CN111340376A - Subway transfer perception service quality assessment method - Google Patents

Subway transfer perception service quality assessment method Download PDF

Info

Publication number
CN111340376A
CN111340376A CN202010128773.3A CN202010128773A CN111340376A CN 111340376 A CN111340376 A CN 111340376A CN 202010128773 A CN202010128773 A CN 202010128773A CN 111340376 A CN111340376 A CN 111340376A
Authority
CN
China
Prior art keywords
transfer
node
service quality
value
network
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
CN202010128773.3A
Other languages
Chinese (zh)
Inventor
华炜欣
冯雪松
张路凯
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.)
Beijing Jiaotong University
Original Assignee
Beijing Jiaotong 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 Beijing Jiaotong University filed Critical Beijing Jiaotong University
Priority to CN202010128773.3A priority Critical patent/CN111340376A/en
Publication of CN111340376A publication Critical patent/CN111340376A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • 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
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/067—Enterprise or organisation modelling
    • 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
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/40—Business processes related to the transportation industry

Landscapes

  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Engineering & Computer Science (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Educational Administration (AREA)
  • Marketing (AREA)
  • Development Economics (AREA)
  • General Business, Economics & Management (AREA)
  • Tourism & Hospitality (AREA)
  • Physics & Mathematics (AREA)
  • Quality & Reliability (AREA)
  • Operations Research (AREA)
  • Game Theory and Decision Science (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a subway transfer perception service quality assessment method, which comprises the following steps: acquiring subway transfer perception service quality data; constructing a subway transfer perception service quality evaluation model based on the subway transfer perception service quality data; based on the subway transfer perception service quality assessment model, the effect of each influence factor on the subway transfer perception service quality assessment is analyzed through a scene, and therefore the best strategy for improving the subway transfer perception service quality is determined. The invention firstly proposes to analyze the influence factors of the subway transfer perception service quality from the perspective of combining the social psychology three-dimensional attribution theory with the characteristics of the actual subway transfer system, realize the evaluation of the subway transfer perception service quality through a probability map model Bayesian network model, learn the Bayesian network topological structure through an improved PC algorithm and quantify the effect of each influence factor on the evaluation of the subway transfer perception service quality, so as to explore the optimal strategy for improving the subway transfer perception service quality.

