CN116156538A - Quality difference cell root cause positioning method based on SMOTE-ReliefF-XGBoost algorithm - Google Patents

Quality difference cell root cause positioning method based on SMOTE-ReliefF-XGBoost algorithm Download PDF

Info

Publication number
CN116156538A
CN116156538A CN202310153935.2A CN202310153935A CN116156538A CN 116156538 A CN116156538 A CN 116156538A CN 202310153935 A CN202310153935 A CN 202310153935A CN 116156538 A CN116156538 A CN 116156538A
Authority
CN
China
Prior art keywords
algorithm
relieff
base station
smote
root cause
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202310153935.2A
Other languages
Chinese (zh)
Inventor
王宁
崔天姿
申凌峰
陈鑫
史文祥
陈任翔
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhengzhou University
Original Assignee
Zhengzhou University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zhengzhou University filed Critical Zhengzhou University
Priority to CN202310153935.2A priority Critical patent/CN116156538A/en
Publication of CN116156538A publication Critical patent/CN116156538A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/04—Arrangements for maintaining operational condition
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/14—Network analysis or design
    • H04L41/142—Network analysis or design using statistical or mathematical methods
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/14—Network analysis or design
    • H04L41/145—Network analysis or design involving simulating, designing, planning or modelling of a network
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D30/00—Reducing energy consumption in communication networks
    • Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Physics & Mathematics (AREA)
  • Algebra (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Physics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

The invention provides a quality difference cell root cause positioning method based on an SMOTE-ReliefF-XGBoost algorithm, which comprises the following steps: establishing a cell base station information matrix, and labeling a base station quality difference root cause according to historical experience; adopting a proper data preprocessing method according to the characteristics of different characteristics of the base station information; equalizing different types of data by using an SMOTE oversampling method; because of redundancy of the data features, firstly calculating weight values of the features by adopting a ReliefF algorithm, and then selecting important features according to a Sequential Forward Selection method; finally, the processed data are sent into 4 independent XGBoost classification models for training, and a genetic algorithm is adopted for parameter tuning; the invention effectively avoids the problem of over-fitting by adopting the SMOTE over-sampling technology, saves the memory resources of a computer and eliminates redundant features by adopting a Relieff feature selection algorithm, improves the performance of the model, and finally reduces the training time of the model by adopting a genetic algorithm to optimize the XGBoost model.

