CN113569902A - Method and system for analyzing satisfaction evaluation quality of aviation enterprise based on Lekter scale - Google Patents

Method and system for analyzing satisfaction evaluation quality of aviation enterprise based on Lekter scale Download PDF

Info

Publication number
CN113569902A
CN113569902A CN202110630665.0A CN202110630665A CN113569902A CN 113569902 A CN113569902 A CN 113569902A CN 202110630665 A CN202110630665 A CN 202110630665A CN 113569902 A CN113569902 A CN 113569902A
Authority
CN
China
Prior art keywords
evaluation
data
evaluation data
clustering
satisfaction
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
CN202110630665.0A
Other languages
Chinese (zh)
Inventor
郭朔辰
丛玮
郑洪峰
王智勇
郑辰子
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.)
Variflight Technology Co ltd
Original Assignee
Variflight Technology Co ltd
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 Variflight Technology Co ltd filed Critical Variflight Technology Co ltd
Priority to CN202110630665.0A priority Critical patent/CN113569902A/en
Publication of CN113569902A publication Critical patent/CN113569902A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06395—Quality analysis or management
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/01—Customer relationship services
    • G06Q30/015—Providing customer assistance, e.g. assisting a customer within a business location or via helpdesk
    • G06Q30/016—After-sales

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • Development Economics (AREA)
  • Mathematical Physics (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • General Engineering & Computer Science (AREA)
  • Educational Administration (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Computational Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Evolutionary Biology (AREA)
  • Databases & Information Systems (AREA)
  • Software Systems (AREA)
  • Algebra (AREA)
  • Accounting & Taxation (AREA)
  • Computing Systems (AREA)
  • Evolutionary Computation (AREA)
  • Finance (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Probability & Statistics with Applications (AREA)
  • Game Theory and Decision Science (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The invention discloses a method and a system for analyzing the satisfaction evaluation quality of an aviation enterprise based on a Lekter scale, which specifically comprise the following steps: collecting aeronautical enterprise satisfaction data by using a Lekter scale to form first evaluation data, removing invalid comments in the first evaluation data, wherein the first evaluation data comprise evaluation objects, evaluation scores and evaluation items, and screening the evaluation items of the first evaluation data to form second evaluation data; randomly selecting a plurality of clustering centers according to the obtained second evaluation data, and clustering the second evaluation data to form a clustering domain to obtain third evaluation data; obtaining the weight of the third evaluation data according to the obtained third evaluation data; according to the calculated weight, the data score of the satisfaction degree of the aviation enterprise is output by adopting a weighted average method.