Description

Subway transfer perception service quality assessment method
Technical Field
The invention relates to the technical field of rail transit, in particular to a subway transfer perception service quality assessment method.
Background
The study of scholars at home and abroad on the sensing service quality of subway transfer starts earlier and the realization methods are various. From the view of implementation process, most of the current researches obtain data of various objective indexes through an actual statistical mode or obtain data of various subjective indexes of passengers for subway transfer evaluation through a questionnaire survey inquiry mode, and then obtain an evaluation result of subway transfer perception service quality through various evaluation methods, such as a non-ensemble behavior selection model, a fuzzy comprehensive evaluation method, a structural equation model, an importance-satisfaction model and the like. But the existing research on the sensing service quality of subway transfer lacks the consideration of the sensing time of subway transfer. In particular, the research on the perception service quality of subway transfer is mostly based on the macroscopic evaluation of passengers on subway transfer, such as the comfort and safety of subway transfer, and the consideration of passenger transfer perception time is lacked. The existing research considering the perception time of the performers adopts a single analysis method, and most of the existing research considers the influence factors through a regression analysis method. Existing studies lack a description of uncertainty in perception time and a consideration of internal interactions of its influencing factors. Therefore, the research and application effect of the current traveler perception time is poor, and most of the research and application effects are characterized in that on the premise of giving a basic model form, the collected data is used for calibrating the model so as to analyze the action effect of the influence factors, but the optimal values of the influence factors are difficult to determine.
Disclosure of Invention
The embodiment of the invention provides a subway transfer perception service quality evaluation method, which overcomes the defects of the prior art.
In order to achieve the purpose, the invention adopts the following technical scheme.
A subway transfer perception service quality assessment method comprises the following steps:
s1, acquiring subway transfer perception service quality data;
s2, constructing a subway transfer perception service quality evaluation model based on the subway transfer perception service quality data;
s3, based on the subway transfer perception service quality assessment model, analyzing the effect of each influence factor on subway transfer perception service quality assessment through a scene, and accordingly determining the best strategy for improving the subway transfer perception service quality.
Preferably, the S1 includes the steps of:
s11, classifying the influence factors of the subway transfer perception service quality, including: the method comprises the following steps of personal attributes and trip characteristics of transfer passengers, basic evaluation factors of transfer and transfer environment factors, wherein the basic evaluation factors of transfer are as follows: actual transfer travel time, perceived transfer travel time, actual transfer travel distance, perceived transfer travel distance, actual transfer wait time, perceived transfer wait time, the transfer environmental factors including: physical and status factors;
and S12, obtaining personal attributes and travel characteristics of the transfer passengers and perception data of the transfer activities through questionnaires, and obtaining actual data of the transfer activities through field verification.
Preferably, the S12 includes the steps of:
s121, questionnaire survey: the method comprises the following steps of adopting a post investigation method, carrying out investigation aiming at the latest transfer activity of a transfer passenger through questionnaire design, wherein the first part of the questionnaire is personal basic information of the passenger and comprises personal attribute variables of the passenger; the second part is travel information of the passengers, and the travel information comprises travel characteristic variables of the passengers and specific transfer paths for transfer behaviors of the passengers; the third part is the perception and evaluation of passengers on the transfer experience of the passengers, and comprises the total time of the passengers, the perceived running time, the perceived running distance and the perceived waiting time, the evaluation of transfer environment factors and the number of trains which are actually waited by the passengers; further, on the questionnaire question option setting, with respect to the relevant questions of time and distance, the option is set to the section value; regarding the evaluation of the passenger on the transfer environment factors, the option is set to adopt a five-point scale;
s122, field verification: according to the transfer path answered by the interviewee and the corresponding transfer date and time period, carrying out experimental reduction on the transfer process of the interviewee, and recording the actual traveling time length of transfer, the actual traveling distance of transfer, the number of stairs in the transfer traveling process and the departure interval of the transfer line in the transfer process; in addition, the actual waiting time for passenger transfer can be approximately estimated by the departure interval of the transfer route and the number of trains that the passenger waits, see formula (1),
Figure BDA0002395223070000031
wherein, twTo transfer an approximate estimate of the actual latency, the unit: the method comprises the following steps of (1) taking minutes; f. ofhrFor the departure interval of the line to be switched in, the unit: minutes/row; n is a radical ofwNumber of trains waiting for passengers, unit: and (4) columns.
Preferably, the S2 includes the steps of:
s21, structure and parameter learning:
a bayesian network is defined as G ═ U, L, where the set of points in the network is U ═ X1，X2...XnIs divided into target variables XCAnd attribute variable XB＝U/XC(ii) a The set of edges between nodes is L and is equal to node XiThe set of directly connected edges is denoted Li(ii) a The subway transfer perception service quality is a target variable of the model, and the influence factors of the subway transfer perception service quality are attribute variables of the model;
s211, calculating mutual information I (X) of any two nodesi,Xj) Where I ≠ j, judge I (X)i,Xj)＞σ1Whether or not, preset sigma1Value of where σ1Not less than 0; if yes, the node X is connectediAnd XjAdd undirected edges in between to the set of edges L, otherwise do not, wherein the expression of mutual information is shown in equation (3), and xiAs discrete random variables XiCorresponding value, xjAs discrete random variables XjThe corresponding value;
Figure BDA0002395223070000032
s212, determining the maximum mutual information of each node in the network, corresponding to L \ LCEach undirected edge in (a) is judged as follows: suppose the endpoints of the undirected edge are X respectivelyqAnd XsWhere q ≠ s, as node XqMaximum mutual information M (X)q) And node XsMaximum mutual information M (X)s) As a reference, I (X) is judgedq,Xs)＞αM(Xq) (0 < α < 1) and I (X)s,Xq)＞αM(Xs) Whether or not it is satisfied, if at least one condition is satisfied, node XqAnd XsThe undirected edge between the two is reserved, otherwise, the edge is deleted from the edge set L to obtain an initial undirected graph G';
s213, after the initial undirected graph G 'is used for replacing a complete undirected graph in the PC algorithm, the PC algorithm is called to delete and initially orient the edges in the network, and G' is obtained;
s214, judging the isolated points and the undirected edges of the G': if the isolated point exists, go to step S215; if no isolated point exists but no undirected edge exists, go to step S216; if no isolated point and no directional edge exist, go to step S217;
s215, removing the isolated points in the primary directional network G', namely deleting random variables corresponding to the isolated points, and returning to the step S212;
s216, obtaining all possible directed acyclic Bayesian network structure set DAGs formed by the undirected edges according to the undirected edges in the preliminary directed network G ″SUsing BIC scoring function to pair DAGSScoring all the network structures in the network structure, and selecting the network with the optimal score as a final oriented network G', wherein the expression of a BIC scoring function is shown in a formula (4),
Figure BDA0002395223070000041