Description

Quality difference cell root cause positioning method based on SMOTE-ReliefF-XGBoost algorithm
Technical Field
The invention belongs to the technical field of communication, and particularly relates to a quality difference cell root cause positioning method based on an SMOTE-ReliefF-XGBoost algorithm.
Background
With the continuous high-speed development of mobile communication technology, wireless networks have become an indispensable part in life and work, and the current mobile communication networks are increasingly large in scale, so that the network structure is increasingly complex, and various network problems are also led to.
The wireless network transmits information through the base stations, and each base station contains the local area network data information and mainly comprises the characteristic information of bandwidth, frequency band, macro station power, air interface uplink/downlink service flow, busy hour uplink/downlink PRB average utilization rate, double-flow duty ratio and the like.
Base station wireless network differences are affected by factors such as the number of users and base station power; if the base station wireless network is to be optimized, the quality difference root cause of the base station needs to be known first, so that the medicine can be applied to the patient. The reason of analyzing the quality difference according to the quality difference base station data is the difficult problem that the wireless network optimization is required to face, if the quality difference base station is not optimized in time, the user experience can be affected, and even the user loss can be caused.
At present, the quality difference root cause of the base station wireless network is mainly analyzed by manpower, so that the time and the labor are wasted; the existing algorithm realizes the automatic quality difference root cause analysis of the base station wireless network, the existing algorithm mainly judges the characteristics through subjective judgment, the application and the use of the characteristic information cannot be realized, and the existing automatic quality difference root cause positioning algorithm is simple and cannot accurately position the quality difference root cause.
Therefore, aiming at the technical problems, the quality difference cell root cause positioning method based on the SMOTE-reliefF-XGBoost algorithm is provided to realize automatic and reliable positioning of the quality difference root cause.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a quality difference cell root cause positioning method based on an SMOTE-reliefF-XGBoost algorithm, which improves the reliability of intelligent quality difference root cause positioning.
The invention is achieved based on the following steps, aiming at the above purpose, and mainly comprising:
s1: establishing a cell base station information matrix, and marking the quality difference reasons of all the base stations according to different state information according to the information such as the bandwidth, the scene, the average user number, the total flow and the like of the cell base stations; the reason for poor quality of the base station is: coverage problems, interference problems, capacity problems, and antenna feed problems, corresponding to tags 1 through 4, respectively;
s2: and (3) tag coding, namely marking the degree of the problem corresponding to the four tags in the S1. Wherein the coverage problem is divided into weak coverage, over coverage and overlapped coverage, corresponding to labels 0-2; the interference problem is divided into external interference and system internal interference, and the corresponding labels are 0-1; the capacity problem is divided into high load, traffic load imbalance among License limited sectors and high load, and other reasons correspond to tags 0-1; the antenna feed problem is divided into a double-current duty ratio problem and an antenna feed disconnection problem, and the two problems correspond to tags 0-1; in addition, the quality difference of some base stations does not belong to any problem, and is called a blank set, and the codes of the sub-problems under each problem are immediately followed, for example, the blank set codes in the coverage problem are 3;
s3: and preprocessing data, and respectively processing the data according to the characteristics of different information of the cell base station. Aiming at discrete information such as the times (times) of RRC connection establishment failure caused by failure of frequency band, area, bandwidth and resource allocation, adopting a one-hot mode for coding processing; for large-unit information such as self-busy average flow (GByte) and air interface uplink traffic flow (GByte) of a downlink cell, adopting a Z-Score method for standardization treatment;
s4: data equalization, namely equalizing the sub-problem data under different quality difference reasons by using an SMOTE method, so that the data quantity of the base stations corresponding to the sub-problem labels under the problem labels is almost the same;
s5: feature selection, namely calculating importance scores of the features of the balanced data under the label problems in the step S4 through a reliefF algorithm, selecting important features corresponding to the label problems through a Sequential Forward Selection method, and eliminating irrelevant features;
s6: model training, namely, sending a matrix formed by the features selected by aiming at each label problem and the sub-problem labels in the step S5 into 4 XGBoost classification models to independently train, adjusting by using a genetic optimization algorithm to obtain super parameters with extremely small loss function (cross entropy function), and testing the performance of the model through cross verification;
s7: and (3) positioning the quality difference root cause, applying the optimal super parameters debugged in the step (S6) to the model, and inputting the cell base station information matrix into the model to obtain the root cause of the quality difference of the base station, thereby optimizing the layout and related configuration of the base station according to the quality difference root cause.
The steps S1-S3 can be specifically described as follows:
suppose that the cell base station information is represented by f= { X 1 ,X 2 ,...,X n } T ＝{F 1 ,F 2 ,...,F m Representation of }, where
