindering the upgrade of marketing strategies from passive response to proactive guidance. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a CRM-based franchise store membership management system, which solves the technical problems mentioned in the background by introducing a cluster evolution sequence.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A CRM-based franchise store membership management system includes the following modules: The cluster set acquisition module is used to acquire the behavioral cluster sets of each store's sales cycle within N historical time windows; each behavioral cluster set contains an equal number of K behavioral clusters. The set sequence construction module is used to construct a cluster set sequence based on the behavior cluster sets of each of the N historical time windows; The evolutionary sequence construction module is used to construct K behavioral cluster evolutionary sequences based on the cluster set sequence; The current behavior acquisition module is used to obtain the target member's current behavior vector within the target time window; The similarity cluster anchoring module is used to anchor the most similar behavior cluster within the store's sales cycle based on the current behavior vector; The evolution path localization module is used to locate the evolution sequence of the behavior cluster to which the most similar behavior cluster belongs, as the behavior evolution path of the target member, and to perform differentiated behavior management on the target member based on the behavior evolution path.

[0006] In some specific embodiments, the cluster set acquisition module is specifically used for: S1-1. Select N historical time windows at equal intervals within the store's sales cycle; S1-2, Obtain M member behavior vectors for each of the N historical time windows; S1-3. Perform K-value clustering on the M member behavior vectors of each window to generate K behavior clusters for each historical time window; S1-4. Combine the K behavior clusters within each historical time window into a behavior cluster set, until N behavior cluster sets are obtained.

[0007] In some specific embodiments, generating K clusters of behaviors for each historical time window includes: S1-3-1. Randomly select K member behavior vectors from M member behavior vectors as the first round cluster centers; S1-3-2, Calculate the vector distance between each member's behavior vector and the K first-round cluster centers; S1-3-4. Assign the member behavior vector to the first-round cluster center with the minimum vector distance; S1-3-5. Traverse the MK member behavior vectors until each member behavior vector is assigned to the first round cluster center with the minimum vector distance, forming K initial behavior clusters; S1-3-6. Calculate the cluster center vector of each of the K initial behavior clusters and define it as the cluster center of the next round. S1-3-7. Calculate the vector distance between the cluster centers of the next round and the cluster centers of the first round; S1-3-8. If the vector distance is greater than the set threshold, the M member behaviors are re-allocated based on the cluster center of the next round until the vector distance between the cluster center of the next round and the cluster center of the previous round is less than the preset threshold, and K behavior clusters are generated.

[0008] In some specific embodiments, the set sequence construction module is specifically used for: S2-1. Based on the order of N historical time windows, generate monotonically increasing window numbers for the historical time windows; S2-2. Assign window numbers to the corresponding set of behavior clusters to obtain N ordered sets of behavior clusters; S2-3. Arrange the N ordered behavior cluster sets in ascending order according to their assigned window numbers to generate a cluster set sequence.

[0009] In some specific embodiments, the evolutionary sequence construction module is specifically used for: S3-1. In the cluster set sequence, match N behavior clusters of the target behavior cluster starting from the smallest window index; S3-2. Obtain the cluster center vectors and window numbers of N behavioral clusters; S3-3. Arrange the N behavior clusters in ascending order based on the window number to generate a behavior cluster evolution sequence; S3-4. Traverse the cluster set sequence and repeatedly execute the generation of behavior cluster evolution sequences until K behavior cluster evolution sequences are obtained.

[0010] In some specific embodiments, the matching of N evolutionary behavior clusters of the target behavior cluster starting from the smallest window index includes: S3-1-1. Anchor the current cluster set corresponding to the smallest window index along the ascending direction of the cluster set sequence; S3-1-2. Select the target row cluster in the current cluster set, and anchor the next row adjacent to the current cluster set as the cluster set; S3-1-3. In the next set of clusters, match the evolutionary behavior cluster that is most similar to the target behavior cluster. S3-1-4. Take the most similar evolutionary behavior cluster as the target behavior cluster in the next behavior cluster set, and iteratively update the most similar evolutionary behavior cluster of the target behavior cluster; S3-1-5. Traverse the N ordered behavior cluster sets of the cluster set sequence until the N evolutionary behavior clusters of the target behavior cluster starting from the smallest window number are matched.