wherein q isiIs node XiParent node set PaiNumber of variable value combinations of (1), miIs node XiNumber of values of (A), NijkIs node XiParent node set PaiWhen the j variable is combined, the node XiIs the number of samples of the kth possible value, and
Figure BDA0002395223070000042
n is the total number of samples;
s217, determining any two adjacent nodes XiAnd XjCutting set V ofij；
S218, in the given cutting set VijOn the premise of (1), judging the condition mutual information I (X)i,Xj|Vij)＞σ2Whether or not, preset sigma2Value of where σ2Not less than 0; if yes, reserving adjacent node XiAnd XjOtherwise, deleting the edge to obtain the final network structure G, wherein the expression of the conditional mutual information is shown in formula (5), and v isijTo cut and collect VijThe value of the combination of the intermediate variables,
Figure BDA0002395223070000051
learning parameters in the final network structure G by adopting a maximum likelihood estimation algorithm based on the obtained final network structure G;
s22, probabilistic reasoning:
on the basis of the Bayesian network obtained by learning, the subway transfer perception service quality is taken as a target variable, the other variables in the network are taken as attribute variables, and the target variable category with the maximum posterior probability is obtained by inference through inputting the value of the attribute variables as evidence, so that the evaluation result of the subway transfer perception service quality is obtained;
s23, model evaluation:
based on a probabilistic reasoning result, introducing an overall index overall prediction accuracy and a Kappa coefficient to analyze and evaluate the overall performance of the model, respectively see formula (8) and formula (9), wherein the overall prediction accuracy represents all sample proportions with correct classification, the Kappa coefficient is used for analyzing the consistency of an actual observed value and a predicted value, the value of the index Kappa coefficient is between-1.00 and 1.00, when the index value is greater than 0.60, the consistency is good, and when the index value is greater than 0.80, the index value is nearly completely consistent;
the accuracy of the category index, the recall degree and the prediction capability of an F measured value analysis model to different categories of target variables are respectively shown in a formula (10), a formula (11) and a formula (12), the values of the accuracy and the recall degree are all between 0.00 and 1.00, the index value is closer to 1.00, the prediction effect is better, and the F measured value is a weighted harmonic value of the two;
Figure BDA0002395223070000052
Figure BDA0002395223070000053
Figure BDA0002395223070000061
Figure BDA0002395223070000062
Figure BDA0002395223070000063
wherein, PoOverall prediction accuracy is an indicator, N' is the number of samples for which the model predicts accuracy, NoIs the total amount of the sample, nj sActual number of samples, n, for the jth class of target variablesj yThe number of predicted samples for the jth class of target variables; n isj' predicting the correct number of samples for the jth class of target variables, JQj、ZHj、FjAccuracy, recall, and F measurement for the jth category of target variables, respectively.
Preferably, the S213 includes:
s2131, based on G', for any two adjacent nodes X in the networkiAnd XjBy node XiAnd XjIs used as a condition set C, the condition independence test is carried out, the order of the condition set is increased in an ascending manner from the zero order, if the condition set C can be foundijSo that at a given CijOn the premise of，XiAnd XjCondition independent, remove node XiAnd XjWhen all nodes in the network complete condition independence test, the side deletion work is completed;
s2132, based on the network obtained in the step S2131, if three nodes X can be found in the undirected graphi,Xj,XkSatisfies node XiAnd XkThere is a non-directional edge between, node XjAnd XkThere is a non-directional edge between, node XiAnd XjThere is no non-directional edge therebetween, at the same time
Figure BDA0002395223070000064
The V structure can be determined: xi→Xk←Xj；
S2133, based on the network obtained in the step S2132, the remaining undirected edges in the network are oriented by a forced orientation principle to obtain G ", wherein the forced orientation principle comprises 2 points: 1. for the triangular shape in the network, if the directions of two sides are known, the direction of the third side is determined according to the principle of no ring formation; 2. any three nodes X in the networki,Xj,XkIf X isiIs XjParent node of, XjIs XkAdjacent nodes of, and XiIs not XkAdjacent node of (2), then order XkIs XjThe parent node of (2).
Preferably, the S217 includes: :
for two adjacent nodes XiAnd XjIn a term of Xj→XiThen cutting set VijIs node XiFather node set PaiAnd is given XiFather node set
Figure BDA0002395223070000071
Determining a cut set VijThe method comprises finding a connection node XiAnd XjAnd passes through Xi j(j＝1,2,...,fi) And put it into the set FPijThen repeating the following 2 stepsTo the set FPijAll non-collision paths in (a) are deleted, including:
s2171 and mixing PaiMiddle node XiAnd XjPut into VijAnd the non-collision path passing through the part of nodes is extracted from the set FPijDeleting;
s2172 and mixing PaiIn which the node blocking the most non-collision path is placed in Vij。
Preferably, the S22 specifically includes:
assume that a Bayesian network contains n variables, divided into attribute variables XB＝{XB1,XB2,...,XBn-1And the target variable XCEach attribute variable XBi∈XBIs r isiOne possible value, the target variable XCWith t possible values, each sample in the sample set D can be defined as { x }B1,xB2,...,xBn-1,xC∈ D, wherein
Figure BDA0002395223070000072
xC∈{xC1,xC2,...,xCtTherefore, the predicted value of the target variable given the network structure G
Figure BDA0002395223070000073
Calculated by formula (6) and formula (7),
Figure BDA0002395223070000074
Figure BDA0002395223070000075
wherein x isBiAs attribute variable XBiValue of (a), xCIs a target variable XCValue of (a) ("nCIs a target variable XCIs a variable value set of the parent node set, piBiAs attribute variable XBiA set of variable values for the parent node set.
Preferably, the S3 includes:
calculating a reference mean value E (HGS) of the target variable by taking the posterior marginal probability P (HGS (j | e ═ phi)) of different types of the target variable as a reference, wherein the calculation formula is shown in a formula (13);
aiming at each attribute variable, on the premise of fixing the values of the other variables, the value of the variable is changed, namely the attribute variable takes a different value Z as ZiThe posterior probabilities P of different classes of target variables are retrieved by probabilistic reasoning (HGS j e Z)i}) and calculate the new mean value of the target variable E (HGS)iThe calculation formula is shown in formula (14);
relative amplitude of change E in the mean value of the target variableiThe effect of different values of the attribute variables is measured, the calculation formula is shown in a formula (15), the single attribute variable value for improving the subway transfer perception service quality can be obtained through scene analysis of the single attribute variable, and then the optimal strategy for improving the subway transfer perception service quality can be explored through combination of a plurality of variable values;
Figure BDA0002395223070000081
Figure BDA0002395223070000082
Figure BDA0002395223070000083
HGS is the subway transfer perception service quality of the target variable, E (HGS) is the reference mean value of the target variable, E (HGS)iDifferent values of Z-Z for a given attribute variableiAs evidence, a new mean of the target variable.
According to the technical scheme provided by the embodiment of the invention, the subway transfer perception service quality evaluation method is mainly characterized in that the subway transfer perception time is used as a basic variable, and the composition of the influence factors and the effect of the influence factors are analyzed through a Bayesian network model, so that a specific strategy for reducing the subway transfer perception time, namely improving the subway transfer perception service quality is provided. The method firstly combines the social psychology three-dimensional attribution theory and the characteristics of the subway transfer system, preliminarily analyzes the influence factors of the subway transfer perception service quality, and acquires the data of related variables by using a questionnaire survey combined with a field verification method. And on the premise of considering interaction of influence factors, the subway transfer perception service quality is evaluated through a probability graph Model-Bayesian Network Model (Bayesian Network Model), and an improved PC algorithm is provided for learning the structure of the Bayesian Network. Based on the constructed subway transfer perception service quality Bayesian network model, the final influence factors of the subway transfer perception service quality and the interaction among the factors can be determined. And finally, carrying out quantitative analysis on the action effect of each influence factor of the subway transfer perception service quality by utilizing scene analysis, thereby determining an optimal strategy for improving the subway transfer perception service quality. The invention can provide effective reference for theoretical innovation and engineering practice of station passenger quality service evaluation in a subway system.
Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
Fig. 1 is a flowchart of a subway transfer awareness service quality assessment method according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a Bayesian network according to an embodiment of the present invention;
FIG. 3 is a flow chart of a Bayesian network modeling process provided by an embodiment of the present invention;
fig. 4 is a flow chart of an improved PC algorithm according to an embodiment of the present invention.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to the same or similar elements or elements having the same or similar function throughout. The embodiments described below with reference to the accompanying drawings are illustrative only for the purpose of explaining the present invention, and are not to be construed as limiting the present invention.
As used herein, the singular forms "a", "an", "the" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and/or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. It will be understood that when an element is referred to as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element or intervening elements may also be present. Further, "connected" or "coupled" as used herein may include wirelessly connected or coupled. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.
It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless expressly so defined herein.
For the convenience of understanding the embodiments of the present invention, the following description will be further explained by taking several specific embodiments as examples in conjunction with the drawings, and the embodiments are not to be construed as limiting the embodiments of the present invention.
The embodiment of the invention provides a subway transfer perception service quality assessment method, as shown in figure 1, comprising the following steps:
s1, acquiring subway transfer perception service quality data, including:
and S11, based on a three-dimensional attribution theory, starting from three aspects of an action person (transfer passenger), an objective stimulus (subway transfer connection system) and a relationship or situation (transfer state), dividing the influence factors of the subway transfer perception service quality into three categories, namely the personal attribute and the travel characteristic of the transfer passenger, the basic evaluation factors of the transfer and the transfer environment factors. The personal attributes of the passengers reflect the differences of different individuals, the judgment and perception of people on a certain event/behavior can be influenced to a certain extent, and the travel characteristics of the passengers serve as the background of a certain travel behavior, so that the travel habits and preferences of people can be reflected; because the passenger transfer process comprises a walking process and a waiting process, basic evaluation factors of one-time transfer are set as actual transfer walking time, sensed transfer walking time, actual transfer walking distance, sensed transfer walking distance, actual transfer waiting time and sensed transfer waiting time; the transfer environment factors comprise physical factors and state factors, the physical factors of the transfer environment start from hardware facilities of a subway transfer system, a key component of a subway transfer station, namely a transfer channel, is selected as a main analysis object, and the state factors of the transfer environment are related to the influence degree of passengers by other transfer passengers in the transfer process. Specific factors influencing the subway transfer perception service quality are shown in table 1.
TABLE 1 Primary influencing factors of subway transfer perception service quality
Figure BDA0002395223070000111
S12, in order to obtain the data related to the variables, a questionnaire is used in data acquisition in combination with a field verification method, wherein the questionnaire can obtain the personal attributes and trip characteristics of the transfer passenger and the perception data of the transfer activity, and the field verification method can obtain the actual data of the transfer activity, specifically:
s121, questionnaire survey: and (4) surveying the latest transfer activity of the transfer passenger by adopting a post investigation method and through questionnaire design. The first part of the questionnaire is personal basic information of the passenger, including personal attribute variables of the passenger; the second part is travel information of the passengers, and the travel information comprises travel characteristic variables of the passengers and specific transfer paths for transfer behaviors of the passengers; the third part is the perception and evaluation of passengers for their transfer experience, including the perceived total time of passenger transfer, the perceived travel time, the perceived travel distance and the perceived waiting time, the evaluation of transfer environment factors and the number of trains that the transfer passengers actually wait (for calculation of the actual waiting time for subsequent transfers). In addition, on the questionnaire question option setting, regarding the relevant questions of time and distance, the option is set to an interval value, for example, the option of converting the perception total time is less than 5 minutes, 5 minutes to 10 minutes, and the like; regarding the passenger's evaluation of the transfer environment factors, the option is set to use a five-point scale.
S122, field verification: according to the transfer path answered by the visitor and the corresponding transfer date (working day or weekend) and time period, the experimental reduction is carried out on the transfer process of the visitor, and the actual traveling time length of transfer, the actual traveling distance of transfer, the number of stairs in the transfer traveling process and the departure interval of the transfer line are recorded in the transfer process. In addition, the actual waiting time for passenger transfer can be approximately estimated by the departure interval of the transfer line and the number of trains that the passenger waits, see equation (1). Since the train arrival interval to the platform is fixed, the waiting time of the passengers for the first train to be switched in can be approximately half of the train departure interval of the switching-in line, provided that the distribution of the arrival of the passengers to the switching-in platform is subject to uniform distribution.
Figure BDA0002395223070000121
Wherein, twTo transfer realityApproximate estimate of latency (unit: min), fhrDeparture interval (unit: min/row), N, for a line changewThe number of trains (unit: rank) waiting for passengers.
S2, constructing a subway transfer perception service quality evaluation model based on the subway transfer perception service quality data.
And (4) taking the data acquired in the step (S1) as an input variable of the model, evaluating the subway transfer perception service quality by constructing a Bayesian network model, and determining the final influence factor of the subway transfer perception service quality. The Bayesian network model is a graph model based on probability theory, and a network of the Bayesian network model consists of a directed acyclic network structure G and a network parameter theta. The network structure G is composed of a node set U and an edge set L of the network, that is, G ═ U, L, each node in the network represents a random variable, and then the node set is U ═ X1,X2...XnAnd edges among the nodes represent conditional dependencies among the random variables. The parameter theta of the Bayesian network is a set of conditional probability distributions, and the parameter theta of each node isiRepresenting node itself XiAnd its father node set PaiConditional probability distribution of (2), i.e. thetai＝P(Xi|πi). Thus, the bayesian network can be represented as a joint probability distribution, as shown in equation (2). Wherein n is the number of nodes in the Bayesian network, piiIs node XiParent node set PaiA set of values of the variable(s). Furthermore, a simple example of a bayesian network is shown in fig. 2. The process of modeling the bayesian network model is shown in fig. 3, where the dashed arrows indicate that the structure of the model needs to be modified back when the model is evaluated poorly.
Figure BDA0002395223070000131
S21, structure and parameter learning
The structural learning of the bayesian network means that the topological structure of the bayesian network is determined by using the input variables as network nodes. The invention provides an improved PC algorithm for structure learning. Definition of shellfishThe leaf network is G ═ { U, L }, where the set of points in the network is U ═ X }1,X2...XnCan be divided into target variables XCAnd attribute variable XB＝U\XC(ii) a The set of edges between nodes is L and is equal to node XiThe set of directly connected edges is denoted Li. In the invention, the sensing service quality of subway transfer is a target variable of a model, and the influencing factors of the sensing service quality of subway transfer are attribute variables of the model. As shown in fig. 4, the improved PC algorithm mainly includes five stages, namely, constructing an initial undirected network, performing edge deletion and initial orientation, outlier culling, performing final orientation of the network, and redundant edge detection. The specific operation steps of each stage are as follows.
The first stage is as follows: constructing initial undirected graph G'
S211, calculating mutual information I (X) of any two nodesi,Xj) Where i ≠ j. Judgment of I (X)i,Xj)＞σ1(where σ1≥0，σ1Value 0.01) is satisfied, if so, the node X is judged to be satisfiediAnd XjThe undirected edges in between are added to the set of edges L, otherwise they are not added. Wherein the expression of mutual information is shown in formula (3), and xiAs discrete random variables XiCorresponding value, xjAs discrete random variables XjThe corresponding value.
Figure BDA0002395223070000132
S212, firstly, determining the maximum mutual information of each node in the network. Then, for L \ LCEach undirected edge in (a) is judged as follows: suppose the endpoints of the undirected edge are X respectivelyqAnd XsWhere q ≠ s, as node XqMaximum mutual information M (X)q) And node XsMaximum mutual information M (X)s) As a reference, I (X) is judgedq,Xs)＞αM(Xq) (0 < α < 1) and I (X)s,Xq)＞αM(Xs) Whether or not it is satisfied, if at least one condition is satisfied, node XqAnd XsReserve the undirected edge between, otherwiseThe edge is deleted from the set of edges L resulting in the initial undirected graph G'.
And a second stage: performing edge deletion and primary orientation to obtain a primary orientation network G ″
S213, after the initial undirected graph G 'is used for replacing a complete undirected graph in the PC algorithm, the PC algorithm is called to delete and initially orient the edges in the network, and G' is obtained. The specific operation steps comprise the following three steps:
s2131, based on G', for any two adjacent nodes X in the networkiAnd XjBy node XiAnd XjThe set of adjacent nodes in (2) is used as a condition set C, and a condition independence test is performed. The order of the condition set starts from zero order and increases in ascending order if the condition set C can be foundijSo that at a given CijUnder the premise of (1), XiAnd XjCondition independent, remove node XiAnd XjAnd no directional edge in between. When all nodes in the network complete the condition independence test, the side deletion work is completed;
s2132, if the network obtained based on the S2131 can find three nodes X in the undirected graphi,Xj,XkSatisfies node XiAnd XkThere is a non-directional edge between, node XjAnd XkThere is a non-directional edge between, node XiAnd XjThere is no non-directional edge therebetween, at the same time
Figure BDA0002395223070000141
The V structure can be determined: xi→Xk←Xj；
And S2133, based on the network obtained in the S2132, the remaining undirected edges in the network are oriented by a mandatory orientation principle to obtain G'. Wherein, the principle of forced orientation includes 2 points: 1. for the triangular shape in the network, if the directions of two sides are known, the direction of the third side is determined according to the principle of no ring formation; 2. any three nodes X in the networki,Xj,XkIf X isiIs XjParent node of, XjIs XkAdjacent nodes of, and XiIs not XkAdjacent node of (2), then order XkIs XjThe parent node of (2).
S214, judging the isolated points and the undirected edges of the G': if the isolated point exists, go to step S215; if no isolated point exists but no undirected edge exists, go to step S216; if no isolated point and no directional edge exist, go to step S217.
And a third stage: removing isolated points
S215, removing the isolated points in the preliminary directional network G', namely deleting the random variables corresponding to the isolated points, and returning to the step S212.
A fourth stage: carrying out final orientation of the network to obtain a final oriented network G'
S216, firstly, obtaining all possible directed acyclic Bayesian network structure set DAGs formed by undirected edges according to the undirected edges in the preliminary directed network G ″S. Then, the DAG is scored by using BIC functionSAll the network structures in the network are scored, and the network with the best score is selected as the final directed network G'. The expression of the BIC scoring function is shown in formula (4).
Figure BDA0002395223070000151
Wherein q isiIs node XiParent node set PaiNumber of variable value combinations of (1), miIs node XiNumber of values of (A), NijkIs node XiParent node set PaiWhen the j variable is combined, the node XiIs the number of samples of the kth possible value, and
Figure BDA0002395223070000152
and N is the total number of samples.
The fifth stage: performing redundant edge detection to obtain the final network structure G
Based on G', if the non-collision path between any two adjacent nodes in the network exceeds 1, redundant edge detection is required. Wherein, the non-collision path refers to a path not including the collision node.
S217, determining any two adjacent nodes XiAnd XjCutting set V ofij. For two adjacent nodes XiAnd XjIn a term of Xj→XiThen cutting set VijIs node XiFather node set PaiAnd is given XiFather node set
Figure BDA0002395223070000153
Determining a cut set VijThe method comprises finding a connection node XiAnd XjAnd passes through Xi j(j＝1,2,...,fi) And put it into the set FPijThen repeat the following 2 steps until set FPijAll non-collision paths in (a) are deleted.
S2171 and mixing PaiMiddle node XiAnd XjPut into VijAnd the non-collision path passing through the part of nodes is extracted from the set FPijDeleting;
s2172 and mixing PaiIn which the node blocking the most non-collision path is placed in Vij。
S218, in the given cutting set VijOn the premise of (1), judging the condition mutual information I (X)i,Xj|Vij)＞σ2(where σ2≥0，σ2Value 0.01) is satisfied. If yes, reserving adjacent node XiAnd XjOtherwise, deleting the edge to obtain the final network structure G. Wherein the expression of the conditional mutual information is shown in formula (5), and vijTo cut and collect VijAnd (4) values of the variable combinations.
Figure BDA0002395223070000154
Generally, the PC algorithm mainly includes three stages, namely, constructing a completely undirected network, performing edge deletion using a conditional independence test, performing orientation of edges in the network by identifying V structures in the network, and using a forced orientation principle. The improved PC algorithm provided by the invention can be known that the algorithm constructs an initial undirected network by introducing mutual information in the first stage, replaces the construction process of a completely undirected network in the first stage of the PC algorithm, and effectively reduces the dimension and times of subsequent condition independence test; after the third stage of the classic PC algorithm is completed, the processes of deleting isolated points, utilizing the BIC scoring function to orient the residual undirected edges in the network, detecting redundant edges and deleting the residual undirected edges are added, namely the third stage to the fifth stage in the algorithm, the defects in the PC algorithm can be perfected, the unreasonable operation is avoided, and the learned network structure is more reasonable.
Based on the bayesian network structure learned in the above five stages, a parameter learning algorithm is used to estimate the conditional probability distribution of each node in the network. The method adopts a Maximum Likelihood estimation (Maximum Likelihood estimation) algorithm to learn the parameters in the network obtained by learning. This part calls the bayesian network toolbox in Matlab directly to implement.
S22 probabilistic reasoning
The probability inference of the bayesian network model refers to calculating the probability of any node or node set value by using the topological structure of the bayesian network and the conditional probability distribution of each node on the premise of giving evidence. The invention adopts clique to combine with the tree transmission algorithm to carry out probabilistic reasoning and is realized by utilizing a Bayesian network tool box in Matlab. And on the basis of the Bayesian network obtained by learning, the subway transfer perception service quality is taken as a target variable, the other variables in the network are taken as attribute variables, and the target variable category with the maximum posterior probability is obtained by inference by inputting the value of the attribute variables as evidence, so that the evaluation result of the subway transfer perception service quality is obtained. Assume that a Bayesian network contains n variables, divided into attribute variables XB＝{XB1,XB2,...,XBn-1And the target variable XC. Each attribute variable XBi∈XBIs r isiOne possible value, the target variable XCWith t possible values, each sample in the sample set D can be defined as { x }B1,xB2,...,xBn-1,xC∈ D, wherein
Figure BDA0002395223070000161
xC∈{xC1,xC2,...,xCt}. Thus, given a network structure G, the predicted values of the target variables
Figure BDA0002395223070000171
Calculated from formula (6) and formula (7).
Figure BDA0002395223070000172
Figure BDA0002395223070000173
Wherein x isBiAs attribute variable XBiValue of (a), xCIs a target variable XCValue of (a) ("nCIs a target variable XCIs a variable value set of the parent node set, piBiAs attribute variable XBiA set of variable values for the parent node set.
S23, model evaluation
And (3) introducing an overall index overall prediction accuracy rate and a Kappa coefficient to analyze and evaluate the overall performance of the model based on a probabilistic reasoning result, wherein the overall index overall prediction accuracy rate and the Kappa coefficient are respectively shown in a formula (8) and a formula (9). The overall index prediction accuracy rate represents the proportion of all classified correct samples, and the Kappa coefficient is used for analyzing the consistency of the actual observed value and the predicted value. The Kappa index is between-1.00 and 1.00, and when the index value is greater than 0.60, the consistency is good, and when the index value is greater than 0.80, the consistency is almost complete. And (3) introducing the accuracy of the category index, the recall degree and the prediction capability of the F measured value analysis model for different categories of target variables, which are respectively shown in a formula (10), a formula (11) and a formula (12). The values of index accuracy and recall degree are both between 0.00 and 1.00, the closer the index value is to 1.00, the better the prediction effect is represented, and the F measurement value is the weighted harmonic value of the index value and the recall degree.
Figure BDA0002395223070000174
Figure BDA0002395223070000175
Figure BDA0002395223070000176
Figure BDA0002395223070000181
Figure BDA0002395223070000182
Wherein, PoOverall prediction accuracy is an indicator, N' is the number of samples for which the model predicts accuracy, NoIs the total amount of the sample, nj sActual number of samples, n, for the jth class of target variablesj yThe number of predicted samples for the jth class of target variables; n isj' predicting the correct number of samples for the jth class of target variables, JQj、ZHj、FjAccuracy, recall, and F measurement for the jth category of target variables, respectively.
S3, based on the subway transfer perception service quality assessment model, determining the best strategy for improving the subway transfer perception service quality by analyzing the action effect of each influence factor on the subway transfer perception service quality assessment through a scene, wherein the strategy comprises the following steps:
based on the subway transfer perception service quality evaluation model established in the step S2, the effect of each attribute variable on the target variable subway transfer perception service quality is quantified by using scene analysis. Specifically, first, a reference mean value e (HGS) of the target variable is calculated with reference to the posterior edge probability P (HGS ═ j | e ═ Φ) of the different types of target variables, and the calculation formula is shown in formula (13). Then, for each attribute variable, on the premise of fixing the values of the other variables, the value of the variable (namely the value of the variable) is changedThe variables are measured at different values of Z ═ Zi) The posterior probabilities P of different classes of target variables are retrieved by probabilistic reasoning (HGS j e Z)i}) and calculate the new mean value of the target variable E (HGS)iThe calculation formula is shown in formula (14). Finally, the relative change amplitude E of the target variable mean value is usedi' to measure the effect of different values of attribute variables, the calculation formula is shown in formula (15). The single attribute variable value for improving the subway transfer perception service quality can be obtained through scene analysis of the single attribute variable, and then the best strategy for improving the subway transfer perception service quality can be explored through combination of a plurality of variable values.
Figure BDA0002395223070000183
Figure BDA0002395223070000184
Figure BDA0002395223070000191
HGS is the subway transfer perception service quality of the target variable, E (HGS) is the reference mean value of the target variable, E (HGS)iDifferent values of Z-Z for a given attribute variableiAs evidence, a new mean of the target variable.
In summary, the embodiment of the present invention provides a subway transfer perception service quality assessment method, which introduces a new assessment analysis process, and applies a probability map model to the assessment of subway transfer perception service quality to implement the proposing and analysis of a subway station transfer service level improvement policy. The method aims at the current situation that the existing subway transfer service quality assessment method generally considers the subway transfer service quality as a latent variable or a qualitative variable, and proposes the subway transfer sensing time as a basic variable of the subway transfer sensing service quality. Aiming at the current situation that objective variables or subjective variables are generally and singly considered in the existing subway transfer perception service quality assessment method, the influence factors of subway transfer perception service quality are analyzed from the perspective of combining a three-dimensional attribution theory with characteristics of a subway transfer system. Aiming at the current situation that the influence factors are not considered by the internal interaction in the conventional subway transfer perception service quality evaluation method, the subway transfer perception service quality is evaluated by a probability graph model-Bayesian network model, and an improved PC algorithm is provided for learning the structure of the Bayesian network. Finally, the constructed subway transfer perception service quality Bayesian network model can be used for exploring the optimal strategy for improving the subway transfer perception service quality.
Those of ordinary skill in the art will understand that: the figures are merely schematic representations of one embodiment, and the blocks or flow diagrams in the figures are not necessarily required to practice the present invention.
From the above description of the embodiments, it is clear to those skilled in the art that the present invention can be implemented by software plus necessary general hardware platform. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which may be stored in a storage medium, such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method according to the embodiments or some parts of the embodiments.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for apparatus or system embodiments, since they are substantially similar to method embodiments, they are described in relative terms, as long as they are described in partial descriptions of method embodiments. The above-described embodiments of the apparatus and system are merely illustrative, and the units described as separate parts may or may not be physically separate, and the parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
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 changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims (8)

1. A subway transfer perception service quality assessment method is characterized by comprising the following steps:
s1, acquiring subway transfer perception service quality data;
s2, constructing a subway transfer perception service quality evaluation model based on the subway transfer perception service quality data;
s3, based on the subway transfer perception service quality assessment model, analyzing the effect of each influence factor on subway transfer perception service quality assessment through a scene, and accordingly determining the best strategy for improving the subway transfer perception service quality.
2. The method according to claim 1, wherein the S1 comprises the following steps:
s11, classifying the influence factors of the subway transfer perception service quality, including: the method comprises the following steps of personal attributes and trip characteristics of transfer passengers, basic evaluation factors of transfer and transfer environment factors, wherein the basic evaluation factors of transfer are as follows: actual transfer travel time, perceived transfer travel time, actual transfer travel distance, perceived transfer travel distance, actual transfer wait time, perceived transfer wait time, the transfer environmental factors including: physical and status factors;
and S12, obtaining personal attributes and travel characteristics of the transfer passengers and perception data of the transfer activities through questionnaires, and obtaining actual data of the transfer activities through field verification.
3. The method according to claim 2, wherein the S12 comprises the steps of:
s121, questionnaire survey: the method comprises the following steps of adopting a post investigation method, carrying out investigation aiming at the latest transfer activity of a transfer passenger through questionnaire design, wherein the first part of the questionnaire is personal basic information of the passenger and comprises personal attribute variables of the passenger; the second part is travel information of the passengers, and the travel information comprises travel characteristic variables of the passengers and specific transfer paths for transfer behaviors of the passengers; the third part is the perception and evaluation of passengers on the transfer experience of the passengers, and comprises the total time of the passengers, the perceived running time, the perceived running distance and the perceived waiting time, the evaluation of transfer environment factors and the number of trains which are actually waited by the passengers; further, on the questionnaire question option setting, with respect to the relevant questions of time and distance, the option is set to the section value; regarding the evaluation of the passenger on the transfer environment factors, the option is set to adopt a five-point scale;
s122, field verification: according to the transfer path answered by the interviewee and the corresponding transfer date and time period, carrying out experimental reduction on the transfer process of the interviewee, and recording the actual traveling time length of transfer, the actual traveling distance of transfer, the number of stairs in the transfer traveling process and the departure interval of the transfer line in the transfer process; in addition, the actual waiting time for passenger transfer can be approximately estimated by the departure interval of the transfer route and the number of trains that the passenger waits, see formula (1),
Figure FDA0002395223060000021
wherein, twTo transfer an approximate estimate of the actual latency, the unit: the method comprises the following steps of (1) taking minutes; f. ofhrFor the departure interval of the line to be switched in, the unit: minutes/row; n is a radical ofwNumber of trains waiting for passengers, unit: and (4) columns.
4. The method according to claim 3, wherein the S2 comprises the following steps:
s21, structure and parameter learning:
a bayesian network is defined as G ═ U, L, where the set of points in the network is U ═ X1，X2...XnIs divided into target variables XCAnd attribute variable XB＝U\XC(ii) a The set of edges between nodes is L and is equal to node XiThe set of directly connected edges is denoted Li(ii) a The subway transfer perception service quality is a target variable of the model, and the influence factors of the subway transfer perception service quality are attribute variables of the model;
s211, calculating mutual information I (X) of any two nodesi,Xj) Where I ≠ j, judge I (X)i,Xj)＞σ1Whether or not, preset sigma1Value of where σ1Not less than 0; if yes, the node X is connectediAnd XjAdd undirected edges in between to the set of edges L, otherwise do not, wherein the expression of mutual information is shown in equation (3), and xiAs discrete random variables XiCorresponding value, xjAs discrete random variables XjThe corresponding value;
Figure FDA0002395223060000022
s212, determining the maximum mutual information of each node in the network, corresponding to L \ LCEach undirected edge in (a) is judged as follows: suppose the endpoints of the undirected edge are X respectivelyqAnd XsWhere q ≠ s, as node XqMaximum mutual information M (X)q) And node XsMaximum mutual information M (X)s) As a reference, I (X) is judgedq,Xs)＞αM(Xq) (0 < α < 1) and I (X)s,Xq)＞αM(Xs) Whether or not it is satisfied, if at least one condition is satisfied, node XqAnd XsThe undirected edge between the two is reserved, otherwise, the edge is deleted from the edge set L to obtain an initial undirected graph G';
s213, after the initial undirected graph G 'is used for replacing a complete undirected graph in the PC algorithm, the PC algorithm is called to delete and initially orient the edges in the network, and G' is obtained;
s214, judging the isolated points and the undirected edges of the G': if the isolated point exists, go to step S215; if no isolated point exists but no undirected edge exists, go to step S216; if no isolated point and no directional edge exist, go to step S217;
s215, removing the isolated points in the primary directional network G', namely deleting random variables corresponding to the isolated points, and returning to the step S212;
s216, obtaining all possible directed acyclic Bayesian network structure set DAGs formed by the undirected edges according to the undirected edges in the preliminary directed network G ″SUsing BIC scoring function to pair DAGSScoring all the network structures in the network structure, and selecting the network with the optimal score as a final oriented network G', wherein the expression of a BIC scoring function is shown in a formula (4),
Figure FDA0002395223060000031
wherein q isiIs node XiParent node set PaiNumber of variable value combinations of (1), miIs node XiNumber of values of (A), NijkIs node XiParent node set PaiWhen the j variable is combined, the node XiIs the number of samples of the kth possible value, and
Figure FDA0002395223060000032
n is the total number of samples;
s217, determining any two adjacent nodes XiAnd XjCutting set V ofij；
S218, in the given cutting set VijOn the premise of (1), judging the condition mutual information I (X)i,Xj|Vij)＞σ2Whether or not, preset sigma2Value of where σ2Not less than 0; if yes, reserving adjacent node XiAnd XjIn betweenAnd if not, deleting the edge to obtain a final network structure G, wherein the expression of the conditional mutual information is shown in formula (5), and v isijTo cut and collect VijThe value of the combination of the intermediate variables,
Figure FDA0002395223060000041
learning parameters in the final network structure G by adopting a maximum likelihood estimation algorithm based on the obtained final network structure G;
s22, probabilistic reasoning:
on the basis of the Bayesian network obtained by learning, the subway transfer perception service quality is taken as a target variable, the other variables in the network are taken as attribute variables, and the target variable category with the maximum posterior probability is obtained by inference through inputting the value of the attribute variables as evidence, so that the evaluation result of the subway transfer perception service quality is obtained;
s23, model evaluation:
based on a probabilistic reasoning result, introducing an overall index overall prediction accuracy and a Kappa coefficient to analyze and evaluate the overall performance of the model, respectively see formula (8) and formula (9), wherein the overall prediction accuracy represents all sample proportions with correct classification, the Kappa coefficient is used for analyzing the consistency of an actual observed value and a predicted value, the value of the index Kappa coefficient is between-1.00 and 1.00, when the index value is greater than 0.60, the consistency is good, and when the index value is greater than 0.80, the index value is nearly completely consistent;
the accuracy of the category index, the recall degree and the prediction capability of an F measured value analysis model to different categories of target variables are respectively shown in a formula (10), a formula (11) and a formula (12), the values of the accuracy and the recall degree are all between 0.00 and 1.00, the index value is closer to 1.00, the prediction effect is better, and the F measured value is a weighted harmonic value of the two;
Figure FDA0002395223060000042
Figure FDA0002395223060000043
Figure FDA0002395223060000051
Figure FDA0002395223060000052
Figure FDA0002395223060000053
wherein, PoOverall prediction accuracy is an indicator, N' is the number of samples for which the model predicts accuracy, NoIs the total amount of the sample, nj sActual number of samples, n, for the jth class of target variablesj yThe number of predicted samples for the jth class of target variables; n'jPredicting the correct number of samples, JQ, for the jth class of target variablesj、ZHj、FjAccuracy, recall, and F measurement for the jth category of target variables, respectively.
5. The method according to claim 4, wherein the S213 comprises:
s2131, based on G', for any two adjacent nodes X in the networkiAnd XjBy node XiAnd XjIs used as a condition set C, the condition independence test is carried out, the order of the condition set is increased in an ascending manner from the zero order, if the condition set C can be foundijSo that at a given CijUnder the premise of (1), XiAnd XjCondition independent, remove node XiAnd XjWhen all nodes in the network complete condition independence test, the side deletion work is completed;
s2132, based on the network obtained in the step S2131, if three nodes X can be found in the undirected graphi,Xj,XkIs full ofFoot node XiAnd XkThere is a non-directional edge between, node XjAnd XkThere is a non-directional edge between, node XiAnd XjThere is no non-directional edge therebetween, at the same time
Figure FDA0002395223060000054
The V structure can be determined: xi→Xk←Xj；
S2133, based on the network obtained in the step S2132, the remaining undirected edges in the network are oriented by a forced orientation principle to obtain G ", wherein the forced orientation principle comprises 2 points: 1. for the triangular shape in the network, if the directions of two sides are known, the direction of the third side is determined according to the principle of no ring formation; 2. any three nodes X in the networki,Xj,XkIf X isiIs XjParent node of, XjIs XkAdjacent nodes of, and XiIs not XkAdjacent node of (2), then order XkIs XjThe parent node of (2).
6. The method according to claim 4, wherein the S217 comprises: :
for two adjacent nodes XiAnd XjIn a term of Xj→XiThen cutting set VijIs node XiFather node set PaiAnd is given XiFather node set
Figure FDA0002395223060000061
Determining a cut set VijThe method comprises finding a connection node XiAnd XjAnd passes through Xi j(j＝1,2,...,fi) And put it into the set FPijThen repeat the following 2 steps until set FPijAll non-collision paths in (a) are deleted, including:
s2171 and mixing PaiMiddle node XiAnd XjPut into VijAnd will pass through this part of the knotNon-collision path of points from set FPijDeleting;
s2172 and mixing PaiIn which the node blocking the most non-collision path is placed in Vij。
7. The method according to claim 4, wherein the S22 specifically includes:
assume that a Bayesian network contains n variables, divided into attribute variables XB＝{XB1,XB2,...,XBn-1And the target variable XCEach attribute variable XBi∈XBIs r isiOne possible value, the target variable XCWith t possible values, each sample in the sample set D can be defined as { x }B1,xB2,...,xBn-1,xC∈ D, wherein
Figure FDA0002395223060000062
xC∈{xC1,xC2,...,xCtTherefore, the predicted value of the target variable given the network structure G
Figure FDA0002395223060000063
Calculated by formula (6) and formula (7),
Figure FDA0002395223060000064
Figure FDA0002395223060000065
wherein x isBiAs attribute variable XBiValue of (a), xCIs a target variable XCValue of (a) ("nCIs a target variable XCIs a variable value set of the parent node set, piBiAs attribute variable XBiA set of variable values for the parent node set.
8. The method according to claim 1, wherein the S3 includes:
calculating a reference mean value E (HGS) of the target variable by taking the posterior marginal probability P (HGS (j | e ═ phi)) of different types of the target variable as a reference, wherein the calculation formula is shown in a formula (13);
aiming at each attribute variable, on the premise of fixing the values of the other variables, the value of the variable is changed, namely the attribute variable takes a different value Z as ZiThe posterior probabilities P of different classes of target variables are retrieved by probabilistic reasoning (HGS j e Z)i}) and calculate the new mean value of the target variable E (HGS)iThe calculation formula is shown in formula (14);
relative amplitude of change E in the mean value of the target variableiThe effect of different values of the attribute variables is measured, the calculation formula is shown in a formula (15), the single attribute variable value for improving the subway transfer perception service quality can be obtained through scene analysis of the single attribute variable, and then the optimal strategy for improving the subway transfer perception service quality can be explored through combination of a plurality of variable values;
Figure FDA0002395223060000071
Figure FDA0002395223060000072
Figure FDA0002395223060000073
HGS is the subway transfer perception service quality of the target variable, E (HGS) is the reference mean value of the target variable, E (HGS)iDifferent values of Z-Z for a given attribute variableiAs evidence, a new mean of the target variable.
CN202010128773.3A 2020-02-28 2020-02-28 Subway transfer perception service quality assessment method Pending CN111340376A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010128773.3A CN111340376A (en) 2020-02-28 2020-02-28 Subway transfer perception service quality assessment method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010128773.3A CN111340376A (en) 2020-02-28 2020-02-28 Subway transfer perception service quality assessment method