Figure BDA0004091636750000031
1≤k≤n,1≤i≤m。
The cell base station tag is formed by L= { L 1 ,L 2 ,L 3 ,L 4 Representation of }, where
Figure BDA0004091636750000032
1≤j≤4。
For the characteristic of larger value in the cell base station information F, according to
Figure BDA0004091636750000033
The formula normalizes its eigenvalues, where μ i Is the characteristic mean value sigma i Is the standard deviation of the features.
And (3) encoding the discrete value characteristics in the cell base station information F by adopting a one-hot mode. And finally, dividing the data into a training set and a testing set.
The S4 may be specifically described as:
first randomly selecting a few samples X from a training set t T is more than or equal to 1 and less than or equal to n, and then according to the Euclidean distance formula
Figure BDA0004091636750000034
The distance from the remaining samples is calculated and one is randomly selected from k (k=3) nearest neighbor samples.
Finally in a few samples X t And randomly selecting a point on the mapping path between the randomly selected samples as a newly sampled sample. And repeatedly sampling for a plurality of times until the data volume of various samples reaches balance.
And in the S5, the ReliefF algorithm randomly selects one sample R from the training set, calculates the distances between the samples of the same kind and different kinds according to the Euclidean distance formula, and selects the samples of the same kind and different kinds of k nearest neighbors respectively.
Each feature weight is iteratively calculated according to:
Figure BDA0004091636750000041
wherein diff (F i ,R,H j ) Representing samples R and H j In feature F i Difference of the upper and lower parts, H j Represents the j nearest neighbor sample in class C e class (R), M j (C) Representing categories
Figure BDA0004091636750000042
The j-th nearest neighbor sample in the list, m represents the iteration times, and k represents the number of selected nearest neighbor samples.
diff(A,R 1 ,R 2 ) Is calculated as follows:
Figure BDA0004091636750000043
the initial feature subset S selected by the Sequential Forward Selection algorithm in S5 is
Figure BDA0004091636750000044
And (3) iteratively adding features into the S one by one according to the feature weight information calculated by the Relieff algorithm until a feature subset which enables the model performance to be optimal is found out.
The nature of the XGBoost classification algorithm in S6 is to integrate several base classifier models into one model set based on the base classifier.
The essence of the genetic optimization algorithm in the S6 is to realize parameter tuning through a computer simulation of the process of biological and human evolution.
The invention has the beneficial effects that:
1. according to the invention, the correlation weight between the feature and the label is calculated by adopting a ReliefF algorithm, and the effectiveness of the selected feature can be improved by supervising the feature selection mode;
2. aiming at the unbalanced characteristic of the data, the invention adopts the SMOTE oversampling algorithm to balance the data quantity of each class, thereby effectively avoiding the influence of overfitting and weight deviation of the model;
3. according to the invention, aiming at each big root cause, a method of dividing and controlling is adopted, and four independent XGBoost classification models are respectively applied to position the quality difference root cause of the base station, so that the reliability of positioning the root cause is effectively improved;
4. according to the invention, the super parameters of the XGBoost classification model are optimized by adopting a genetic algorithm, so that the training time of the model can be effectively reduced and the performance of the model can be improved.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following description will briefly explain the embodiments of the present invention or the drawings used in the description of the prior art, and it is obvious that the drawings described below are only some embodiments of the present invention, and other drawings can be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a general system flow chart of a bad cell root cause positioning method based on the SMOTE-ReliefF-XGBoost algorithm.
FIG. 2 is a block diagram of an overall system algorithm of a bad cell root cause positioning method based on the SMOTE-ReliefF-XGBoost algorithm.
Detailed Description
The present invention will be described in detail with reference to the flow embodiments shown in the drawings. However, the present invention is not limited to the embodiments, and variations in structure, method or function according to the method, which are made by those skilled in the art, are included in the scope of the present invention based on the examples of the present invention.
The invention provides a quality difference cell root cause positioning method based on an SMOTE-ReliefF-XGBoost algorithm, which is shown by referring to figures 1-2 and comprises the following steps:
s1: establishing a cell base station information matrix, and marking the quality difference reasons of all the base stations according to different state information according to the information such as the bandwidth, the scene, the average user number, the total flow and the like of the cell base stations; the reason for poor quality of the base station is: coverage problems, interference problems, capacity problems, and antenna feed problems, corresponding to tags 1 through 4, respectively;
s2: and (3) tag coding, namely marking the degree of the problem corresponding to the four tags in the S1. Wherein the coverage problem is divided into weak coverage, over coverage and overlapped coverage, corresponding to labels 0-2; the interference problem is divided into external interference and system internal interference, and the corresponding labels are 0-1; the capacity problem is divided into high load, traffic load imbalance among License limited sectors and high load, and other reasons correspond to tags 0-1; the antenna feed problem is divided into a double-current duty ratio problem and an antenna feed disconnection problem, and the two problems correspond to tags 0-1; in addition, the quality difference of some base stations does not belong to any problem, and is called a blank set, and the codes of the sub-problems under each problem are immediately followed, for example, the blank set codes in the coverage problem are 3;