Description

Method and system for analyzing satisfaction evaluation quality of aviation enterprise based on Lekter scale
Technical Field
The invention relates to the technical field of evaluation quality detection methods, in particular to a method and a system for analyzing the satisfaction evaluation quality of an aviation enterprise based on a Lekter scale.
Background
With the development of internet technology, the living standard of people is increasingly improved, along with the upgrading of travel requirements, the number of people who select airplanes for traveling is increased year by year, and the passenger satisfaction research of aviation enterprises is particularly important. With the arrival of the big data era and the continuous promotion of informatization of mobile interconnection technology, the number of available passenger resources is explosively increased, and the value and convenience brought to people by big data are obvious at present. The method has the advantages that the mining and analysis of big data are utilized to carry out the research on the service satisfaction quality of the aviation enterprise, the level of the service satisfaction of the aviation enterprise is scientifically evaluated, the evaluation of the aviation enterprise is provided for users, meanwhile, the reference is provided for the service satisfaction quality of the aviation industry, the recovery cost of the service satisfaction data of the aviation enterprise is high, the recovery efficiency is low, the recovery samples are limited, and the evaluation result cannot reflect the real service level. The reason is that at present, no accurate quality evaluation method and effective quality control mechanism are adopted for quality evaluation of the service satisfaction degree of the aviation enterprise.
Although the service quality evaluation of the aviation enterprises already has a mature theoretical framework, the application field still has many defects. For example, when passengers travel by taking flights, a plurality of service points contacted with an airline company and an airport are provided, most passengers have weak perception on part of services and pay attention to the airline enterprises, but pay attention to the other part of services and ignore the situation of the airline enterprises, and the experience of the passengers on the air travel can be reduced as a whole, so that the service quality evaluation of the airline enterprises needs a calculation method capable of really finding the service points which the passengers are interested in.
Generally, the research on the service quality evaluation of the aviation enterprises is still in the starting stage, the method is applied to digital service resources, and an automatic quality evaluation method based on a fine granularity evaluation model is yet to be researched.
Disclosure of Invention
In order to solve the technical problems in the background art, the invention provides a method for analyzing the satisfaction evaluation quality of an aviation enterprise based on a Lekter scale, which comprises the following steps:
s1: collecting aeronautical enterprise satisfaction data by using a Lekter scale to form first evaluation data, wherein the first evaluation data comprise evaluation objects, evaluation scores and evaluation items, and the first evaluation data are subjected to evaluation item screening to form second evaluation data;
s2: according to the second evaluation data obtained in the step S1, a plurality of clustering centers are selected optionally, and the second evaluation data are clustered according to the clustering centers to form a clustering domain, so that third evaluation data are obtained;
s3: obtaining a weight of the third evaluation data from the third evaluation data obtained in S2;
s4: and outputting the data score of the aviation enterprise satisfaction degree by adopting a weighted average method according to the weight obtained in the S3.
Further, the method for collecting the aircraft enterprise satisfaction data by using the Lekter scale forms first evaluation data, and specifically comprises the following steps: collecting the aircraft enterprise satisfaction data by using a Likter scale, and removing the aircraft enterprise satisfaction data containing invalid comments to serve as first evaluation data, wherein the first evaluation data comprises evaluation objects, evaluation scores of the evaluation objects and evaluation items corresponding to the evaluation objects and the evaluation scores; wherein the invalid comment comprises first evaluation data in which the highest-score or lowest-score evaluation accounts for 100% of the total number of evaluations.
Further, the screening of the evaluation items of the first evaluation data to form second evaluation data specifically includes: the first evaluation data is firstly processed by a standard matrix, then the correlation coefficient matrix is solved according to the standard matrix, the characteristic root of the correlation coefficient matrix is solved, the minimum value in the characteristic root is removed, and the accumulated contribution value alpha is enabled to bepCan be not less than a predetermined value, the cumulative contribution αpRepresents oneOr the ratio of a plurality of characteristic roots to the total sum of the characteristic roots, then removing the evaluation items corresponding to the removed characteristic roots in the first data and forming second evaluation data by using the remaining first evaluation data;
further, the performing of the standard matrix processing on the first evaluation data specifically includes: and carrying out reduced data correlation processing of variance and mean value on the first evaluation data, wherein the expression is as follows:
Figure BDA0003103616180000031
wherein i represents the ith bit of n evaluation objects, j represents the jth of m evaluation items, and xijA rating score value of the jth rating item indicating the ith rating object,
Figure BDA0003103616180000032
mean value s of evaluation scores of n evaluation objects for j evaluation itemjA standard deviation representing the evaluation score of the j-th evaluation item for the n evaluation objects;
preferably, the minimum value in the removed feature root specifically is: the characteristic root lambda obtained by solving the correlation coefficient matrixmSorted by size, referenced to a cumulative contribution apExpression, root of feature λmThe maximum value of (1) is taken into (a)pThe expression is as follows:
Figure BDA0003103616180000033
when p in the above formula takes a value of alphapWhen the value of (A) is greater than a predetermined value, the calculation is stopped and will not participate in alphapCalculated characteristic root λmDiscarding while keeping the characteristic root λmThe corresponding evaluation items in the first evaluation data are discarded to form second evaluation data.
Further, the clustering process is performed on the second evaluation data according to the clustering center to form a clustering domain, and third evaluation data is obtained, specifically:
randomly selecting k second evaluation data as clustering centers, sequentially allocating the second evaluation data which are not selected as the clustering centers to the selected k clustering centers according to Euclidean distances, and forming k clustering domains around the clustering centers after allocation is finished; summing and averaging all the p-dimensional vectors included in each of the k clustering domains, taking the averaged mean vector as a new clustering center, and calculating the change distance between the new clustering centers and the previous generation clustering center;
if the change distance between the plurality of new cluster center positions and the previous cluster center position is not more than a predetermined value, terminating the calculation and forming third evaluation data;
if the change distance between the plurality of new cluster center positions and the previous cluster center position is greater than the predetermined value, the clustering process is repeated until the change distance between the plurality of new cluster center positions and the previous cluster center position is not greater than the predetermined value.
Further, the weighting of the third evaluation data obtained in S2 includes:
calculating a conflict quantization index R of the third evaluation datajThe method specifically comprises the following steps: determining conflicting quantization indices R for multiple different cluster domains or multiple different p-dimensional vectors in a single cluster domainjConflict quantization index RjIs expressed as follows, r in the formulaijIs to evaluate the correlation coefficient between the ith evaluation index and the jth evaluation index:
Figure BDA0003103616180000041
preferably, the indicator R is quantified by the conflictjCalculating the information amount A contained in the jth third evaluation datajThe expression is as follows;
Figure BDA0003103616180000042
preferably, the information amount A is usedjCalculating the weight W of the jth third evaluation datajThe expression is as follows;
Figure BDA0003103616180000043
wherein, the clustering domain participates in the conflict quantization index RjInformation amount AjWeight WjWhen calculating, its value is taken by its clustering center cjIs a reference value.
Further, the data screening module: collecting aeronautical enterprise satisfaction data by using a Lekter scale to form first evaluation data, wherein the first evaluation data comprise evaluation objects, evaluation scores and evaluation items, and the first evaluation data are subjected to evaluation item screening to form second evaluation data;
a data clustering module: randomly selecting a plurality of clustering centers according to the second evaluation data obtained by the data screening module, and clustering the second evaluation data according to the clustering centers to form a clustering domain to obtain third evaluation data;
a weight analysis module: obtaining the weight of the third evaluation data according to the third evaluation data obtained in the clustering module;
satisfaction output module: and outputting the data score of the satisfaction degree of the aviation enterprise by adopting a weighted average method according to the weight obtained by the weight analysis module.
Further, the data processing module is specifically configured to: collecting the aircraft enterprise satisfaction data by using a Likter scale, and removing the aircraft enterprise satisfaction data containing invalid comments to serve as first evaluation data, wherein the first evaluation data comprises evaluation objects, evaluation scores of the evaluation objects and evaluation items corresponding to the evaluation objects and the evaluation scores; wherein the invalid comment comprises first evaluation data in which the evaluation of the highest score or the lowest score accounts for 100% of the total number of evaluations;
preferably, the data processing module is specifically configured to: the first evaluation data is firstly processed by a standard matrix and then the correlation coefficient of the first evaluation data is obtained according to the standard matrixSolving the characteristic root of the correlation coefficient matrix, and removing the minimum value in the characteristic root to ensure that the accumulated contribution value alphapCan be not less than a predetermined value, the cumulative contribution αpRepresenting the ratio of one or more characteristic roots to the total sum of the characteristic roots, removing the evaluation items corresponding to the removed characteristic roots in the first data, and forming second evaluation data by using the remaining first evaluation data;
preferably, the data processing module is specifically configured to: and carrying out reduced data correlation processing of variance and mean value on the first evaluation data, wherein the expression is as follows:
Figure BDA0003103616180000061
wherein i represents the ith bit of n evaluation objects, j represents the jth of m evaluation items, and xijA rating score value of the jth rating item indicating the ith rating object,
Figure BDA0003103616180000062
mean value s of evaluation scores of n evaluation objects for j evaluation itemjA standard deviation representing the evaluation score of the j-th evaluation item for the n evaluation objects;
preferably, the data processing module is specifically configured to: the characteristic root lambda obtained by solving the correlation coefficient matrixmSorted by size, referenced to a cumulative contribution apExpression, root of feature λmThe maximum value of (1) is taken into (a)pThe expression is as follows:
Figure BDA0003103616180000063
when p in the above formula takes a value of alphapWhen the value of (A) is greater than a predetermined value, the calculation is stopped and will not participate in alphapCalculated characteristic root λmDiscarding while keeping the characteristic root λmThe corresponding evaluation items in the first evaluation data are discarded to form second evaluation data.
Further, the clustering module is specifically configured to: randomly selecting k second evaluation data as clustering centers, sequentially allocating the second evaluation data which are not selected as the clustering centers to the selected k clustering centers according to Euclidean distances, and forming k clustering domains around the clustering centers after allocation is finished; summing and averaging all the p-dimensional vectors included in each of the k clustering domains, taking the averaged mean vector as a new clustering center, and calculating the change distance between the new clustering centers and the previous generation clustering center;
if the change distance between the plurality of new cluster center positions and the previous cluster center position is not more than a predetermined value, terminating the calculation and forming third evaluation data;
if the change distance between the plurality of new cluster center positions and the previous cluster center position is greater than the predetermined value, the clustering process is repeated until the change distance between the plurality of new cluster center positions and the previous cluster center position is not greater than the predetermined value.
Further, the weight analysis module is specifically configured to: determining conflicting quantization indices R for multiple different cluster domains or multiple different p-dimensional vectors in a single cluster domainjConflict quantization index RjIs expressed as follows, r in the formulaijIs to evaluate the correlation coefficient between the ith evaluation index and the jth evaluation index:
Figure BDA0003103616180000071
preferably, the weight analysis module is specifically configured to: by quantifying the index R by conflictjCalculating the information amount A contained in the jth third evaluation datajThe expression is as follows;
Figure BDA0003103616180000072
preferably, the weight analysis module is specifically configured to: passing informationQuantity AjCalculating the weight W of the jth third evaluation datajThe expression is as follows;
Figure BDA0003103616180000073
wherein, the clustering domain participates in the conflict quantization index RjInformation amount AjWeight WjWhen calculating, its value is taken by its clustering center cjIs a reference value.
According to the method for analyzing the satisfaction evaluation quality of the aviation enterprise based on the Lekter scale, the dimension reduction, clustering and weight analysis processing are carried out on the obtained passenger aviation evaluation original data through a big data processing mode, interference data can be removed, the service item of the passenger is found out accurately, and the problems of high cost, high difficulty, high subjectivity and the like of a method for manually processing the evaluation data are solved technically.
Drawings
Fig. 1 is a schematic flow structure diagram of a method for analyzing the satisfaction evaluation quality of an aviation enterprise based on a lectt scale, which is provided by the invention.
Detailed Description
The embodiment provides a method and a system for analyzing the satisfaction evaluation quality of an aviation enterprise based on a Lekter scale, and important data are screened out by performing dimensionality reduction, clustering and weight analysis on the data based on big data processing.
In order to achieve the technical effect, the method for analyzing the satisfaction evaluation quality of the aviation enterprise based on the Lekter scale comprises the following steps of:
s1: collecting aeronautical enterprise satisfaction data by using a Lekter scale to form first evaluation data, removing invalid comments in the first evaluation data, wherein the first evaluation data comprise evaluation objects, evaluation scores and evaluation items, and screening the evaluation items of the first evaluation data to form second evaluation data; the method specifically comprises the following steps:
s101: flight evaluation original data of passengers are collected, and under the development of current big data and the Internet, besides the traditional opinion form collection, the flight evaluation original data of the passengers can be collected through different APPs; for objectivity and accuracy of data, in this embodiment, extreme data with all options in flight evaluation raw data being highest scores or lowest scores may be removed, and it is noted that a standard for data removal may be adjusted according to selection;
s102: forming first evaluation data on the basis of flight evaluation original data with the pole data removed;
classifying and screening the first evaluation data according to the service item categories provided by the aviation enterprises, and storing the obtained first evaluation data according to a triple structure of < evaluation object, evaluation item, evaluation score >;
in an embodiment of the present invention, the python open source framework pandas is used to convert the first rating data to be stored in a structure of < rating object, rating item, rating score > triple.
S103: performing standard matrix processing on the first evaluation data;
in order to find out the service items provided by the airlines which the passengers really intend, the first evaluation data needs to be screened, namely, the total number of types of the evaluation items is reduced, and useless interference data is further removed compared with S101; meanwhile, in order to facilitate the arrangement of the first evaluation data, in the embodiment of the present invention, the evaluation item x is definedjFor m dimensions, i.e. representing a total of m service items provided by the airline, the expression of the evaluation item is xj＝(x1，x2，…，xm)TDefining an evaluation object xiN, i.e. representing a total of n participating passengers, and since each evaluation object is an m-dimensional vector, the evaluation object can be represented as: x is the number ofi＝(xn1，xn2，…，xnm)TAccording to xiAnd xjThe evaluation score of the jth evaluation item of the ith evaluation object can be represented as xij，(i＝1，2，…，n；j＝1，2，…，m)；
For example: the evaluation score of the a-th passenger on the b-th service item provided by the airline enterprise is 4, and x can be representedab＝4；
In order to reduce the correlation between the original evaluation score data, x needs to be setijConverted into a standardized index
Figure BDA0003103616180000091
The expression is as follows:
Figure BDA0003103616180000092
wherein,
Figure BDA0003103616180000093
is the mean of all the evaluation scores in the jth evaluation item, SjFor the standard deviation of all the evaluation scores in the jth evaluation item, the expression is as follows:
Figure BDA0003103616180000094
to be provided with
Figure BDA0003103616180000095
Constructing a sample matrix, performing digital processing on the sample matrix to obtain a standard matrix Z (as follows), and performing standard matrix processing on the original data to obtain each variable xijThe average value of the data is changed into 0, the standard deviation is changed into 1, and dimensional difference and magnitude difference among different original data can be eliminated;
Figure BDA0003103616180000101
s104: a correlation matrix R is obtained according to the standard matrix Z, so that the mutual influence among evaluation score data can be further eliminated;
Figure BDA0003103616180000102
depending on the nature of the correlation matrix, ρ can be obtainedij(see below), where ρii＝1,ρij＝ρji：
Figure BDA0003103616180000103
According to rhoijR is known as:
Figure BDA0003103616180000104
s105: solving the correlation matrix R to obtain m characteristic roots of the correlation matrix R, wherein the mth characteristic root represents the mth service item provided by the aviation enterprise, namely each characteristic root corresponds to one evaluation item; in the actual working process, as the number of service items is large, in order to reduce the workload, the total number of the service items needs to be reduced;
according to the mathematical expression, the characteristic equation of the correlation matrix is as follows: l R- λ E | ═ 0, where λ is an unknown number and E is an identity matrix; solving the characteristic equation to obtain m characteristic roots lambda related to the unknown number lambdam：
In the embodiment of the invention, the m-dimensional evaluation data needs to be reduced to p-dimensional evaluation data, namely, the total number of service item categories provided by the whole aviation enterprise is reduced, and the data accumulated contribution rate alpha is screened outpThe service item category is larger, so that the service point which is really the intention of passengers is found, and the interference of useless data is reduced, wherein the cumulative contribution rate alphapIs expressed as follows, in practical operation, will be1、λ2、...、λmSequentially arranged according to the size sequence and preferentially brings the maximum value into alphapUntil the obtained cumulative contribution rate alpha in the formula (2)pNot less than 90%, stopping calculation, and not participating in calculated lambdamThe corresponding evaluation item is subtracted from the first evaluation data, and the remaining data is used to form a second evaluation dataEvaluating the data, wherein the original data can be reduced from an m-dimensional vector to a p-dimensional vector;
Figure BDA0003103616180000111
s2: randomly selecting a plurality of clustering centers according to the second evaluation data obtained in the step S1, and clustering the second evaluation data to form a clustering domain to obtain third evaluation data; note that the aviation evaluation data at this time has been screened by S1, and the first evaluation data is converted from the m-dimension to the p-dimension; the method specifically comprises the following steps:
s201: in order to provide a reference for iterative clustering, in the embodiment of the present invention, let Zk(f +1) is a vector value of k initial samples randomly taken from the second evaluation data, and it is to be noted that the aviation evaluation data at this time is still n passengers, but the evaluation items of the n passengers are reduced from m to p, that is, the total data volume is reduced; with cjInitial clustering center: c. Cj∈{Z1(f+1)，Z2(f+1)，…，ZK(f +1) }, cluster center cjAs a p-dimensional vector, i.e. the cluster centre cjRepresenting a multi-dimensional set comprising a set of all the evaluation scores of a passenger for p evaluation items; due to cjIs a vector of dimension p, then Z1(f+1)＝{z11(f+1),z12(f+1),…,z1p(f +1) }, f is the number of iterative operations, and in the initial state, f is 0, and any cluster center may also be represented as Zj(1) (j ═ 1,2, …, k); to enable the final completion of the step of S2, it is necessary that there be a set C of k cluster centerskSo that the second evaluation data can be respectively clustered at a cluster center c nearest theretojThe specific function expression of the neighborhood is as follows, wherein xiIs shown in the second evaluation data except that it is sorted out as a cluster center cjData other than the data;
Figure BDA0003103616180000121
s202: at the first iteration, the data is sorted out as the cluster center cjThe extra data being assigned to one of Z by Euclidean distancej(1) (ii) a After all data are made to cluster center cjAfter the distances are all minimum, the first iteration can be completed;
s203: after the first iteration is completed, a cluster domain { S ] formed by the first iteration can be obtained1(f)，S2(f)，…，SK(f) Each cluster domain contains Nj(f) Since f is 1, any cluster field can be represented as Sj(1)，(Sj(1)∈{S1(1)，S2(1)，…，SK(1)})；
S204: clustering domain Sj(1) N contained in (1)j(f) The mean vector of each sample, and the obtained mean vector is used as a new clustering center, the new clustering center can be expressed as: { Z1(2)，Z2(2)，…，ZK(2) }; and repeating iteration and operation until the moving distance between the new cluster centers and the previous cluster center is less than a preset value, and stopping the operation to form third evaluation data.