[0011] In some specific embodiments, the matching of the evolutionary behavior cluster most similar to the target behavior cluster includes: S3-1-3-1, Anchor the next row to K candidate behavior clusters within the cluster set; S3-1-3-2. Obtain the front cluster center vector of the target behavior cluster, and the back cluster center vectors of the K candidate behavior clusters; S3-1-3-3, Calculate the K inter-cluster similarities between the front cluster center vector and the back cluster center vector; S3-1-3-4. Select the maximum inter-cluster similarity among the K inter-cluster similarities; S3-1-3-5. Based on the maximum inter-cluster similarity, anchor the evolutionary behavior cluster that is most similar to the target behavior cluster among the K candidate behavior clusters.

[0012] In some specific embodiments, the similar cluster anchoring module is specifically used for: S5-1. Obtain the window number of the target time window within the store's sales cycle; S5-2. Based on the window number, anchor the corresponding window number in the cluster set sequence to the cluster set set; S5-3. Obtain the K behavior clusters and their cluster center vectors from the set of behavior clusters; S5-4. Calculate the K current similarities between the current action vector and the K cluster center vectors; S5-5. Select the largest current similarity among the K current similarities; S5-6. Based on the maximum current similarity, anchor the most similar behavior cluster among the K behavior clusters.

[0013] This invention provides a CRM-based franchise store membership management system, which has the following beneficial effects: This invention generates a set of behavior clusters containing K behavior clusters, enabling the structured preservation of member segmentation results at each business stage. Based on this, the behavior cluster sets are arranged chronologically according to window numbers to form a cluster set sequence, and K continuous behavior cluster evolution sequences are constructed. This transforms the originally discrete periodic segmentation into a traceable typical member group behavior evolution path. Furthermore, after obtaining the current behavior vector of a target member within a target time window, the complete evolution path to which it belongs can be located. This allows the cluster center vectors of each time window in the evolution path to be analyzed into interpretable behavioral features, thereby identifying the typical patterns of the target member. Finally, based on the future behavioral trends revealed by this path, differentiated behavior management of target members is implemented, improving the level of store member management. Attached Figure Description

[0014] Figure 1 This is a structural block diagram of a franchise store membership management system based on CRM according to the present invention; Figure 2 This is a flowchart illustrating a CRM-based franchise store membership management system according to the present invention. Figure 3 This is a schematic diagram of the matching process for the evolutionary behavior clusters described in this invention; Figure 4 This is a schematic diagram of the anchoring process for the most similar evolutionary behavior cluster described in this invention. Figure 5 This is a schematic diagram of the anchoring process for the most similar behavior cluster described in this invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] Example 1: Please refer to Figures 1 to 2 This invention provides a CRM-based franchise store membership management system, comprising the following modules: The cluster set acquisition module is used to acquire the behavioral cluster sets of each store's sales cycle within N historical time windows; each behavioral cluster set contains an equal number of K behavioral clusters. Specifically, the store sales cycle refers to a complete business cycle, such as a week, a month, or a promotional season.

[0017] The behavior cluster refers to a set of members with similar consumption patterns obtained by clustering the member behavior vectors of M members within a certain historical time window.

[0018] The set sequence construction module is used to construct a cluster set sequence based on the behavior cluster sets of each of the N historical time windows; The set of behavior clusters represents the set of all K behavior clusters within a historical time window; Therefore, the cluster set sequence represents an ordered structure formed by arranging the behavioral cluster sets corresponding to N historical time windows in chronological order.

[0019] The evolutionary sequence construction module is used to construct K behavioral cluster evolutionary sequences based on the cluster set sequence. Specifically, the behavioral cluster evolutionary sequence represents the evolution path of the consumption behavior patterns of a group of members with similar behavioral characteristics across multiple time windows within the sales cycle. The current behavior acquisition module is used to obtain the target member's current behavior vector within the target time window; The similarity cluster anchoring module is used to anchor the most similar behavior cluster within the store's sales cycle based on the current behavior vector; The evolution path localization module is used to locate the evolution sequence of the behavior cluster to which the most similar behavior cluster belongs, as the behavior evolution path of the target member, and to perform differentiated behavior management on the target member based on the behavior evolution path.

[0020] Specifically, the behavioral evolution path represents a behavioral cluster evolution sequence matched with the behavioral characteristics of a target member in the current time window. This sequence describes the typical consumption behavior evolution process of a member group with similar historical behavioral patterns during the sales cycle. It includes the behavioral pattern characteristics and evolution trends of the member group in each time window. Therefore, based on the future behavioral trends revealed by this behavioral evolution path, corresponding behavioral management can be carried out for potential churn, repurchase, or upgrade behaviors of store members in future sales cycles.

[0021] For example, the differentiated behavior management means that personalized coupons are pushed in a targeted manner, the frequency of marketing outreach is adjusted, or exclusive membership level benefits are set according to the behavioral evolution path of the target members.

[0022] In the specific execution process, through the preset dimension-semantic mapping table, the value of each dimension of the cluster center vector of the behavior cluster corresponding to each time window in the behavior cluster evolution sequence is mapped back to its corresponding original behavior feature. For example, when the cluster center vector of a certain behavior cluster is [3.2, 45.6, 0.8, 0.1, 0.1, 0.7], the system knows the following through the dimension-semantic mapping table: Dimension 0 = 3.2 purchases in the last 7 days (high frequency) Dimension 1 = Average order value 45.6 yuan The second dimension = beverages account for 80%. 6th Dimension = 70% Discount Usage Rate (Highly Sensitive) Therefore, this behavior cluster can be identified as "high-value price-sensitive users", and a discount coupon is recommended instead of a full reduction coupon.

[0023] The dimension-semantic mapping table is generated synchronously when constructing member behavior vectors, recording the correspondence between each vector dimension and behavioral characteristics such as purchase frequency, average order value, category preference, and discount sensitivity; based on the mapping results, the behavioral evolution trend of the member group is quantitatively analyzed, and corresponding differentiated management strategies are generated.

[0024] Example 2: See Figures 3 to 5 The technical solution of this embodiment 2 differs from that of embodiment 1 in that it discloses the specific application steps of each module in embodiment 1.

[0025] Specifically, in this embodiment, the cluster set acquisition module is used for: S1-1. Select N historical time windows at equal intervals within the store's sales cycle; S1-2, Obtain M member behavior vectors for each of the N historical time windows; The member behavior vector is a d-dimensional numerical vector constructed within the corresponding historical time window based on the member's consumption frequency, consumption amount, category preference, discount usage behavior, and interaction events.

[0026] Specifically, in this embodiment, the behavioral characteristics of the member behavior vector are as follows: The d-dimensional numerical vector includes the following behavioral feature dimensions: (1) Consumption activity: the number of transactions within the corresponding historical time window; (2) Consumption value: The cumulative consumption amount within the corresponding historical time window; (3) Average transaction value: calculated by dividing the total amount spent by the number of transactions; (4) Category preference percentage: According to the unified product classification system of the headquarters, the consumption amount percentage of each primary category is calculated; (5) Discount Sensitivity: The proportion of orders that used coupons or participated in promotional activities within the corresponding historical time window to the total number of orders; (6) Interaction depth: including at least one of the following: number of times coupons were claimed but not redeemed, number of times members participated in member activities, and number of reviews / complaints.

[0027] All of the above behavioral features were normalized before constructing vectors to eliminate differences in dimensions.

[0028] S1-3. Perform K-value clustering on the M member behavior vectors of each window to generate K behavior clusters for each historical time window; S1-4. Combine the K behavior clusters within each historical time window into a behavior cluster set, until N behavior cluster sets are obtained.

[0029] In this embodiment, by dividing the store sales cycle into equally spaced historical time windows, and constructing a normalized member behavior vector based on multi-dimensional behavioral features within each window, the generated behavior cluster can accurately reflect the differences in consumption behavior of different member groups at specific business stages.

[0030] Furthermore, steps S1-3 include: S1-3-1. Randomly select K member behavior vectors from M member behavior vectors as the first round cluster centers; S1-3-2, Calculate the vector distance between each member's behavior vector and the K first-round cluster centers; The formula for calculating the vector distance is: ; in, Represents a member behavior vector. Represents the cluster center vector. Indicates the number of dimensions of a vector. and These represent member behavior vectors respectively. with cluster center vector The feature components in the i-th dimension.

[0031] S1-3-4. Assign the member behavior vector to the first-round cluster center with the minimum vector distance; S1-3-5. Traverse the MK member behavior vectors until each member behavior vector is assigned to the first round cluster center with the minimum vector distance, forming K initial behavior clusters; S1-3-6. Calculate the cluster center vector of each of the K initial behavior clusters and define it as the cluster center of the next round. S1-3-7. Calculate the vector distance between the cluster centers of the next round and the cluster centers of the first round; S1-3-8. If the vector distance is greater than the set threshold, the M member behaviors are re-allocated based on the cluster center of the next round until the vector distance between the cluster center of the next round and the cluster center of the previous round is less than the preset threshold, and K behavior clusters are generated.

[0032] In this embodiment, by clustering member behavior vectors within each historical time window, the active members of the current period are divided into K groups with clear business meanings, enabling stores to implement operational management based on typical behavior patterns (high frequency, low amount, high value sensitivity, concentrated category preference, etc.) within the business cycle.

[0033] Specifically, in this embodiment, the set sequence construction module is used for: S2-1. Based on the order of N historical time windows, generate monotonically increasing window numbers for the historical time windows; Specifically, the window sequence number represents the chronological order of each historical time window within the store's sales cycle, and is used to uniquely determine the relative position of each behavior cluster set within the sales cycle.

[0034] S2-2. Assign window numbers to the corresponding set of behavior clusters to obtain N ordered sets of behavior clusters; S2-3. Arrange the N ordered behavior cluster sets in ascending order according to their assigned window numbers to generate a cluster set sequence.

[0035] In this embodiment, by assigning window numbers to each historical time window and arranging the behavior cluster sets in order, the discrete member grouping results within the period are restored to the time axis of the store sales cycle, so that the behavior changes of the same type of member group correspond to the business stage (such as promotion launch, weekend peak, end-of-month clearance).

[0036] Specifically, in this embodiment, the evolutionary sequence construction module is used for: S3-1. In the cluster set sequence, match N behavior clusters of the target behavior cluster starting from the smallest window index; S3-2. Obtain the cluster center vectors and window numbers of N behavioral clusters; S3-3. Arrange the N behavior clusters in ascending order based on the window number to generate a behavior cluster evolution sequence; S3-4. Traverse the cluster set sequence and repeatedly execute the generation of behavior cluster evolution sequences until K behavior cluster evolution sequences are obtained.

[0037] In this embodiment, by tracking the continuity status of each initial behavior cluster in each window during the sales cycle, K typical member behavior evolution paths are generated, enabling stores to identify interventionable behavior evolution patterns (gradual churn of promotion-dependent users or high-frequency beverage customers upgrading to package deals, etc.).

[0038] Furthermore, step S3-1 includes: S3-1-1. Anchor the current cluster set corresponding to the smallest window index along the ascending direction of the cluster set sequence; S3-1-2. Select the target row cluster in the current cluster set, and anchor the next row adjacent to the current cluster set as the cluster set; S3-1-3. In the next set of clusters, match the evolutionary behavior cluster that is most similar to the target behavior cluster. Specifically, the most similar evolutionary behavior cluster refers to the behavior cluster in the next set of behavior clusters whose cluster center vector has the highest similarity to the cluster center vector of the current target behavior cluster.

[0039] S3-1-4. Take the most similar evolutionary behavior cluster as the target behavior cluster in the next behavior cluster set, and iteratively update the most similar evolutionary behavior cluster of the target behavior cluster; S3-1-5. Traverse the N ordered behavior cluster sets of the cluster set sequence until the N evolutionary behavior clusters of the target behavior cluster starting from the smallest window number are matched.

[0040] In this embodiment, by matching the most similar behavior clusters based on cluster center vector similarity between adjacent time windows, it is ensured that each evolutionary path always corresponds to the behavior continuation process of the same type of member group.

[0041] Furthermore, step S3-1-3 includes: S3-1-3-1, Anchor the next row to K candidate behavior clusters within the cluster set; S3-1-3-2. Obtain the front cluster center vector of the target behavior cluster, and the back cluster center vectors of the K candidate behavior clusters; S3-1-3-3, Calculate the K inter-cluster similarities between the front cluster center vector and the back cluster center vector; The preferred inter-cluster similarity is cosine similarity, which is calculated as follows: ; in, Indicates the vector of the front cluster center. Indicates the vector at the center of the cluster. Represents the vector dot product. , These represent the magnitudes of the vectors at the front and back cluster centers, respectively. S3-1-3-4. Select the maximum inter-cluster similarity among the K inter-cluster similarities; S3-1-3-5. Based on the maximum inter-cluster similarity, anchor the evolutionary behavior cluster that is most similar to the target behavior cluster among the K candidate behavior clusters.

[0042] It should be noted that when classifying the behavior of store members, the focus should be on the structural similarity of member behavior patterns, rather than the absolute value of consumption.

[0043] For example, a customer with an average order value of 30 yuan who comes 3 times a week and a customer with an average order value of 60 yuan who comes 6 times a week, although the amounts are different, both belong to the "high-frequency stable type" and should be regarded as a continuation on the same evolutionary path.

[0044] Therefore, this embodiment preferably uses cosine similarity to measure the behavioral consistency between cross-window behavioral clusters, so that the system can identify member groups with different consumption intensity but similar behavioral structures (such as high-frequency low-spending and high-frequency high-spending users) belonging to the same evolutionary path, avoiding misclassification of groups due to differences in absolute consumption levels.

[0045] Specifically, in this embodiment, the similar cluster anchoring module is used for: S5-1. Obtain the window number of the target time window within the store's sales cycle; S5-2. Based on the window number, anchor the corresponding window number in the cluster set sequence to the cluster set set; S5-3. Obtain the K behavior clusters and their cluster center vectors from the set of behavior clusters; S5-4. Calculate the K current similarities between the current action vector and the K cluster center vectors; S5-5. Select the largest current similarity among the K current similarities; S5-6. Based on the maximum current similarity, anchor the most similar behavior cluster among the K behavior clusters.

[0046] Furthermore, when classifying the real-time behavior of target members, it should be ensured that the comparison between their current behavior and that of historical groups occurs in the same business phase (such as both being promotional weeks or the end of the month) to avoid misclassification due to misaligned periods.

[0047] Therefore, this embodiment first obtains the window number of the target time window, anchors the set of historical behavior clusters with the same number accordingly, and then matches the current behavior vector to the most similar behavior cluster based on cosine similarity, so that the system can accurately identify the typical member group to which the member currently belongs.

[0048] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means.

[0049] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0050] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0051] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A CRM-based franchise store membership management system, characterized in that, include: The cluster set acquisition module is used to acquire the behavioral cluster sets of each store's sales cycle within N historical time windows; each behavioral cluster set contains an equal number of K behavioral clusters. The set sequence construction module is used to construct a cluster set sequence based on the behavior cluster sets of each of the N historical time windows; The evolutionary sequence construction module is used to construct K behavioral cluster evolutionary sequences based on the cluster set sequence; The current behavior acquisition module is used to obtain the target member's current behavior vector within the target time window; The similarity cluster anchoring module is used to anchor the most similar behavior cluster within the store's sales cycle based on the current behavior vector; The evolution path localization module is used to locate the evolution sequence of the behavior cluster to which the most similar behavior cluster belongs, as the behavior evolution path of the target member, and to perform differentiated behavior management on the target member based on the behavior evolution path.

2. The franchise store membership management system based on CRM according to claim 1, characterized in that, The cluster set acquisition module is specifically used for: S1-1. Select N historical time windows at equal intervals within the store's sales cycle; S1-2, Obtain M member behavior vectors for each of the N historical time windows; S1-3. Perform K-value clustering on the M member behavior vectors of each window to generate K behavior clusters for each historical time window; S1-4. Combine the K behavior clusters within each historical time window into a behavior cluster set, until N behavior cluster sets are obtained.

3. The franchise store membership management system based on CRM according to claim 2, characterized in that, The K behavior clusters generated for each historical time window include: S1-3-1. Randomly select K member behavior vectors from M member behavior vectors as the first round cluster centers; S1-3-2, Calculate the vector distance between each member's behavior vector and the K first-round cluster centers; S1-3-4. Assign the member behavior vector to the first-round cluster center with the minimum vector distance; S1-3-5. Traverse the MK member behavior vectors until each member behavior vector is assigned to the first round cluster center with the minimum vector distance, forming K initial behavior clusters; S1-3-6. Calculate the cluster center vector of each of the K initial behavior clusters and define it as the cluster center of the next round. S1-3-7. Calculate the vector distance between the cluster centers of the next round and the cluster centers of the first round; S1-3-8. If the vector distance is greater than the set threshold, the M member behaviors are re-allocated based on the cluster center of the next round until the vector distance between the cluster center of the next round and the cluster center of the previous round is less than the preset threshold, and K behavior clusters are generated.

4. The franchise store membership management system based on CRM according to claim 1, characterized in that, The set sequence construction module is specifically used for: S2-1. Based on the order of N historical time windows, generate monotonically increasing window numbers for the historical time windows; S2-2. Assign window numbers to the corresponding set of behavior clusters to obtain N ordered sets of behavior clusters; S2-3. Arrange the N ordered behavior cluster sets in ascending order according to their assigned window numbers to generate a cluster set sequence.

5. A franchise store membership management system based on CRM according to claim 1, characterized in that, The evolutionary sequence construction module is specifically used for: S3-1. In the cluster set sequence, match N behavior clusters of the target behavior cluster starting from the smallest window index; S3-2. Obtain the cluster center vectors and window numbers of N behavioral clusters; S3-3. Arrange the N behavior clusters in ascending order based on the window number to generate a behavior cluster evolution sequence; S3-4. Traverse the cluster set sequence and repeatedly execute the generation of behavior cluster evolution sequences until K behavior cluster evolution sequences are obtained.

6. A franchise store membership management system based on CRM according to claim 5, characterized in that, The matching of the target behavior cluster, starting from the smallest window index, comprises N evolutionary behavior clusters, including: S3-1-1. Anchor the current cluster set corresponding to the smallest window index along the ascending direction of the cluster set sequence; S3-1-2. Select the target row cluster in the current cluster set, and anchor the next row adjacent to the current cluster set as the cluster set; S3-1-3. In the next set of clusters, match the evolutionary behavior cluster that is most similar to the target behavior cluster. S3-1-4. Take the most similar evolutionary behavior cluster as the target behavior cluster in the next behavior cluster set, and iteratively update the most similar evolutionary behavior cluster of the target behavior cluster; S3-1-5. Traverse the N ordered behavior cluster sets of the cluster set sequence until the N evolutionary behavior clusters of the target behavior cluster starting from the smallest window number are matched.

7. A franchise store membership management system based on CRM according to claim 6, characterized in that, The evolutionary behavior clusters that are most similar to the target behavior cluster include: S3-1-3-1, Anchor the next row to K candidate behavior clusters within the cluster set; S3-1-3-2. Obtain the front cluster center vector of the target behavior cluster, and the back cluster center vectors of the K candidate behavior clusters; S3-1-3-3, Calculate the K inter-cluster similarities between the front cluster center vector and the back cluster center vector; S3-1-3-4. Select the maximum inter-cluster similarity among the K inter-cluster similarities; S3-1-3-5. Based on the maximum inter-cluster similarity, anchor the evolutionary behavior cluster that is most similar to the target behavior cluster among the K candidate behavior clusters.

8. A franchise store membership management system based on CRM according to claim 7, characterized in that, The similar cluster anchoring module is specifically used for: S5-1. Obtain the window number of the target time window within the store's sales cycle; S5-2. Based on the window number, anchor the corresponding window number in the cluster set sequence to the cluster set set; S5-3. Obtain the K behavior clusters and their cluster center vectors from the set of behavior clusters; S5-4. Calculate the K current similarities between the current action vector and the K cluster center vectors; S5-5. Select the largest current similarity among the K current similarities; S5-6. Based on the maximum current similarity, anchor the most similar behavior cluster among the K behavior clusters.

Images