s3: and preprocessing data, and respectively processing the data according to the characteristics of different information of the cell base station. Aiming at discrete information such as the times (times) of RRC connection establishment failure caused by failure of frequency band, area, bandwidth and resource allocation, adopting a one-hot mode for coding processing; for large-unit information such as self-busy average flow (GByte) and air interface uplink traffic flow (GByte) of a downlink cell, adopting a Z-Score method for standardization treatment;
s4: data equalization, namely equalizing the sub-problem data under different quality difference reasons by using an SMOTE method, so that the data quantity of the base stations corresponding to the sub-problem labels under the problem labels is almost the same;
s5: feature selection, namely calculating importance scores of the features of the balanced data under the label problems in the step S4 through a reliefF algorithm, selecting important features corresponding to the label problems through a Sequential Forward Selection method, and eliminating irrelevant features;
s6: model training, namely, sending a matrix formed by the features selected by aiming at each label problem and the sub-problem labels in the step S5 into 4 XGBoost classification models to independently train, adjusting by using a genetic optimization algorithm to obtain super parameters with extremely small loss function (cross entropy function), and testing the performance of the model through cross verification;
s7: and (3) positioning the quality difference root cause, applying the optimal super parameters debugged in the step (S6) to the model, and inputting the cell base station information matrix into the model to obtain the root cause of the quality difference of the base station, thereby optimizing the layout and related configuration of the base station according to the quality difference root cause.
Wherein, the steps S1-S3 can be specifically described as follows:
suppose that the cell base station information is represented by f= { X 1 ,X 2 ,...,X n } T ＝{F 1 ,F 2 ,...,F m Representation of }, where
Figure BDA0004091636750000071
1≤k≤n,1≤i≤m。
The cell base station tag is formed by L= { L 1 ,L 2 ,L 3 ,L 4 Representation of }, where
Figure BDA0004091636750000072
1≤j≤4。
For the characteristic of larger value in the cell base station information F, according to
Figure BDA0004091636750000073
The formula normalizes its eigenvalues, where μ i Is the characteristic mean value sigma i Is the standard deviation of the features.
And (3) encoding the discrete value characteristics in the cell base station information F by adopting a one-hot mode. And finally, dividing the data into a training set and a testing set.
Specifically, S4 may be specifically described as:
first randomly selecting a few samples X from a training set t T is more than or equal to 1 and less than or equal to n, and then according to the Euclidean distance formula
Figure BDA0004091636750000074
The distance from the remaining samples is calculated and one is randomly selected from k (k=3) nearest neighbor samples.
Finally in a few samples X t And randomly selecting a point on the mapping path between the randomly selected samples as a newly sampled sample. And repeatedly sampling for a plurality of times until the data volume of various samples reaches balance.
Specifically, as shown in fig. 2, in S5, the ReliefF algorithm randomly selects one sample R from the training set, calculates distances between the samples of the same kind and different kinds according to the euclidean distance formula, and selects k nearest neighbor samples of the same kind and different kinds.
Each feature weight is iteratively calculated according to:
Figure BDA0004091636750000075
wherein diff (F i ,R,H j ) Representing samples R and H j In feature F i Difference of the upper and lower parts, H j Represents the j nearest neighbor sample in class C e class (R), M j (C) Representing categories
Figure BDA0004091636750000076
The j-th nearest neighbor sample in the list, m represents the iteration times, and k represents the number of selected nearest neighbor samples.
diff(A,R 1 ,R 2 ) Is calculated as follows:
Figure BDA0004091636750000081
specifically, the first selected by the Sequential Forward Selection algorithm in S5The initial feature subset S is
Figure BDA0004091636750000082
And (3) iteratively adding the features into the feature subset S one by one according to the feature weight information calculated by the Relieff algorithm in the step (S4) until the feature subset which enables the model performance to be optimal is found out.
As shown in fig. 2, the nature of the XGBoost classification algorithm in S6 is to integrate several base classifier models into one model set based on the base classifier, the base classifier generally adopts a decision tree, and the principle of the XGBoost classification algorithm is that the XGBoost classification algorithm is continuously increased on the basis of one base classifier until the accuracy rate of the algorithm is not significantly increased any more.
Specifically, suppose there are K base classifiers f in XGBoost k K is 1.ltoreq.K, for sample X of i (1.ltoreq.i.ltoreq.n) i The output of the jth (1.ltoreq.j.ltoreq.4) XGBoost model for the ith sample is:
Figure BDA0004091636750000083
the following objective function is constructed:
Figure BDA0004091636750000084
training of XGBoost model by minimizing the objective function argmin Obj j Realizing the method.
Specifically, the optimization process is realized according to the genetic optimization algorithm in the step S6, and the essence of the optimization process is that parameter tuning is realized through a process of simulating biological and human evolution by a computer.
The foregoing description of the preferred embodiments of the present invention has been provided for the purpose of illustrating the general principles of the present invention and is not to be construed as limiting the scope of the invention in any way. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention, and other embodiments of the present invention as will occur to those skilled in the art without the exercise of inventive faculty, are intended to be included within the scope of the present invention.