Publications (1)

Publication Number Publication Date
CN111340376A true CN111340376A (en) 2020-06-26

Family

ID=71185705

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010128773.3A Pending CN111340376A (en) 2020-02-28 2020-02-28 Subway transfer perception service quality assessment method

Country Status (1)

Country Link
CN (1) CN111340376A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111814859A (en) * 2020-06-30 2020-10-23 南京航空航天大学 Three-dimensional space class correction method for XCT slice classification

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106124175A (en) * 2016-06-14 2016-11-16 电子科技大学 A kind of compressor valve method for diagnosing faults based on Bayesian network
CN108320040A (en) * 2017-01-17 2018-07-24 国网重庆市电力公司 Acquisition terminal failure prediction method and system based on Bayesian network optimization algorithm
US20190385105A1 (en) * 2018-06-13 2019-12-19 Zhejiang University Latent Ability Model Construction Method, Parameter Calculation Method, and Labor Force Assessment Apparatus

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106124175A (en) * 2016-06-14 2016-11-16 电子科技大学 A kind of compressor valve method for diagnosing faults based on Bayesian network
CN108320040A (en) * 2017-01-17 2018-07-24 国网重庆市电力公司 Acquisition terminal failure prediction method and system based on Bayesian network optimization algorithm
US20190385105A1 (en) * 2018-06-13 2019-12-19 Zhejiang University Latent Ability Model Construction Method, Parameter Calculation Method, and Labor Force Assessment Apparatus

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
FENG, XUESONG1 等: "Bayesian network modeling explorations of strategies on reducing perceived transfer time for urban rail transit service improvement in different seasons", 《CITIES》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111814859A (en) * 2020-06-30 2020-10-23 南京航空航天大学 Three-dimensional space class correction method for XCT slice classification
CN111814859B (en) * 2020-06-30 2021-09-14 南京航空航天大学 Three-dimensional space class correction method for XCT slice classification

Similar Documents

Publication Publication Date Title
Amérigo et al. A theoretical and methodological approach to the study of residential satisfaction
Contreras-Nieto et al. Bridge maintenance prioritization using analytic hierarchy process and fusion tables
Pendyala et al. An exploration of the relationship between timing and duration of maintenance activities
Helms Keeping up with the Joneses: Neighborhood effects in housing renovation
Liu et al. Incorporating household gathering and mode decisions in large‐scale no‐notice evacuation modeling
Arentze et al. Parametric action decision trees: Incorporating continuous attribute variables into rule-based models of discrete choice
Asanbe et al. Teachers’ performance evaluation in higher educational institution using data mining technique
Ebden et al. Network analysis on provenance graphs from a crowdsourcing application
Zumkeller et al. Panel surveys
Habib et al. Investigating structural changes in commuting mode choice preferences with repeated cross-sectional travel survey data: the contexts of Greater Toronto and Hamilton (GTHA) area
Öztürk A hybrid type-2 fuzzy performance evaluation model for public transport services
CN111340376A (en) Subway transfer perception service quality assessment method
Olaru et al. Modelling behavioural rules for daily activity scheduling using fuzzy logic
Suganya et al. Prediction of the level of air pollution using adaptive neuro-fuzzy inference system
Hawas Development and calibration of route choice utility models: factorial experimental design approach
Gleeson Estimating housing mortality
Khan Modelling and Microsimulation of Activity Generation, Activity Scheduling and Mobility Assignment
Kulkarni Development of performance prediction models using expert opinions
Bottero et al. Calculating composite indicators for sustainability
Duarte et al. Representing collective thinking through cognitive networks
Janssens et al. Identifying behavioral principles underlying activity patterns by means of Bayesian networks
Firera et al. Community Service Monitoring Information System at The Level of Community Harmony of Citizens
Chen Models of count with endogenous choices
Mawarni et al. Website Implementation for Making Scholarship Acceptance Decisions At SMK N 1 Talangpadang Using The Analytical Hieararchy Process Method
Rezapour et al. Properties of a transport instrument for measuring psychological impacts of delay on commuters, mokken scale analysis

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
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20200626