S3: for the third evaluation data obtained in S2, the weight of the third evaluation data is obtained, specifically:
s301: in order to judge the data similarity among a plurality of different third evaluation data, the measurement needs to be carried out through a conflict quantization index;
in an embodiment of the invention, the third evaluation data comprises two levels of data, each referring to a different cluster domain SK(f) Between and same cluster domain SK(f) The specific operation process among the different second evaluation data is as follows:
j different cluster domain SK(f) With other cluster domains, the jth cluster domain SK(f) The conflict quantization index between different second evaluation data in (1) is RjThe conflict quantization index is RjThe expression of (a) is as follows:
Figure BDA0003103616180000131
wherein r isijEvaluating a correlation coefficient between the ith third evaluation data and the jth third evaluation data, wherein k in the formula refers to k clustering domains or k second evaluation data in the same clustering domain;
Figure BDA0003103616180000132
the objective weight of the aviation evaluation data is comprehensively measured by contrast strength and conflict, and is AjIndicating the amount of information contained in the jth third evaluation data, AjCan be expressed as:
Figure BDA0003103616180000133
Ajthe larger the information content contained in the jth third evaluation data is, the greater the relative importance of the index is; the objective weight of the jth third evaluation data is Wj：
Figure BDA0003103616180000134
S4: calculating the satisfaction score of the third evaluation data by using the objective weight of the third evaluation data obtained in S3 and using a weighted average method
Figure BDA0003103616180000135
The embodiment of the application also provides a system for analyzing the satisfaction evaluation quality of an aviation enterprise based on the Lekter scale, which comprises the following modules:
the data screening module: collecting aeronautical enterprise satisfaction data by using a Lekter scale to form first evaluation data, wherein the first evaluation data comprise evaluation objects, evaluation scores and evaluation items, and the first evaluation data are subjected to evaluation item screening to form second evaluation data;
a data clustering module: randomly selecting a plurality of clustering centers according to the second evaluation data obtained by the data screening module, and clustering the second evaluation data according to the clustering centers to form a clustering domain to obtain third evaluation data;
a weight analysis module: obtaining the weight of the third evaluation data according to the third evaluation data obtained in the clustering module;
satisfaction output module: and outputting the data score of the satisfaction degree of the aviation enterprise by adopting a weighted average method according to the weight obtained by the weight analysis module.
The data processing module is specifically configured to: collecting the aircraft enterprise satisfaction data by using a Likter scale, and removing the aircraft enterprise satisfaction data containing invalid comments to serve as first evaluation data, wherein the first evaluation data comprises evaluation objects, evaluation scores of the evaluation objects and evaluation items corresponding to the evaluation objects and the evaluation scores; the invalid comments comprise first evaluation data with the evaluation of the highest score or the lowest score accounting for 100% of the total evaluation number;
preferably, the data processing module is specifically configured to: the first evaluation data is firstly processed by a standard matrix, then the correlation coefficient matrix is solved according to the standard matrix, the characteristic root of the correlation coefficient matrix is solved, the minimum value in the characteristic root is removed, and the accumulated contribution value alpha is enabled to bepCan be not less than a predetermined value, the contribution value alpha is accumulatedpRepresenting the ratio of one or more characteristic roots to the total sum of the characteristic roots, removing the evaluation items corresponding to the removed characteristic roots in the first data, and forming second evaluation data by using the remaining first evaluation data;
preferably, the data processing module is specifically configured to: and carrying out reduced data correlation processing of variance and mean value on the first evaluation data, wherein the expression is as follows:
Figure BDA0003103616180000141
in the formula, i represents n scoresThe ith position in the price object, j represents the jth of m evaluation items, xijA rating score value of the jth rating item indicating the ith rating object,
Figure BDA0003103616180000151
mean value s of evaluation scores of n evaluation objects for j evaluation itemjA standard deviation representing the evaluation score of the j-th evaluation item for the n evaluation objects;
preferably, the data processing module is specifically configured to: the characteristic root lambda obtained by solving the correlation coefficient matrixmSorted by size, referenced to a cumulative contribution apExpression, root of feature λmThe maximum value of (1) is taken into (a)pThe expression is as follows:
Figure BDA0003103616180000152
when p in the above formula takes a value of alphapWhen the value of (A) is greater than a predetermined value, the calculation is stopped and will not participate in alphapCalculated characteristic root λmDiscarding while keeping the characteristic root λmThe corresponding evaluation items in the first evaluation data are discarded to form second evaluation data.
The clustering module is specifically configured to: randomly selecting k second evaluation data as clustering centers, sequentially allocating the second evaluation data which are not selected as the clustering centers to the selected k clustering centers according to Euclidean distances, and forming k clustering domains around the clustering centers after allocation is finished; summing and averaging all the p-dimensional vectors included in each of the k clustering domains, taking the averaged mean vector as a new clustering center, and calculating the change distance between the new clustering centers and the previous generation clustering center;
if the change distance between the plurality of new cluster center positions and the previous cluster center position is not more than a predetermined value, terminating the calculation and forming third evaluation data;
if the change distance between the plurality of new cluster center positions and the previous cluster center position is greater than the predetermined value, the clustering process is repeated until the change distance between the plurality of new cluster center positions and the previous cluster center position is not greater than the predetermined value.
The weight analysis module is specifically configured to: determining conflicting quantization indices R for multiple different cluster domains or multiple different p-dimensional vectors in a single cluster domainjConflict quantization index RjIs expressed as follows, r in the formulaijIs to evaluate the correlation coefficient between the ith evaluation index and the jth evaluation index:
Figure BDA0003103616180000161
preferably, the weight analysis module is specifically configured to: by quantifying the index R by conflictjCalculating the information amount A contained in the jth third evaluation datajThe expression is as follows;
Figure BDA0003103616180000162
preferably, the weight analysis module is specifically configured to: by the information amount AjCalculating the weight W of the jth third evaluation datajThe expression is as follows;
Figure BDA0003103616180000163
wherein, the clustering domain participates in the conflict quantization index RjInformation amount AjWeight WjWhen calculating, its value is taken by its clustering center cjIs a reference value.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be able to cover the technical scope of the present invention and the equivalent alternatives or modifications according to the technical solution and the inventive concept of the present invention within the technical scope of the present invention.

Claims (10)

1. A method for analyzing the satisfaction degree evaluation quality of an aviation enterprise based on a Likter scale is characterized by comprising the following steps:
s1: collecting aeronautical enterprise satisfaction data by using a Lekter scale to form first evaluation data, wherein the first evaluation data comprise evaluation objects, evaluation scores and evaluation items, and the first evaluation data are subjected to evaluation item screening to form second evaluation data;
s2: according to the second evaluation data obtained in the step S1, a plurality of clustering centers are selected optionally, and the second evaluation data are clustered according to the clustering centers to form a clustering domain, so that third evaluation data are obtained;
s3: obtaining a weight of the third evaluation data from the third evaluation data obtained in S2;
s4: and outputting the data score of the aviation enterprise satisfaction degree by adopting a weighted average method according to the weight obtained in the S3.
2. The method for analyzing airline satisfaction rating quality based on the litterb scale of claim 1, wherein:
the method for collecting the satisfaction data of the aviation enterprises by using the Lekter scale forms first evaluation data, and specifically comprises the following steps: collecting the aircraft enterprise satisfaction data by using a Likter scale, and removing the aircraft enterprise satisfaction data containing invalid comments to serve as first evaluation data, wherein the first evaluation data comprises evaluation objects, evaluation scores of the evaluation objects and evaluation items corresponding to the evaluation objects and the evaluation scores; wherein the invalid comment comprises first evaluation data in which the highest-score or lowest-score evaluation accounts for 100% of the total number of evaluations.
3. The method for analyzing airline satisfaction rating quality based on the litterb scale of claim 1, wherein:
the screening of the evaluation items of the first evaluation data to form second evaluation data specifically comprises the following steps: the first evaluation data is processed by a standard matrix, and then the correlation is calculated according to the standard matrixCounting a matrix, solving a characteristic root of the correlation coefficient matrix, and removing a minimum value in the characteristic root to ensure that the contribution value alpha is accumulatedpCan be not less than a predetermined value, the cumulative contribution αpAnd the evaluation items corresponding to the removed characteristic roots in the first data are removed, and second evaluation data is formed by the remaining first evaluation data.
4. The method for analyzing airline satisfaction rating quality based on the litterb scale of claim 1, wherein: the standard matrix processing is performed on the first evaluation data, and specifically includes: and carrying out reduced data correlation processing of variance and mean value on the first evaluation data, wherein the expression is as follows:
Figure FDA0003103616170000021
wherein i represents the ith bit of n evaluation objects, j represents the jth of m evaluation items, and xijA rating score value of the jth rating item indicating the ith rating object,
Figure FDA0003103616170000023
mean value s of evaluation scores of n evaluation objects for j evaluation itemjA standard deviation representing the evaluation score of the j-th evaluation item for the n evaluation objects;
preferably, the minimum value in the removed feature root specifically is: the characteristic root lambda obtained by solving the correlation coefficient matrixmSorted by size, referenced to a cumulative contribution apExpression, root of feature λmThe maximum value of (1) is taken into (a)pThe expression is as follows:
Figure FDA0003103616170000022
when p in the above formula takes a value of alphapWhen the value of (A) is greater than a predetermined value, the calculation is stopped and will not participate in alphapCalculated characteristic root λmDiscarding while keeping the characteristic root λmThe corresponding evaluation items in the first evaluation data are discarded to form second evaluation data.
5. The method for analyzing the satisfaction evaluation quality of the aviation enterprise based on the litters scale as claimed in claim 1 or 4, wherein the second evaluation data is clustered according to a clustering center to form a clustering domain, and third evaluation data is obtained, specifically:
randomly selecting k second evaluation data as clustering centers, sequentially allocating the second evaluation data which are not selected as the clustering centers to the selected k clustering centers according to Euclidean distances, and forming k clustering domains around the clustering centers after allocation is finished; summing and averaging all the p-dimensional vectors included in each of the k clustering domains, taking the averaged mean vector as a new clustering center, and calculating the change distance between the new clustering centers and the previous generation clustering center;
if the change distance between the plurality of new cluster center positions and the previous cluster center position is not more than a predetermined value, terminating the calculation and forming third evaluation data;
if the change distance between the plurality of new cluster center positions and the previous cluster center position is greater than the predetermined value, the clustering process is repeated until the change distance between the plurality of new cluster center positions and the previous cluster center position is not greater than the predetermined value.
6. The method for analyzing the quality of an airline satisfaction rating based on the Lekter scale according to claim 1 or 5, characterized in that,
the weighting of the third evaluation data obtained in S2 includes:
calculating a conflict quantization index R of the third evaluation datajThe method specifically comprises the following steps: determining conflicting quantization indices for multiple different cluster domains or multiple different p-dimensional vectors in a single cluster domainMark RjConflict quantization index RjIs expressed as follows, r in the formulaijIs to evaluate the correlation coefficient between the ith evaluation index and the jth evaluation index:
Figure FDA0003103616170000031
preferably, the indicator R is quantified by the conflictjCalculating the information amount A contained in the jth third evaluation datajThe expression is as follows;
Figure FDA0003103616170000041
preferably, the information amount A is usedjCalculating the weight W of the jth third evaluation datajThe expression is as follows;
Figure FDA0003103616170000042
wherein, the clustering domain participates in the conflict quantization index RjInformation amount AjWeight WjWhen calculating, its value is taken by its clustering center cjIs a reference value.