Claims (7)

1. A quality difference cell root cause positioning method based on an SMOTE-ReliefF-XGBoost algorithm is characterized by comprising the following steps:
s1: establishing a cell base station information matrix, and marking the quality difference reasons of all the base stations according to different state information according to the information such as the bandwidth, the scene, the average user number, the total flow and the like of the cell base stations; the reason for poor quality of the base station is: coverage problems, interference problems, capacity problems, and antenna feed problems, corresponding to tags 1 through 4, respectively;
s2: tag coding, namely marking the degrees of problems corresponding to the four tags in the step S1; wherein the coverage problem is divided into weak coverage, over coverage and overlapped coverage, corresponding to labels 0-2; the interference problem is divided into external interference and system internal interference, and the corresponding labels are 0-1; the capacity problem is divided into high load, traffic load imbalance among License limited sectors and high load, and other reasons correspond to tags 0-1; the antenna feed problem is divided into a double-current duty ratio problem and an antenna feed disconnection problem, and the two problems correspond to tags 0-1; in addition, the quality difference of some base stations does not belong to any problem, and is called a blank set, and the codes of the sub-problems under each problem are immediately followed, for example, the blank set codes in the coverage problem are 3;
s3: data preprocessing, namely respectively processing the data according to the characteristics of different information of the cell base station; aiming at discrete information such as the times of RRC connection establishment failure caused by failure of frequency band, area, bandwidth and resource allocation, adopting a one-hot mode coding process; for large-unit information such as self-busy average flow (GByte) and air interface uplink traffic flow (GByte) of a downlink cell, adopting a Z-Score method for standardization treatment;
s4: data equalization, namely equalizing the sub-problem data under different quality difference reasons by using an SMOTE method, so that the data quantity of the base stations corresponding to the sub-problem labels under the problem labels is almost the same;
s5: feature selection, namely calculating importance scores of the features of the balanced data under the label problems in the step S4 through a reliefF algorithm, selecting important features corresponding to the label problems through a Sequential Forward Selection method, and eliminating irrelevant features;
s6: model training, namely, sending a matrix formed by the features selected by aiming at each label problem and the sub-problem labels in the step S5 into 4 XGBoost classification models to independently train, adjusting by using a genetic optimization algorithm to obtain super parameters with extremely small loss function (cross entropy function), and testing the performance of the model through cross verification;
s7: and (3) positioning the quality difference root cause, applying the optimal super parameters debugged in the step (S6) to the model, and inputting the cell base station information matrix into the model to obtain the root cause of the quality difference of the base station, thereby optimizing the layout and related configuration of the base station according to the quality difference root cause.
2. The method for positioning root cause of poor quality cell based on SMOTE-ReliefF-XGBoost algorithm according to claim 1, wherein the steps S1-S3 can be specifically described as:
suppose that the cell base station information is represented by f= { X 1 ,X 2 ,…,X n } T ＝{F 1 ,F 2 ,…,F m Representation of }, where
Figure FDA0004091636730000021
The cell base station tag is formed by L= { L 1 ,L 2 ,L 3 ,L 4 Represented by }, where->
Figure FDA0004091636730000022
For the feature of larger value in the cell base station information F, according to +.>
Figure FDA0004091636730000023
The formula normalizes its eigenvalues, where μ i Is the characteristic mean value sigma i Is the standard deviation of the features; for the discrete value characteristics in the cell base station information F, adopting a one-hot mode for coding; and finally, dividing the data into a training set and a testing set.
3. The method for positioning root cause of poor quality cell based on SMOTE-ReliefF-XGBoost algorithm according to claim 1, wherein the SMOTE algorithm in S4 can be specifically described as:
first randomly selecting a few samples X from a training set t T is more than or equal to 1 and less than or equal to n, and then according to the Euclidean distance formula
Figure FDA0004091636730000024
Calculating the distance between the sample and the rest samples, randomly selecting one from k (k=3) nearest neighbor samples, and finally, selecting a few samples X t Randomly selecting a point on a mapping path between the randomly selected samples as a newly sampled sample; and repeatedly sampling for a plurality of times until the data volume of various samples reaches balance.
4. The method for positioning root causes of cells with poor quality based on the SMOTE-ReliefF-XGBoost algorithm according to claim 1, wherein the ReliefF algorithm in S5 randomly selects a sample R from a training set, calculates distances between the samples of the same kind and different kinds according to a euclidean distance formula, and selects k nearest neighbor samples of the same kind and different kinds respectively; each feature weight is iteratively calculated according to:
Figure FDA0004091636730000031
wherein diff (F i ,R,H j ) Representing samples R and H j In feature F i Difference of the upper and lower parts, H j Represents the j nearest neighbor sample in class C e class (R), M j (C) Representing categories
Figure FDA0004091636730000032
The j-th nearest neighbor sample in the list, m represents the iteration times, and k represents the number of the selected nearest neighbor samples; diff (A, R) 1 ,R 2 ) Is calculated as follows:
Figure FDA0004091636730000033
5. the method for positioning root cause of poor quality cell based on SMOTE-ReliefF-XGBoost algorithm as set forth in claim 1, wherein the initial feature subset S selected by Sequential Forward Selection algorithm in S5 is
Figure FDA0004091636730000034
And (5) iteratively adding features into the S one by one according to the feature weight information calculated by the Relieff algorithm until the best model performance is found out.
6. The method for positioning root cause of poor quality cell based on SMOTE-ReliefF-XGBoost algorithm according to claim 1, wherein the nature of XGBoost classification algorithm in S6 is to integrate several base classifier models into one model set based on base classifier.
7. The method for positioning root cause of poor quality cell based on SMOTE-ReliefF-XGBoost algorithm according to claim 1, wherein the essence of the genetic optimization algorithm in S6 is to realize parameter tuning through the process of computer simulation of biological and human evolution.
CN202310153935.2A 2023-02-23 2023-02-23 Quality difference cell root cause positioning method based on SMOTE-ReliefF-XGBoost algorithm Pending CN116156538A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310153935.2A CN116156538A (en) 2023-02-23 2023-02-23 Quality difference cell root cause positioning method based on SMOTE-ReliefF-XGBoost algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310153935.2A CN116156538A (en) 2023-02-23 2023-02-23 Quality difference cell root cause positioning method based on SMOTE-ReliefF-XGBoost algorithm

