Electric power service sensitivity grading method and system based on customer grouping

Abstract

The invention discloses an electric power service sensitivity grading method and system based on customer grouping, and relates to the technical field of service sensitivity grading, and the method comprises the steps: screening out a non-forward feedback work order from historical feedback customer service work orders, and extracting sensitive information from the non-forward feedback work order; for each power customer, based on the sensitive information, calculating to obtain comprehensive sensitivity and personalized sensitivity of the corresponding sensitive point; according to the comprehensive sensitivity and the personalized sensitivity of the sensitive point corresponding to each power customer, calculating to obtain the target sensitivity of the sensitive point corresponding to each power customer; based on a preset sensitivity level, obtaining a target sensitivity level of each power customer according to each target sensitivity by using an improved k-means clustering algorithm; according to the method, sensitive points and sensitivities of different customer groups are accurately found, different marketing strategies are made for different customer groups, lean services are provided for customers, and the customer service satisfaction degree is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of service sensitivity level classification, and more specifically, to a method and system for classifying power service sensitivity levels based on customer grouping. Background Art

[0002] With the continuous construction and development of smart grids, a large amount of customer data assets have been accumulated, including customer service work order data and business data. In the era of customer first, the reasonable configuration of power services will directly determine the company's costs and customer service satisfaction. Therefore, power supply companies urgently need to conduct scientific analysis on customer service quality sensitivity, identify the sensitive points and sensitivity of different customer groups, formulate different marketing strategies for different customer groups, and provide customers with lean services to improve customer service satisfaction; so that power grid companies can further expand the depth and breadth of services and provide data support for the formulation of future power demand-side response policies. Summary of the invention

[0003] The object of the present invention is to provide a method and system for classifying power service sensitivity levels based on customer grouping, so as to solve the problems existing in the above-mentioned background technology.

[0004] The above technical objectives of the present invention are achieved through the following technical solutions:

[0005] In a first aspect, the present application provides a method for classifying the sensitivity levels of electric power services based on customer grouping, comprising the following specific steps:

[0006] Filter out non-positive feedback work orders from historical feedback customer service work orders, and extract sensitive information from the non-positive feedback work orders, including sensitive points of each power customer;

[0007] For each electricity customer, based on sensitive information, the comprehensive sensitivity and personalized sensitivity of the corresponding sensitive point are calculated respectively;

[0008] According to the comprehensive sensitivity and personalized sensitivity of the sensitive points corresponding to each power customer, the target sensitivity of the sensitive points corresponding to each power customer is calculated;

[0009] Based on the preset sensitivity level, the target sensitivity level of each power customer is obtained according to each target sensitivity and using the improved k-means clustering algorithm.

[0010] The beneficial effects of the present invention are as follows: in this scheme, firstly, by analyzing customer service work orders, customer work orders are divided into demand type and feedback type, and in the feedback type of work orders, the service quality sensitivity is only for the feedback type of work orders that do not contain praise business types; secondly, the sensitive points of the work orders are extracted, that is, customer complaints, reports, etc.; then, the sensitivity of each sensitive point of the customer is predicted by combining the comprehensive sensitivity of each sensitive point of the customer and the personalized sensitivity of each sensitive point, that is, the target sensitivity; finally, based on the sensitivity of each sensitive point, the level is divided, and the improved k-means clustering algorithm with high computational efficiency, stability and good effect is applied; the customer service quality sensitivity prediction is realized, so as to find the sensitive points and sensitivities of different customer groups, formulate different marketing strategies for different customer groups, provide customers with lean services, so as to improve customer service satisfaction, enable power grid companies to further expand the depth and breadth of services, and provide data support for the formulation of future power demand side response policies.

[0011] Based on the above technical solution, the present invention can also be improved as follows.

[0012] Furthermore, the above historical feedback customer service work orders include praise type and non-positive feedback work orders, and non-positive feedback work orders include complaint type, opinion type, suggestion type, report type and urging type; the comprehensive sensitivity of a power customer corresponding to a sensitive point is specifically:

[0013] Among them, x i =m i / N i , i=1,2,…6;

[0014] Where P1 represents the comprehensive sensitivity of power customer A to sensitive point B, x i represents the proportion of sensitive point B corresponding to power customer A in work order type i, i corresponds to the number of work order types in historical feedback customer service work orders, m i Indicates the number of work orders with sensitive point B in work order type i, N i represents the number of work orders in work order type i, ω i Represents the weight of the i-th work order type.

[0015] Furthermore, the personalized sensitivity of one of the above-mentioned power customers corresponding to the sensitive point is specifically:

[0016]

[0017] Where P2 represents the personalized sensitivity of power customer A to sensitive point B, ω i represents the weight of the i-th work order type, P1 represents the comprehensive sensitivity of power customer A to sensitive point B, N iIndicates the number of work orders in work order type i, n i Indicates the number of work orders with sensitive point B and work order type i in historical feedback customer service work orders.

[0018] Furthermore, the target sensitivity of each of the above-mentioned power customers corresponding to the sensitive point is specifically:

[0019] P = P1 + P2;

[0020] Where P2 represents the personalized sensitivity of power customer A to sensitive point B, P1 represents the comprehensive sensitivity of power customer A to sensitive point B, and P represents the target sensitivity of power customer A to sensitive point B.

[0021] Furthermore, the target sensitivity level of each power customer is obtained according to each target sensitivity and using the improved k-means clustering algorithm, specifically:

[0022] A1, establish a corresponding number of clusters according to the number of sensitivity levels, and apply the initial cluster center algorithm to determine the initial cluster center of each cluster;

[0023] A2, calculate the Euclidean distance between the target sensitivity of each electricity customer and each initial cluster center, and assign each target sensitivity to the cluster with the closest Euclidean distance;

[0024] A3, calculate the average value of all data objects in each cluster and use the average value as the new cluster center of the cluster;

[0025] A4, loop A2 to A3 until the new cluster centers of each cluster obtained after two iterations are the same.

[0026] Furthermore, the above-mentioned initial cluster center algorithm is used to determine the initial cluster center of each cluster, specifically:

[0027] A11, determining the target sensitivity of each power customer with the smallest value as the first target value;

[0028] A12, calculating the Euclidean distance between the target sensitivities of other power customers and the first target value, and arranging the target sensitivities in ascending order according to the Euclidean distance;

[0029] A13, in the sequence of target sensitivities arranged in ascending order, assigning the target sensitivity numbered 1+(i-1)(n-2) / (k-1) to the kth cluster, where i=1, 2...k, and n represents the total number of target sensitivities;

[0030] A14, for the target sensitivities not classified into the clusters, calculating the Euclidean distance between each target sensitivity not classified into the clusters and each target sensitivity classified into the clusters, and classifying each target sensitivity not classified into the clusters into the cluster with the smallest Euclidean distance;

[0031] A15, calculate the average value of the sensitivity of each target in each cluster, and determine the average value as the initial cluster center.

[0032] The beneficial effect of adopting the above further scheme is that the initial cluster center algorithm avoids random steps, making the clustering results more reliable. At the same time, the calculated initial cluster center points are evenly distributed and can better represent the overall distribution of the original data, which can appropriately reduce the number of iterations.

[0033] In a second aspect, the present application provides a system for classifying the sensitivity level of electric service based on customer grouping, which is applied to a method for classifying the sensitivity level of electric service based on customer grouping in any one of the first aspects, including:

[0034] The first module is used to filter out non-positive feedback work orders from historical feedback customer service work orders, and extract sensitive information from the non-positive feedback work orders, and the sensitive information includes the sensitive points of each power customer;

[0035] The second module is used to calculate the comprehensive sensitivity and personalized sensitivity of the corresponding sensitive point for each power customer based on the sensitive information;

[0036] The third module is used to calculate the target sensitivity of the sensitive point corresponding to each power customer based on the comprehensive sensitivity and personalized sensitivity of the sensitive point corresponding to each power customer;

[0037] The fourth module is used to obtain the target sensitivity level of each power customer based on the preset sensitivity level and according to each target sensitivity and using the improved k-means clustering algorithm.

[0038] Furthermore, the fourth module includes:

[0039] The first submodule is used to establish a corresponding number of clusters according to the number of sensitivity levels, and to determine the initial cluster center of each cluster using an initial cluster center algorithm;

[0040] The second submodule is used to calculate the Euclidean distance between the target sensitivity of each power customer and each initial cluster center, and assign each target sensitivity to the cluster with the closest Euclidean distance;

[0041] The third submodule is used to calculate the average value of all data objects in each cluster and use the average value as the new cluster center of the cluster;

[0042] The fourth submodule is used for the second submodule and the third submodule until the new cluster centers of each cluster obtained after two iterations are the same.

[0043] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements any one of the methods in the first aspect when executing the computer program.

[0044] In a fourth aspect, the present application provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions enable a computer to execute any one of the methods in the first aspect.

[0045] Compared with the prior art, the present invention has at least the following beneficial effects:

[0046] In this application, firstly, by analyzing the customer service work orders, the customer work orders are divided into demand type and feedback type. Among the feedback type work orders, the service quality sensitivity is only for the feedback type work orders that do not contain praise business types; secondly, the sensitive points of the work orders are extracted, that is, customer complaints, reports, etc.; then, the sensitivity of each sensitive point of the customer is predicted by combining the comprehensive sensitivity of each sensitive point of the customer and the personalized sensitivity of each sensitive point, that is, the target sensitivity; finally, based on the sensitivity of each sensitive point, the level is divided, and the improved k-means clustering algorithm with high computational efficiency, stability and good effect is applied; the customer service quality sensitivity prediction is realized, so as to find the sensitive points and sensitivities of different customer groups, formulate different marketing strategies for different customer groups, provide customers with lean services, so as to improve customer service satisfaction, enable power grid companies to further expand the depth and breadth of services, and provide data support for the formulation of future power demand-side response policies.

[0047] In the present application, random steps are avoided through the initial cluster center algorithm, making the clustering results more reliable. At the same time, the calculated initial cluster center points are evenly distributed and can better represent the overall distribution of the original data, which can appropriately reduce the number of iterations. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0049] Figure 1 A method flow chart of a division method in an embodiment of the present invention;

[0050] Figure 2 A connection diagram of a partitioning system in an embodiment of the present invention;

[0051] Figure 3Schematic diagram of the connection of electronic equipment in an embodiment of the present invention. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0053] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0055] In the description of the embodiments of the present invention, "plurality" means at least 2.

[0056] Embodiment 1:

[0057] The present application embodiment provides a method for classifying the sensitivity level of electric power services based on customer grouping, such as Figure 1 As shown, the specific steps include:

[0058] S1, filter out non-positive feedback work orders from historical feedback customer service work orders, and extract sensitive information from the non-positive feedback work orders. The sensitive information includes the sensitive points of each power customer.

[0059] Among them, historical feedback customer service tickets include praise type and non-positive feedback tickets. Non-positive feedback tickets include complaint type, opinion type, suggestion type, report type and urging type.

[0060] Specifically, for example, the content of the customer complaint work order is: "

Frequent power outages

[0061] S2, for each electricity customer, based on the sensitive information, the comprehensive sensitivity and personalized sensitivity of the corresponding sensitive point are calculated respectively.

[0062] Among them, the prediction of service quality sensitivity can be based on the customer service work orders in the past three years. First, the sensitive points or sensitive contents of each work order are extracted, and then the comprehensive sensitivity of each sensitive point is predicted. The comprehensive sensitivity reflects which services all customers of the power grid are more sensitive or less sensitive to. Secondly, the personalized sensitivity of individual customers is predicted. The personalized sensitivity reflects which services individual customers are more sensitive or less sensitive to.

[0063] Optionally, the comprehensive sensitivity of the sensitive point corresponding to one of the above-mentioned power customers is specifically:

[0064] Among them, x i =m i / N i , i=1,2,…6;

[0065] Where P1 represents the comprehensive sensitivity of power customer A to sensitive point B, x i represents the proportion of sensitive point B corresponding to power customer A in work order type i, i corresponds to the number of work order types in historical feedback customer service work orders, m i Indicates the number of work orders with sensitive point B in work order type i, N i represents the number of work orders in work order type i, ω i Represents the weight of the i-th work order type.

[0066] Optionally, the personalized sensitivity of the sensitive point corresponding to one of the above-mentioned power customers is specifically:

[0067]

[0068] Where P2 represents the personalized sensitivity of power customer A to sensitive point B, ω i represents the weight of the i-th work order type, P1 represents the comprehensive sensitivity of power customer A to sensitive point B, N i Indicates the number of work orders in work order type i, n i Indicates the number of work orders with sensitive point B and work order type i in historical feedback customer service work orders.

[0069] Specifically, regarding the weights in the above formula, there is no unified standard for assigning weights. It can be an expert scoring method or some objective calculation methods, but the objective calculation methods are calculated based on some rules of the data itself and may not be in line with the actual situation. Here, weights can be assigned according to the actual situation; among them, the complaint type, opinion type, suggestion type, report type and urging type of work orders indicate that the power grid's services in this regard need to be improved, and the sensitivity of these types of services should be increased. The weights are all positive numbers, and the size of the weights is determined according to the intensity of the negative emotions they express; for example, the weights of complaint type, opinion type, suggestion type, report type and urging type can be set to 0.3, 0.2, 0.05, 0.3 and 0.15 respectively.

[0070] S3, according to the comprehensive sensitivity and personalized sensitivity of the sensitive points corresponding to each power customer, the target sensitivity of the sensitive point corresponding to each power customer is calculated.

[0071] Optionally, the target sensitivity of the sensitive point corresponding to each of the above-mentioned electricity customers is specifically:

[0072] P = P1 + P2;

[0073] Where P2 represents the personalized sensitivity of power customer A to sensitive point B, P1 represents the comprehensive sensitivity of power customer A to sensitive point B, and P represents the target sensitivity of power customer A to sensitive point B.

[0074] S4, based on the preset sensitivity level, the target sensitivity level of each power customer is obtained according to each target sensitivity and using the improved k-means clustering algorithm.

[0075] Among them, the sensitivity level division is based on the target sensitivity of the customer service quality calculated above, usually divided into 3-5 levels, depending on the specific situation; in this embodiment, the service quality sensitivity level can be divided into 5 levels; because it is not clear in advance which service quality sensitivity level category the user belongs to, clustering technology can be used to classify data without category information; clustering technology is usually also called unsupervised learning, in clustering, there is no classification information representing the data category. Clustering simply means grouping similar things into one group and dissimilar things into different groups.

[0076] Specifically, clustering algorithms mainly include hierarchical clustering algorithms and partition-based clustering algorithms. Hierarchical clustering algorithms need to repeatedly calculate the distance between two objects, and their time complexity is high ο(n^3), where n is the number of data objects. In addition, once two clusters are merged, they will not be undone. If one step is wrong, all steps will be wrong. In this embodiment, a partition-based clustering algorithm is used.

[0077] Optionally, the target sensitivity level of each power customer is obtained according to each target sensitivity and using an improved k-means clustering algorithm, specifically:

[0078] A1, establish a corresponding number of clusters according to the number of sensitivity levels, and apply the initial cluster center algorithm to determine the initial cluster center of each cluster.

[0079] Optionally, the above-mentioned initial cluster center algorithm is used to determine the initial cluster center of each cluster, specifically:

[0080] A11, determining the target value with the smallest sensitivity among the electricity customers as the first target value.

[0081] A12, calculating the Euclidean distance between the target sensitivities of other power customers and the first target value, and arranging the target sensitivities in ascending order according to the Euclidean distance.

[0082] A13, in the sequence of target sensitivities arranged in ascending order, the target sensitivity numbered 1+(i-1)(n-2) / (k-1) is assigned to the kth cluster, where i=1, 2...k, and n represents the total number of target sensitivities.

[0083] A14, for the target sensitivities not classified into the cluster, the Euclidean distance between each target sensitivity not classified into the cluster and each target sensitivity classified into the cluster is calculated, and each target sensitivity not classified into the cluster is classified into the cluster with the smallest Euclidean distance.

[0084] A15, calculate the average value of the sensitivity of each target in each cluster, and determine the average value as the initial cluster center.

[0085] Among them, the initial cluster center algorithm avoids random steps, making the clustering results more reliable. At the same time, the calculated initial cluster center points are evenly distributed and can better represent the overall distribution of the original data, which can appropriately reduce the number of iterations.

[0086] A2, calculate the Euclidean distance between the target sensitivity of each electricity customer and each initial cluster center, and assign each target sensitivity to the cluster with the closest Euclidean distance.

[0087] A3, calculate the average value of all data objects in each cluster and use the average value as the new cluster center of the cluster.

[0088] A4, loop A2 to A3 until the new cluster centers of each cluster obtained after two iterations are the same.

[0089] Embodiment 2:

[0090] The embodiment of the present application provides a system for classifying the sensitivity level of electric service based on customer grouping, which is applied to a method for classifying the sensitivity level of electric service based on customer grouping in any one of the embodiments 1, such as Figure 2 As shown, including:

[0091] The first module is used to filter out non-positive feedback work orders from historical feedback customer service work orders, and extract sensitive information from the non-positive feedback work orders, where the sensitive information includes sensitive points of each power customer.

[0092] The second module is used to calculate the comprehensive sensitivity and personalized sensitivity of the corresponding sensitive point for each electricity customer based on the sensitive information.

[0093] The third module is used to calculate the target sensitivity of the sensitive point corresponding to each power customer based on the comprehensive sensitivity and personalized sensitivity of the sensitive point corresponding to each power customer.

[0094] The fourth module is used to obtain the target sensitivity level of each power customer based on the preset sensitivity level and according to each target sensitivity and using the improved k-means clustering algorithm.

[0095] Optionally, the fourth module may include:

[0096] The first submodule is used to establish a corresponding number of clusters according to the number of sensitivity levels, and to apply an initial cluster center algorithm to determine an initial cluster center of each cluster.

[0097] The second submodule is used to calculate the Euclidean distance between the target sensitivity of each power customer and each initial cluster center, and allocate each target sensitivity to the cluster with the closest Euclidean distance.

[0098] The third submodule is used to calculate the average value of all data objects in each cluster and use the average value as the new cluster center of the cluster.

[0099] The fourth submodule is used for the second submodule and the third submodule until the new cluster centers of each cluster obtained after two iterations are the same.

[0100] Embodiment 3:

[0101] The present application embodiment provides an electronic device, such as Figure 3 As shown, it includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, any method in Embodiment 1 is implemented.

[0102] Embodiment 4:

[0103] An embodiment of the present application provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions enable a computer to execute any method in Example 1.

[0104] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for classifying power service sensitivity levels based on customer grouping, characterized in that: The specific steps include: Screening out non-positive feedback work orders from historical feedback customer service work orders, and extracting sensitive information from the non-positive feedback work orders, wherein the sensitive information includes sensitive points of each power customer; For each electricity customer, based on the sensitive information, the comprehensive sensitivity and personalized sensitivity of the corresponding sensitive point are calculated respectively; According to the comprehensive sensitivity and personalized sensitivity of the sensitive points corresponding to each power customer, the target sensitivity of the sensitive points corresponding to each power customer is calculated; Based on the preset sensitivity level, the target sensitivity level of each power customer is obtained according to each target sensitivity and using an improved k-means clustering algorithm.

2. A method for classifying power service sensitivity levels based on customer grouping according to claim 1, characterized in that: The historical feedback customer service work orders include praise type and non-positive feedback work orders, and the non-positive feedback work orders include complaint type, opinion type, suggestion type, report type and urging type; the comprehensive sensitivity of a power customer corresponding to a sensitive point is specifically: Among them, x i =m i / N i , i=1,2,…6; Where P1 represents the comprehensive sensitivity of power customer A to sensitive point B, x i represents the proportion of sensitive point B corresponding to power customer A in work order type i, i corresponds to the number of work order types in historical feedback customer service work orders, m i Indicates the number of work orders with sensitive point B in work order type i, N i represents the number of work orders in work order type i, ω i Represents the weight of the i-th work order type.

3. A method for classifying power service sensitivity levels based on customer grouping according to claim 2, characterized in that: The personalized sensitivity of one of the power customers corresponding to the sensitive point is as follows: Where P2 represents the personalized sensitivity of power customer A to sensitive point B, ω i represents the weight of the i-th work order type, P1 represents the comprehensive sensitivity of power customer A to sensitive point B, N i Indicates the number of work orders in work order type i, n i Indicates the number of work orders with sensitive point B and work order type i in historical feedback customer service work orders.

4. A method for classifying power service sensitivity levels based on customer grouping according to claim 3, characterized in that: The target sensitivity of each power customer corresponding to the sensitive point is specifically: P = P1 + P2; Where P2 represents the personalized sensitivity of power customer A to sensitive point B, P1 represents the comprehensive sensitivity of power customer A to sensitive point B, and P represents the target sensitivity of power customer A to sensitive point B.

5. The method for classifying the sensitivity level of electric power services based on customer grouping according to claim 1 is characterized in that: The target sensitivity level of each power customer is obtained according to each target sensitivity and using the improved k-means clustering algorithm, specifically: A1, establish a corresponding number of clusters according to the number of sensitivity levels, and apply the initial cluster center algorithm to determine the initial cluster center of each cluster; A2, calculate the Euclidean distance between the target sensitivity of each electricity customer and each initial cluster center, and assign each target sensitivity to the cluster with the closest Euclidean distance; A3, calculate the average value of all data objects in each cluster and use the average value as the new cluster center of the cluster; A4, loop A2 to A3 until the new cluster centers of each cluster obtained after two iterations are the same.

6. A method for classifying power service sensitivity levels based on customer grouping according to claim 5, characterized in that: The initial cluster center algorithm is applied to determine the initial cluster center of each cluster, specifically: A11, determining the target sensitivity of each power customer with the smallest value as the first target value; A12, calculating the Euclidean distance between the target sensitivity of other power customers and the first target value, and arranging the target sensitivities in ascending order according to the Euclidean distance; A13, in the sequence of target sensitivities arranged in ascending order, assigning the target sensitivity numbered 1+(i-1)(n-2) / (k-1) to the kth cluster, where i=1, 2...k, and n represents the total number of target sensitivities; A14, for the target sensitivities not classified into the clusters, calculating the Euclidean distance between each target sensitivity not classified into the clusters and each target sensitivity classified into the clusters, and classifying each target sensitivity not classified into the clusters into the cluster with the smallest Euclidean distance; A15, calculate the average value of the sensitivity of each target in each cluster, and determine the average value as the initial cluster center.

7. A system for classifying the sensitivity level of electric service based on customer grouping, applied to a method for classifying the sensitivity level of electric service based on customer grouping according to any one of claims 1 to 6, characterized in that: include: The first module is used to filter out non-positive feedback work orders from historical feedback customer service work orders, and extract sensitive information from the non-positive feedback work orders, wherein the sensitive information includes sensitive points of each power customer; The second module is used to calculate the comprehensive sensitivity and personalized sensitivity of the corresponding sensitive point for each power customer based on the sensitive information; The third module is used to calculate the target sensitivity of the sensitive point corresponding to each power customer based on the comprehensive sensitivity and personalized sensitivity of the sensitive point corresponding to each power customer; The fourth module is used to obtain the target sensitivity level of each power customer based on the preset sensitivity level and according to each of the target sensitivities and using the improved k-means clustering algorithm.

8. The power service sensitivity level classification system based on customer grouping according to claim 7 is characterized in that: The fourth module includes: The first submodule is used to establish a corresponding number of clusters according to the number of sensitivity levels, and to determine the initial cluster center of each cluster using an initial cluster center algorithm; The second submodule is used to calculate the Euclidean distance between the target sensitivity of each power customer and each initial cluster center, and assign each target sensitivity to the cluster with the closest Euclidean distance; The third submodule is used to calculate the average value of all data objects in each cluster and use the average value as the new cluster center of the cluster; The fourth submodule is used for the second submodule and the third submodule until the new cluster centers of each cluster obtained after two iterations are the same.

9. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method according to any one of claims 1 to 6 is implemented.

10. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, which enable a computer to execute the method of any one of claims 1-6.

Images