7. A system for analyzing the satisfaction evaluation quality of an aviation enterprise based on a Lekter scale is characterized by comprising the following modules:
the data screening module: collecting aeronautical enterprise satisfaction data by using a Lekter scale to form first evaluation data, wherein the first evaluation data comprise evaluation objects, evaluation scores and evaluation items, and the first evaluation data are subjected to evaluation item screening to form second evaluation data;
a data clustering module: randomly selecting a plurality of clustering centers according to the second evaluation data obtained by the data screening module, and clustering the second evaluation data according to the clustering centers to form a clustering domain to obtain third evaluation data;
a weight analysis module: obtaining the weight of the third evaluation data according to the third evaluation data obtained in the clustering module;
satisfaction output module: and outputting the data score of the satisfaction degree of the aviation enterprise by adopting a weighted average method according to the weight obtained by the weight analysis module.
8. The system for analyzing airline satisfaction rating quality based on the litterb scale of claim 7, wherein:
the data processing module is specifically configured to: collecting the aircraft enterprise satisfaction data by using a Likter scale, and removing the aircraft enterprise satisfaction data containing invalid comments to serve as first evaluation data, wherein the first evaluation data comprises evaluation objects, evaluation scores of the evaluation objects and evaluation items corresponding to the evaluation objects and the evaluation scores; wherein the invalid comment comprises first evaluation data in which the evaluation of the highest score or the lowest score accounts for 100% of the total number of evaluations;
preferably, the data processing module is specifically configured to: the first evaluation data is firstly processed by a standard matrix, then the correlation coefficient matrix is solved according to the standard matrix, the characteristic root of the correlation coefficient matrix is solved, the minimum value in the characteristic root is removed, and the accumulated contribution value alpha is enabled to bepCan be not less than a predetermined value, the cumulative contribution αpRepresenting the ratio of one or more characteristic roots to the total sum of the characteristic roots, removing the evaluation items corresponding to the removed characteristic roots in the first data, and forming second evaluation data by using the remaining first evaluation data;
preferably, the data processing module is specifically configured to: and carrying out reduced data correlation processing of variance and mean value on the first evaluation data, wherein the expression is as follows:
Figure FDA0003103616170000051
wherein i represents the ith bit of n evaluation objects, j represents the jth of m evaluation items, and xijIs shown asThe evaluation score value of the j-th evaluation item of the i evaluation objects,
Figure FDA0003103616170000052
mean value s of evaluation scores of n evaluation objects for j evaluation itemjA standard deviation representing the evaluation score of the j-th evaluation item for the n evaluation objects;
preferably, the data processing module is specifically configured to: the characteristic root lambda obtained by solving the correlation coefficient matrixmSorted by size, referenced to a cumulative contribution apExpression, root of feature λmThe maximum value of (1) is taken into (a)pThe expression is as follows:
Figure FDA0003103616170000061
when p in the above formula takes a value of alphapWhen the value of (A) is greater than a predetermined value, the calculation is stopped and will not participate in alphapCalculated characteristic root λmDiscarding while keeping the characteristic root λmThe corresponding evaluation items in the first evaluation data are discarded to form second evaluation data.
9. The system for analyzing the quality of an airline satisfaction rating based on the Lekter scale according to claim 7 or 8, wherein,
the clustering module is specifically configured to: randomly selecting k second evaluation data as clustering centers, sequentially allocating the second evaluation data which are not selected as the clustering centers to the selected k clustering centers according to Euclidean distances, and forming k clustering domains around the clustering centers after allocation is finished; summing and averaging all the p-dimensional vectors included in each of the k clustering domains, taking the averaged mean vector as a new clustering center, and calculating the change distance between the new clustering centers and the previous generation clustering center;
if the change distance between the plurality of new cluster center positions and the previous cluster center position is not more than a predetermined value, terminating the calculation and forming third evaluation data;
if the change distance between the plurality of new cluster center positions and the previous cluster center position is greater than the predetermined value, the clustering process is repeated until the change distance between the plurality of new cluster center positions and the previous cluster center position is not greater than the predetermined value.
10. The system for analyzing the quality of an airline satisfaction rating based on the Lekter scale according to claim 7 or 9, characterized in that,
the weight analysis module is specifically configured to: determining conflicting quantization indices R for multiple different cluster domains or multiple different p-dimensional vectors in a single cluster domainjConflict quantization index RjIs expressed as follows, r in the formulaijIs to evaluate the correlation coefficient between the ith evaluation index and the jth evaluation index:
Figure FDA0003103616170000071
preferably, the weight analysis module is specifically configured to: by quantifying the index R by conflictjCalculating the information amount A contained in the jth third evaluation datajThe expression is as follows;
Figure FDA0003103616170000072
preferably, the weight analysis module is specifically configured to: by the information amount AjCalculating the weight W of the jth third evaluation datajThe expression is as follows;
Figure FDA0003103616170000073
wherein, the clustering domain participates in the conflict quantization index RjInformation amount AjWeight WjWhen calculating, it is takenClustering center c of value withjIs a reference value.
CN202110630665.0A 2021-06-07 2021-06-07 Method and system for analyzing satisfaction evaluation quality of aviation enterprise based on Lekter scale Pending CN113569902A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110630665.0A CN113569902A (en) 2021-06-07 2021-06-07 Method and system for analyzing satisfaction evaluation quality of aviation enterprise based on Lekter scale

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110630665.0A CN113569902A (en) 2021-06-07 2021-06-07 Method and system for analyzing satisfaction evaluation quality of aviation enterprise based on Lekter scale

Publications (1)

Publication Number Publication Date
CN113569902A true CN113569902A (en) 2021-10-29

Family

ID=78161053

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110630665.0A Pending CN113569902A (en) 2021-06-07 2021-06-07 Method and system for analyzing satisfaction evaluation quality of aviation enterprise based on Lekter scale

Country Status (1)

Country Link
CN (1) CN113569902A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114595244A (en) * 2022-03-11 2022-06-07 北京字节跳动网络技术有限公司 Collapse data aggregation method and device, electronic equipment and storage medium

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114595244A (en) * 2022-03-11 2022-06-07 北京字节跳动网络技术有限公司 Collapse data aggregation method and device, electronic equipment and storage medium
CN114595244B (en) * 2022-03-11 2023-10-17 抖音视界有限公司 Method and device for aggregating crash data, electronic equipment and storage medium

Similar Documents

Publication Publication Date Title
CN103617435B (en) Image sorting method and system for active learning
CN110503245B (en) Prediction method for large-area delay risk of airport flight
CN112700325A (en) Method for predicting online credit return customers based on Stacking ensemble learning
CN110928764A (en) Automated mobile application crowdsourcing test report evaluation method and computer storage medium
CN105069470A (en) Classification model training method and device
CN108038627B (en) Object evaluation method and device
CN111079941B (en) Credit information processing method, credit information processing system, terminal and storage medium
CN111160696A (en) Big data based detected person grading method
CN114764682B (en) Rice safety risk assessment method based on multi-machine learning algorithm fusion
WO2017071369A1 (en) Method and device for predicting user unsubscription
CN114429245A (en) Analysis display method of engineering cost data
Nurdina et al. Naive Bayes and KNN for Airline Passenger Satisfaction Classification: Comparative Analysis
CN113569902A (en) Method and system for analyzing satisfaction evaluation quality of aviation enterprise based on Lekter scale
CN113837266B (en) Software defect prediction method based on feature extraction and Stacking ensemble learning
CN111341096B (en) Bus running state evaluation method based on GPS data
CN116128160B (en) Method, system, equipment and medium for predicting peak passenger flow of railway station
CN112241820A (en) Risk identification method and device for key nodes in fund flow and computing equipment
CN113297336B (en) Seasonal classification method, device and machine-readable medium for flights
CN115374858A (en) Intelligent diagnosis method for process industrial production quality based on hybrid integration model
CN111654853B (en) Data analysis method based on user information
CN114936204A (en) Feature screening method and device, storage medium and electronic equipment
CN108986554B (en) Airspace sector crowding degree dynamic identification method based on fuzzy comprehensive judgment
CN114445141A (en) Customer demand obtaining method
CN108255949B (en) Method for subdividing and selecting behavior preference of exhibition park when tourists visit exhibition
Sun Multiple criteria decision analysis techniques in aircraft design and evaluation processes

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