Publications (1)

Publication Number Publication Date
CN116156538A true CN116156538A (en) 2023-05-23

Family

ID=86354081

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310153935.2A Pending CN116156538A (en) 2023-02-23 2023-02-23 Quality difference cell root cause positioning method based on SMOTE-ReliefF-XGBoost algorithm

Country Status (1)

Country Link
CN (1) CN116156538A (en)

Similar Documents

Publication Publication Date Title
CN110324170B (en) Data analysis equipment, multi-model co-decision system and method
CN109495920B (en) Wireless communication network feature imaging method, equipment and computer program product
CN114269007A (en) Method, device and method storage medium for determining energy-saving strategy of base station
WO2022057321A1 (en) Method and apparatus for detecting anomalous link, and storage medium
CN112513882B (en) Methods, devices and computer readable media related to detection of cell conditions in a wireless cellular network
CN113709779A (en) Cellular network fault diagnosis method
CN108551491A (en) The mobile Internet of Things cloud system of heterogeneous network spatial multi time based on artificial intelligence
CN113348691B (en) Method and apparatus for fault prediction in network management
Moysen et al. On the potential of ensemble regression techniques for future mobile network planning
Manalastas et al. A data-driven framework for inter-frequency handover failure prediction and mitigation
Iqbal et al. Convolutional neural network-based deep Q-network (CNN-DQN) resource management in cloud radio access network
CN113676357B (en) Decision method for edge data processing in power internet of things and application thereof
CN115422260A (en) Base station activity analysis method and device, electronic equipment and readable storage medium
Bakri et al. Channel stability prediction to optimize signaling overhead in 5G networks using machine learning
Liu et al. GA-AdaBoostSVM classifier empowered wireless network diagnosis
CN111935761B (en) Automatic positioning method and system for quality difference root cause of mobile communication network
CN116156538A (en) Quality difference cell root cause positioning method based on SMOTE-ReliefF-XGBoost algorithm
US20240129801A1 (en) Systems and methods for determining, monitoring, and predicting application capacity in multiple radio access networks
Sánchez-González et al. On learning mobility patterns in cellular networks
CN110072197A (en) A kind of preferably intelligent switching and selecting method of emergency communication transmission channel
CN113824670B (en) 5G slice air interface protocol stack scheduling method and device and computing equipment
CN114205238A (en) Network resource optimization and model training method, device, storage medium and equipment
US20240380676A1 (en) Model configuration method and apparatus
Ge Distinguished capabilities of Artificial Intelligence wireless communication systems
WO2024027424A1 (en) Data quality measurement method and device

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination