Base station evaluation method and computer program product

Abstract

The invention discloses a base station evaluation method and a computer program product, and relates to the technical field of wireless. The base station evaluation method comprises the following steps: processing obtained multi-dimensional index data of a base station to obtain a multi-view data matrix; on the basis of the multi-view data matrix, a public view map is constructed, and the public view map is used for representing the weight of each view; clustering processing is carried out on the public view map to obtain a clustering result of the base station, and the clustering result is the category of the base station in a preset number of clustering clusters; and based on the clustering result and the weight of each view in the base station, evaluating the base station to obtain the grade of the base station.

Description

Technical Field

[0001] This application relates to the field of wireless technology, and in particular to a base station evaluation method and computer program product. Background Technology

[0002] Base station classification assessment refers to classifying and ranking base stations according to specific standards in order to optimize network performance, resource allocation, or operation and maintenance strategies.

[0003] In related technologies, base station classification assessment generally adopts the expert experience method, that is, key assessment dimensions such as traffic, coverage, and energy consumption are manually selected by domain experts, and fixed weight coefficients are assigned to each dimension based on subjective experience. Finally, the comprehensive score of the base station is calculated by linear weighted summation, and the level is divided according to the preset threshold range.

[0004] However, this method relies on manually setting the weights of the indicators, which has obvious subjective flaws. For example, different experts may have different evaluations of the same base station, resulting in inconsistent results, and it is also inefficient. Summary of the Invention

[0005] This application provides a base station evaluation method and computer program product to at least solve the problem of low efficiency caused by reliance on manual labor in base station hierarchical evaluation in related technologies.

[0006] In a first aspect, embodiments of this application provide a base station evaluation method, including: The acquired multi-dimensional indicator data of the base stations are processed to obtain a multi-view data matrix; Based on the multi-view data matrix, a common view graph is constructed, which is used to represent the weight of each view; Clustering is performed on the public view map to obtain the clustering results of the base stations. The clustering results are the categories of the base stations in a preset number of clusters. Based on the clustering results and the weights of each view within the base station, the base station is evaluated to obtain its level.

[0007] Secondly, embodiments of this application provide a computer program product, the computer program product including a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, which, when executed by a computer, implement the steps of the method described in the first aspect.

[0008] In this embodiment, the acquired multi-dimensional indicator data of the base stations is first processed to obtain a multi-view data matrix. Then, based on the multi-view data matrix, a common view graph is constructed. The common view graph is used to represent the weight of each view. Next, the common view graph is clustered to obtain the clustering results of the base stations. The clustering results represent the category of the base station within a preset number of clusters. Finally, based on the clustering results and the weights of each view within the base station, the base station is evaluated to obtain its level. This embodiment constructs a common view graph based on the multi-dimensional indicator data of the base stations. The graph reflects the relationship between various indicators. Clustering the graph then yields the category of the base station within a preset number of clusters. The base stations are then classified according to their category. This method eliminates the need for manual indicator selection and achieves high efficiency in obtaining the classification results for different base stations. Attached Figure Description

[0009] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart of the base station evaluation method provided in the embodiments of this application; Figure 2 This is a schematic diagram of the base station evaluation device provided in the embodiments of this application; Figure 3 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0010] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0011] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0012] Among related technologies, the main methods for base station classification and evaluation include expert experience method, entropy weight method, and single-view evaluation method.

[0013] Among them, the expert experience method involves domain experts manually selecting key evaluation dimensions, such as traffic, coverage, and energy consumption, and assigning fixed weight coefficients to each dimension based on subjective experience. Finally, the comprehensive score of the base station is calculated by linear weighted summation, and then the levels are divided according to the preset threshold range, such as L1-L4.

[0014] The entropy weight method first requires manual division of evaluation indicators into several categories, such as network performance and energy efficiency. Then, representative indicators are selected from each category, and the objective weights are determined by calculating the information entropy of each indicator data. Information entropy reflects the degree of dispersion. Finally, the weighted sum is used to obtain the comprehensive score of the base station, and the level is divided based on the threshold.

[0015] The single-view evaluation method constructs a base station-index data matrix and employs collaborative recommendation algorithms, such as user-based collaborative filtering, to calculate the similarity between base stations, or applies swarm intelligence algorithms, such as improved ant colony optimization, to dynamically search for the optimal weight combination. This method first standardizes multi-source indicators, constructs an M×N dimensional evaluation matrix, and solves for the implicit weights of each indicator through matrix factorization or intelligent optimization, ultimately outputting a comprehensive score vector for the base station.

[0016] However, the aforementioned expert-based method relies on manually setting indicator weights, resulting in significant subjectivity. Different experts may disagree on their assessments of the same base station, leading to inconsistent results. This method lacks a data-driven, automated mechanism and cannot respond in real-time to network changes, such as sudden traffic fluctuations or equipment failures. Furthermore, once experience is solidified, it struggles to adapt to new service demands, causing assessment results to become disconnected from reality. This method also has high maintenance costs, requiring continuous investment of expert resources for weight adjustments.

[0017] While the entropy weight method calculates objective weights based on data fluctuations, it has inherent limitations. Its index classification relies on manual pre-setting, and the algorithm is sensitive to data quality, easily becoming biased in the presence of noise or missing values. Traditional entropy weight methods employ a static weighting model, failing to adjust the evaluation strategy based on dynamic factors such as network load and time period.

[0018] The single-view evaluation method compresses multiple base station indicator data into a single matrix, which easily loses the non-linear relationships between indicators and fails to reflect the complex relationships in real business scenarios.

[0019] Therefore, this application proposes a base station evaluation method and computer program product to at least solve the problems of reliance on manual intervention, adaptability to business factors, and lack of dimensional interaction.

[0020] The following is in conjunction with the appendix Figures 1 to 3 The present application provides a detailed description of a base station evaluation method and computer program product through specific embodiments and application scenarios.

[0021] like Figure 1 The diagram shown is a flowchart of a base station evaluation method provided in an embodiment of this application. Figure 1 As shown, the base station evaluation method may include the contents shown in S101 to S104.

[0022] In S101, the acquired multi-dimensional indicator data of the base station is processed to obtain a multi-view data matrix.

[0023] The multidimensional indicator data may include base station traffic, users, coverage, equipment and geographical environment, or other indicator data, depending on the actual application. This embodiment does not limit the data.

[0024] In this embodiment, multi-dimensional indicator data of the base station is first collected, and then a preset number of indicator data can be selected from them to obtain a multi-view data matrix.

[0025] A multi-view data matrix refers to a collection of data matrices obtained by observing or describing the same set of objects from multiple different perspectives. Each perspective (view) provides different characteristics or representations of the objects, thus forming a multi-view data structure.

[0026] Multi-view data matrices are used to improve the performance of data analysis tasks, such as clustering or classification, by integrating information from multiple views.

[0027] In S102, a common view graph is constructed based on the multi-view data matrix. The common view graph is used to represent the weight of each view.

[0028] This embodiment constructs a graph for each view to obtain a common view graph.

[0029] In S103, clustering processing is performed on the common view map to obtain the clustering results of the base stations. The clustering results are the categories of the base stations in a preset number of clusters.

[0030] This embodiment performs clustering based on the data structure information of the public view map to obtain the clustering result of each base station, that is, the category of each base station in a preset number of clusters.

[0031] The number of clusters can be set based on experience or determined based on the differences between adjacent feature values. This embodiment does not limit the number of clusters, depending on the actual application.

[0032] In S104, the base station is evaluated based on the clustering results and the weights of each view within the base station to obtain the base station's level.

[0033] This embodiment maps the clustering results to the base station value grading system, which can determine the base station's grade.

[0034] The base station value grading system can be pre-set or determined based on the data of each base station, depending on the actual application. This embodiment does not impose any limitations.

[0035] In this embodiment, the acquired multi-dimensional indicator data of the base stations is first processed to obtain a multi-view data matrix. Then, based on the multi-view data matrix, a common view graph is constructed. The common view graph is used to represent the weight of each view. Next, the common view graph is clustered to obtain the clustering results of the base stations. The clustering results represent the category of the base station within a preset number of clusters. Finally, based on the clustering results and the weights of each view within the base station, the base station is evaluated to obtain its level. This embodiment constructs a common view graph based on the multi-dimensional indicator data of the base stations. The graph reflects the relationship between various indicators. Clustering the graph then yields the category of the base station within a preset number of clusters. The base stations are then classified according to their category. This method eliminates the need for manual indicator selection and achieves high efficiency in obtaining the classification results for different base stations.

[0036] In one possible implementation of this application, the multidimensional indicator data of the acquired base station is processed to obtain a multi-view data matrix, which may include: forming a data matrix from the multidimensional data of the acquired base station, the data matrix including the original data of multiple views; preprocessing the data matrix to obtain a preprocessed data matrix; and associating the preprocessed data matrix with multi-view data based on the unique identifier of the base station to obtain a multi-view data matrix.

[0037] This embodiment divides the indicator data of different dimensions into multiple independent views, and performs standardized preprocessing on the original data of each view to ensure that the data of each view meets the requirements of the clustering algorithm.

[0038] First, the acquired data is filtered by view. Specifically, K multi-dimensional view data of the base station are collected, such as traffic, users, coverage, equipment, and geographical environment. From these, V key views are selected to form a data matrix. ,in, This represents the raw data for view v, where N is the total number of base stations. For view v, the first column is used to identify the uniqueness of the base station and is not included in the calculation. In this embodiment, the unique identifier (ID) of the base station is used to identify the uniqueness of the base station. The view combination can be dynamically adjusted according to the evaluation requirements.

[0039] Then, the features are preprocessed. In one possible implementation of this application, the data matrix is ​​preprocessed to obtain a preprocessed data matrix, which may include: deleting views in the data matrix whose feature missing rate exceeds the missing threshold to obtain a first data matrix; and standardizing the first data matrix to obtain a preprocessed data matrix.

[0040] Feature cleaning may include calculating the missing rate for each view: , If the missing rate is greater than the missing rate threshold, the view is discarded, for example. The view is discarded. The missing rate threshold can be set according to the actual situation and is subject to the actual application. This embodiment does not limit it. The missing rate is detected feature by feature. Features with more than 50% are removed. Irrelevant features such as test fields and constant values ​​are checked and removed. At the same time, invalid base station samples with missing base station IDs are discarded to ensure data integrity and uniqueness.

[0041] Feature classification and normalization can include establishing a data feature specification table, clarifying the data type, effective range and processing rules of each feature, and ensuring the quality of clustering features.

[0042] In one example, taking Physical Resource Block (PRB) utilization as an example, the PRB utilization type is defined as a numerical feature, ranging from [0, 100]. The processing flow first performs missing value imputation using the mean of base stations in the same area, then performs anomaly handling, truncating values ​​exceeding [0, 100] to the boundary; values ​​less than 0 are truncated to 0, and values ​​greater than 100 are truncated to 100. Finally, data standardization is performed. , where μ is the characteristic mean and σ is the characteristic standard deviation.

[0043] Secondly, sample alignment is performed, which involves associating multi-view data based on base station IDs to ensure cross-view data consistency. Base stations with missing views are removed to ensure that the number of base stations in each view is consistent.

[0044] Finally, the data is output to generate a standardized multi-view data matrix. As input to the clustering algorithm, this multi-view data matrix represents multiple views.

[0045] In one possible implementation of this application, constructing a common view map based on a multi-view data matrix may include: calculating the similarity between base stations using a Gaussian kernel function based on the multi-view data matrix; processing the similarity using the K-nearest neighbor algorithm to obtain a single-view map; and constructing a common view map using a weighted fusion algorithm based on the single-view map and the weights of the views.

[0046] In this embodiment, a graph is constructed for each view, and then a multi-view weighted algorithm is used to form a common view graph.

[0047] The construction of the single-view map is shown below: For the data matrices of each view after data processing The similarity between base stations is calculated using the Gaussian kernel function, as shown in the following formula:

[0048] in, Indicates base station In view eigenvectors, The table shows the median distance between base stations in this view.

[0049] To improve computational efficiency, this embodiment employs a sparsity strategy based on the K-Nearest Neighbors (KNN) algorithm. Optimization is performed to obtain a single-view atlas, as shown in the formula below:

[0050] in, Indicates base station The top K largest non-zero similarities.

[0051] The construction of the public view graph is as follows: A common graph matrix is ​​constructed using a multi-view weighted fusion strategy, as shown in the following formula:

[0052] in, The weights of the views are determined using a dynamic weighted fusion strategy.

[0053] The weight of a view can be configured in various ways, such as data quality configuration, business requirement configuration, or a combination of data quality and business requirement configuration. The choice can be made flexibly according to the evaluation scenario and the actual application shall prevail. This application does not impose any restrictions.

[0054] In one possible implementation of this application, the weight dynamic configuration based on data quality, i.e., the base station evaluation method may further include: obtaining the view information of each view; and determining the weight coefficient of each view based on the view information, wherein the weight coefficient is proportional to the view information.

[0055] This embodiment dynamically allocates weights based on the amount of information in the view; views with more information are assigned higher weight coefficients. The information content of each view is calculated as follows:

[0056] in, , A larger value indicates less effective information. Time weight is set to zero. The maximum weight is obtained at this time. To establish a positive correlation between information content and weight, the initial weight is calculated through an inverse linear transformation. , No. The weight calculation method for each view is as follows:

[0057] in, This mode achieves dynamic allocation of weights across multiple views through an automated weight optimization algorithm, requiring no manual intervention.

[0058] In one example, the information content scores for the network performance view, user value view, and coverage view are 0.85, 0.70, and 0.92, respectively. The data weight vector is obtained after calculation using the weighting formula. It accurately reflects the actual data contribution of each view.

[0059] In one possible implementation of this application, the weight configuration based on service requirements, i.e., the base station evaluation method, may further include: obtaining the baseline weights of each view in a pre-set service scenario, wherein the sum of the baseline weights of each view in the service scenario is one; adjusting the baseline weights of each view based on weight fine-tuning parameters to obtain the weight coefficient of each view, wherein the weight fine-tuning parameters include the adjustment parameters of each view, the weight fine-tuning parameters are less than a first value, and the sum of the adjustment parameters of each view is zero.

[0060] This embodiment employs a two-layer mechanism of baseline weighting and dynamic adjustment. A scenario-weight knowledge base is pre-built, defining baseline weight configurations for typical business scenarios, such as traffic assurance and energy-saving optimization. Among these, the first... k The baseline weight vector stored in each scene template is: , Then, based on the business objectives, one or more baseline weight vectors are matched and dynamically weighted, as shown in the calculation method below:

[0061] in, Let be the priority coefficient for the i-th scene. .

[0062] This mode supports fine-tuning parameters through weights. , must meet and This parameter can be set if it is believed that the weights between certain views need to be adjusted based on the baseline weights, ultimately generating a business weight vector. If not set The final business weight is then determined as follows: If set The final business weight is then determined as follows: .

[0063] In one example, network performance view, user value view, and coverage view are used. Figure 3 Taking a single view as an example, a preset network performance template User growth template and wide coverage template The business objectives for this period are set as "network expansion (60%) + user retention (40%)", and the configuration is as follows: Finally, the strategy weights are calculated. =[0.46,0.44,0.1], if not set The final business weights involved in the calculation If the actual business needs require an increase of 0.06 in the coverage factor, it can be set. =[-0.03,-0.03,0.06], finally obtained .

[0064] In one possible implementation of this application, the base station evaluation method may further include: obtaining the view information of each view; determining a first weight coefficient for each view based on the view information; obtaining the baseline weight of each view in a pre-set service scenario; adjusting the baseline weight in each view based on the weight fine-tuning parameters to obtain a second weight coefficient for each view; and determining the weight coefficient for each view based on the first weight coefficient and the second weight coefficient.

[0065] This embodiment combines dynamic configuration of data quality weights and weight configuration of business requirements to construct a dual-channel weight fusion mechanism of data quality and business strategy. This mechanism enables adaptive allocation of weights across multiple views, quantifies the inherent value of a view based on its information content, and reflects management intent based on current business objectives. It effectively balances the lack of flexibility of the inherent weighting method and the problem of the strategy losing focus in a purely data-driven approach.

[0066] In one example, dynamic weight configuration based on vector dot product is used to achieve intelligent weight allocation through a collaborative enhancement mechanism. When both data quality weight and business weight consider a view to be important, the weight enhancement effect is triggered. When the difference between the two is large, automatic smoothing is initiated to avoid extreme allocation. When both weights are low, the view is downweighted. This method can highlight the consensus dimension and suppress the disagreement dimension.

[0067] Determining the weight coefficient of each view based on the first weight coefficient and the second weight coefficient may include: determining the weight fusion coefficient based on the first weight coefficient and the second weight coefficient, wherein the elements in the weight fusion coefficient are the product of the elements in the weight coefficient and the elements in the second weight coefficient; and normalizing the weight fusion coefficient to obtain the weight coefficient of each view.

[0068] Among them, the data weight vector is dynamically configured by combining the weights of data quality. The business weight vector is generated based on the weight configuration of business requirements. Calculate element-wise product to achieve weight fusion The final adjusted result is obtained by normalization, where the first... k The normalized weights of each view are calculated as follows:

[0069] This embodiment focuses on the synergistic effect of two weight vectors in the same view.

[0070] In one instance, the weights are dynamically configured and calculated based on data quality. And the weights calculated based on business needs Weighted fusion yields After normalization, the final weights are: The dynamic configuration of data quality weights and the weight configuration of business requirements are considered important simultaneously. Figure 2 The weight is increased, and the view of disagreement between the two parties is strengthened. Figure 1 Then smooth the process, and consider unimportant views Figure 3 The weights are suppressed.

[0071] In one instance, a dynamic weighting configuration based on linear weighting is used, through adjustable mixing coefficients. This achieves a dynamic balance between data quality and business objectives. A higher weight indicates a greater emphasis on the amount of information in the view. The smaller the weight, the more it leans towards business objectives. exist Adjustments are made between these parameters to achieve a smooth transition between the amount of view information and business objectives.

[0072] Determining the weight coefficient of each view based on the first weight coefficient and the second weight coefficient may include: adjusting the first weight coefficient and the second weight coefficient using a preset mixing coefficient to obtain the weight coefficient of each view, wherein the preset mixing coefficient is proportional to the amount of view information.

[0073] Among them, the data weights are dynamically configured by combining data quality weights. And the business weight vector generated based on business requirements and weight configuration Using mixing coefficients in weights and The linear adjustment is as follows:

[0074] This embodiment focuses on the dynamic balance and collaborative optimization of the two weight vectors between business strategy guidance and real-time data feedback.

[0075] In one instance, the weights are dynamically configured and calculated based on data quality. And the weights calculated based on business needs , At that time, it was believed that the data weight accounted for a large proportion, and the final weight was determined accordingly. It will be more focused .

[0076] The final weighting result is obtained by selecting from the above various configurations. Similarity matrix of each view Weighted fusion is performed to obtain the final common view graph used for clustering. .

[0077] In one example, the input data is a matrix. (where n is the number of base stations), data graphs (eigenvalues ​​and eigenvectors) are used to discover structural information in the data and perform clustering to obtain the final clustering results. The main steps are: calculating the matrix. Laplace matrix ,in, The degree matrix is ​​a diagonal matrix. Eigenvalue decomposition of the Laplacian matrix L yields an increasing sequence of eigenvalues. and the corresponding feature vector The number of clusters can be manually determined. Alternatively, the optimal clustering data can be determined based on the differences between adjacent feature values. Before choosing The eigenvector corresponding to the smallest eigenvalue For feature vectors Clustering methods such as K-means are used to obtain the final clustering results. Each base station is explicitly divided into one of the k clusters, and the category label of each base station is determined. .

[0078] In one example, the final clustering results are mapped to the base station value grading system, specifically as follows: for category labels... i The base station set calculates the mean vector of metrics for each view within the cluster. ,like The average performance metrics within the network performance view cluster are used for a weighted comprehensive evaluation based on the final weight vector calculated in step two. ,right k Each category label is categorized by Arranging the fractions in descending order yields a monotonically decreasing result. , correspond The highest category label, correspond The lowest category label. Dynamic hierarchical mapping is performed based on the sorting results, with base station value levels defined from high to low as Level 1, Level 2, ..., Level N. N≤k The mapping table of dynamic mapping and cluster relationships is shown in Table 1 below.

[0079] Table 1

[0080] Based on the base station value classification results mapped in Table 1, we will customize the corresponding wireless network planning strategy accordingly.

[0081] This application embodiment divides indicator data of different dimensions into multiple independent views, dynamically calculates the contribution of each view, integrates multi-view information, performs clustering based on data graphs, and dynamically maps the clustering results to base station value levels to obtain the final evaluation result. It does not require manual selection of indicators, can automatically capture the nonlinear interaction effect between indicators, adapts to network changes by dynamically updating view weights, and finally outputs a hierarchical result with business interpretability.

[0082] like Figure 2 The diagram shown is a schematic representation of a base station evaluation device provided in an embodiment of this application. Figure 2 As shown, the base station evaluation device may include: a processing module 201, a construction module 202, a clustering module 203, and an evaluation module 204.

[0083] The system includes a processing module 201, which processes the acquired multi-dimensional indicator data of the base station to obtain a multi-view data matrix; a construction module 202, which constructs a common view map based on the multi-view data matrix, wherein the common view map is used to represent the weight of each view; a clustering module 203, which performs clustering processing on the common view map to obtain the clustering result of the base station, wherein the clustering result is the category of the base station in a preset number of clusters; and an evaluation module 204, which evaluates the base station based on the clustering result and the weight of each view within the base station to obtain the level of the base station.

[0084] In this embodiment, the processing module 201 first processes the acquired multidimensional indicator data of the base station to obtain multiple views. Then, the construction module 202 constructs a common view graph based on these views, which represents the weight of each view. Next, the clustering module 203 performs clustering based on the common view graph to obtain the clustering results of the base station, including categories. Finally, the evaluation module 204 evaluates the base station based on the clustering results and the weights of each view within the base station to obtain the base station's level. This embodiment constructs a common view graph based on the multidimensional indicator data of the base station. The graph reflects the relationships between various indicators. Clustering is then performed on the graph to obtain the clustering results of the base station. The base station is then classified according to the clustering results. This method eliminates the need for manual indicator selection and achieves high efficiency in obtaining the classification results for different base stations.

[0085] In one possible implementation of this application, the processing module 201 is configured to: form a data matrix from the acquired multidimensional data of the base station, the data matrix including the original data of multiple views; preprocess the data matrix to obtain a preprocessed data matrix; and perform multi-view data association on the preprocessed data matrix based on the unique identifier of the base station to obtain a multi-view data matrix.

[0086] In one possible implementation of this application, the processing module 201 is configured to: delete views in the data matrix whose feature missing rate exceeds a missing threshold to obtain a first data matrix; and perform standardization processing on the first data matrix to obtain a preprocessed data matrix.

[0087] In one possible implementation of this application, the construction module 202 is configured to: calculate the similarity between base stations using a Gaussian kernel function based on the multi-view data matrix; process the similarity using the K-nearest neighbor algorithm to obtain a single-view map; and construct a common view map using a weighted fusion algorithm based on the single-view map and the weights of the views.

[0088] In one possible implementation of this application, the construction module 202 is configured to: obtain the view information of each view; and determine the weight coefficient of each view based on the view information, wherein the weight coefficient is proportional to the view information.

[0089] In one possible implementation of this application, the construction module 202 is configured to: obtain the baseline weights of each view in a pre-set business scenario, wherein the sum of the baseline weights of each view in the business scenario is one; adjust the baseline weights of each view based on the weight fine-tuning parameters to obtain the weight coefficient of each view, wherein the weight fine-tuning parameters include the adjustment parameters of each view, the weight fine-tuning parameters are less than a first value, and the sum of the adjustment parameters of each view is zero.

[0090] In one possible implementation of this application, the construction module 202 is configured to: obtain the view information of each view; determine a first weight coefficient for each view based on the view information; obtain the baseline weight of each view in a pre-set business scenario; adjust the baseline weight of each view based on the weight fine-tuning parameters to obtain a second weight coefficient for each view; and determine the weight coefficient for each view based on the first weight coefficient and the second weight coefficient.

[0091] In one possible implementation of this application, the construction module 202 is configured to: determine a weight fusion coefficient based on the first weight coefficient and the second weight coefficient, wherein the elements in the weight fusion coefficient are the product of the elements in the first weight coefficient and the elements in the second weight coefficient; and normalize the weight fusion coefficient to obtain the weight coefficient of each view.

[0092] In one possible implementation of this application, the construction module 202 is configured to: adjust the first weight coefficient and the second weight coefficient using a preset mixing coefficient to obtain a weight coefficient for each view, wherein the preset mixing coefficient is proportional to the amount of view information.

[0093] The functions of the base station evaluation device in this application have been implemented. Figure 1 The method embodiments shown are described in detail. Therefore, for any parts not covered in detail in this embodiment, please refer to the relevant descriptions in the foregoing embodiments, which will not be repeated here.

[0094] like Figure 3 As shown, this application embodiment also provides an electronic device 300, including a processor 301, a memory 302, and a program or instructions stored in the memory 302 and executable on the processor 301. When the program or instructions are executed by the processor 301, they implement the various processes of the above-described base station evaluation processing method embodiment and achieve the same technical effect. To avoid repetition, they will not be described again here.

[0095] Optionally, embodiments of this application also provide a computer-readable storage medium storing a computer program. When executed by a processor, this computer program implements the various processes of the above-described base station evaluation method embodiments and achieves the same technical effects. To avoid repetition, it will not be described again here. The computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.

[0096] Optionally, this application embodiment also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions, which, when executed by a computer, implement the various processes of the above-described base station evaluation method embodiment and achieve the same technical effect. To avoid repetition, they will not be described again here.

[0097] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0098] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, 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 is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0099] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A base station evaluation method, characterized in that, include: The acquired multi-dimensional indicator data of the base stations are processed to obtain a multi-view data matrix; Based on the multi-view data matrix, a common view graph is constructed, which is used to represent the weight of each view; Clustering is performed on the public view map to obtain the clustering results of the base stations. The clustering results are the categories of the base stations in a preset number of clusters. Based on the clustering results and the weights of each view within the base station, the base station is evaluated to obtain its level.

2. The method according to claim 1, characterized in that, The process of obtaining the multi-dimensional indicator data of the base station to obtain a multi-view data matrix includes: The acquired multidimensional data from the base station is used to form a data matrix, which includes the original data from multiple views. The data matrix is ​​preprocessed to obtain a preprocessed data matrix; Based on the unique identifier of the base station, the preprocessed data matrix is ​​correlated with multiple views to obtain a multi-view data matrix.

3. The method according to claim 2, characterized in that, The data matrix is ​​preprocessed to obtain a preprocessed data matrix, including: Views in the data matrix whose feature missing rate exceeds the missing threshold are deleted to obtain the first data matrix; The first data matrix is ​​standardized to obtain a preprocessed data matrix.

4. The method according to claim 1, characterized in that, The construction of a common view map based on the multi-view data matrix includes: Based on the multi-view data matrix, the similarity between base stations is calculated using a Gaussian kernel function; The similarity is processed using the K-nearest neighbor algorithm to obtain a single-view map; Based on the single-view atlas and the weights of the views, a weighted fusion algorithm is used to construct a common view atlas.

5. The method according to claim 4, characterized in that, The method further includes: Obtain the amount of view information for each view; Based on the amount of view information, a weight coefficient is determined for each view, wherein the weight coefficient is proportional to the amount of view information.

6. The method according to claim 4, characterized in that, The method further includes: Obtain the baseline weights of each view in a pre-set business scenario, wherein the sum of the baseline weights of each view in the business scenario is one; Based on the weight fine-tuning parameters, the baseline weights in each view are adjusted to obtain the weight coefficient of each view. The weight fine-tuning parameters include the adjustment parameters of each view. The weight fine-tuning parameters are less than a first value, and the sum of the adjustment parameters of each view is zero.

7. The method according to claim 4, characterized in that, The method further includes: Obtain the amount of view information for each view; Based on the amount of view information, a first weight coefficient is determined for each view; Obtain the baseline weights of each view in the pre-defined business scenario; Based on the weight fine-tuning parameters, the baseline weights in each view are adjusted to obtain the second weight coefficient for each view; The weight coefficient for each view is determined based on the first weight coefficient and the second weight coefficient.

8. The method according to claim 7, characterized in that, Determining the weight coefficient for each view based on the first weight coefficient and the second weight coefficient includes: Based on the first weight coefficient and the second weight coefficient, a weight fusion coefficient is determined, wherein the elements in the weight fusion coefficient are the product of the elements in the first weight coefficient and the elements in the second weight coefficient; The weight fusion coefficients are normalized to obtain the weight coefficients for each view.

9. The method according to claim 7, characterized in that, Determining the weight coefficient for each view based on the first weight coefficient and the second weight coefficient includes: The first weight coefficient and the second weight coefficient are adjusted using a preset mixing coefficient to obtain the weight coefficient of each view, wherein the preset mixing coefficient is proportional to the amount of information in the view.

10. A computer program product, characterized in that, The computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions that, when executed by a computer, implement the steps of the method as described in any one of claims 1 to 9.

Images