CN110751523A - Method and device for discovering potential high-value passengers - Google Patents
Method and device for discovering potential high-value passengers Download PDFInfo
- Publication number
- CN110751523A CN110751523A CN201911002633.5A CN201911002633A CN110751523A CN 110751523 A CN110751523 A CN 110751523A CN 201911002633 A CN201911002633 A CN 201911002633A CN 110751523 A CN110751523 A CN 110751523A
- Authority
- CN
- China
- Prior art keywords
- passenger
- airline
- travel
- information
- potential
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000009826 distribution Methods 0.000 claims abstract description 86
- 238000012545 processing Methods 0.000 claims abstract description 35
- 239000011159 matrix material Substances 0.000 claims description 29
- 230000008569 process Effects 0.000 claims description 15
- 230000010006 flight Effects 0.000 claims description 14
- 238000000605 extraction Methods 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 6
- 238000005065 mining Methods 0.000 abstract description 4
- 238000004364 calculation method Methods 0.000 description 10
- 238000010586 diagram Methods 0.000 description 9
- 238000007781 pre-processing Methods 0.000 description 5
- 238000005070 sampling Methods 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 230000006399 behavior Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008450 motivation Effects 0.000 description 2
- 230000005526 G1 to G0 transition Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000008707 rearrangement Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0255—Targeted advertisements based on user history
- G06Q30/0256—User search
-
- 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)
- Strategic Management (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Finance (AREA)
- Theoretical Computer Science (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Accounting & Taxation (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Primary Health Care (AREA)
- Tourism & Hospitality (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses a method and a device for discovering potential high-value passengers, which are used for discovering hidden travel intentions behind the travel of passengers based on processing large-scale travel data (historical passenger ticket booking information) of the passengers, obtaining travel intention distribution of each passenger, and further determining the potential high-value passengers of a predetermined airline company based on the travel intention distribution information of each passenger. Therefore, the potential high-value passenger discovering scheme based on the passenger travel intention is provided, the passenger travel intention can reflect the potential demand of the passenger on the airline, correspondingly, the potential value of the passenger can be determined more accurately by mining the travel intention distribution information hidden behind the passenger travel data, the potential high-value passenger with less records of the current passenger can be discovered, and the neglect of the traditional passenger value measuring method to the potential high-value passenger is avoided.
Description
Technical Field
The application belongs to the field of prediction theory and method technology application, and particularly relates to a method and a device for discovering potential high-value passengers.
Background
With the high-speed development of economy in China, aviation travel is increasingly common. Typically, the customer life cycle includes a latent phase, a developing phase, a stationary phase, and a declining phase. For civil aviation passengers, although the current real value of high-value passengers in the potential period and the development period is not high, the future value-added potential is high, and the marketing value is high.
The existing passenger value measurement methods include a number method, a mileage method and an RFM (license frequency Monety) model method. Wherein, the times method measures the value of passengers by accumulating the times of passengers taking the airplane; the mileage method measures the passenger value by accumulating the passenger boarding mileage; the RFM model is used for measuring passenger value by performing weighted calculation on three indexes, namely the latest passenger boarding date, the passenger boarding frequency and the passenger boarding expense. The methods all use historical travel data of passenger individuals as passenger value measurement bases, only can reflect the value of passengers based on historical travel conditions, and are difficult to accurately find potential high-value passengers; particularly, for passengers with few current boarding times, no historical boarding data is available to predict the probability that the passengers will take an airline which is not taken in the future in history, which causes the problem of cold start.
Therefore, it is necessary in the art to provide an implementation scheme capable of accurately and effectively discovering potential high-value passengers.
Disclosure of Invention
In view of this, an object of the present application is to provide a method and an apparatus for discovering a potential high-value passenger, which are used to accurately and effectively discover the potential high-value passenger, so as to provide convenience for the marketing work of an airline company.
Therefore, the application discloses the following technical scheme:
a method of discovering a potential high-value passenger, comprising:
obtaining historical passenger ticket booking information of an airline company;
processing the historical passenger booking information into a passenger airline travel record document in a predefined format;
processing the travel record document of the passenger airline based on a preset theme model to obtain travel intention distribution information of each passenger;
potential high-value passengers of a predetermined airline are determined based on travel intention distribution information of the respective passengers.
Preferably, the processing the passenger booking information into a passenger airline travel record document with a predefined format includes:
carrying out data deduplication processing on the passenger ticket booking information;
determining the number of annual passengers of each airline of an airline company;
screening a route P before the number of annual passengers to determine passenger travel intention distribution information;
processing passenger booking information corresponding to a P airline before the number of annual passengers into a passenger airline travel record document with a predefined format;
wherein, the P is an integer larger than 1.
Preferably, in the method, the processing the travel record document of the passenger airline based on the theme model to obtain the travel intention distribution information of the passenger includes:
initializing a first hyper-parameter α and a second hyper-parameter β of the preset topic model, randomly extracting the travel intention z of each of the previous P airlines, and constructing an initial state of a Markov chain;
extracting the travel intention of each route in the travel record document of the passenger routes one by one, and finishing one iteration each time the travel intention of one route is extracted;
repeatedly executing the iteration process until the extracted information reaches the target probability distribution;
acquiring a passenger travel intention distribution matrix theta of an airline company and a airline distribution matrix phi under all travel intentions based on an extraction result, wherein the theta and the phi form travel intention distribution information;
wherein the first hyperparameter α is a hyperparameter of the θ, and the second hyperparameter β is a hyperparameter of the φ.
The above method, preferably, the determining a potentially high-value passenger of a predetermined airline based on travel intention information of the passenger includes:
constructing a passenger potential value discovery business model based on passenger trip intention based on passenger boarding prior information, passenger loyalty, airline market share, passenger potential airline demand information and passenger trip intention information;
and based on the passenger potential value discovery business model, discovering potential high-value passengers of the predetermined airline company from the passengers corresponding to the historical passenger booking information.
Preferably, the method for constructing a passenger potential value discovery service model based on the passenger travel intention based on the passenger boarding prior information, the passenger loyalty, the airline market share, the passenger potential airline demand information, and the passenger travel intention information includes:
calculating passenger boarding prior information based on the passenger boarding total times and the total ticket booking record numbers of all airlines;
calculating the loyalty of the passenger to the predetermined airline company based on the total number of times of taking the predetermined airline company and the historical total number of times of taking the airplane;
calculating the market share of the predetermined airline company on a certain airline based on the total number of flights the predetermined airline company opens on the certain airline and the total number of flights all the airline companies open on the certain airline;
calculating the potential airline demand information of the passenger for the airlines based on the total number of times of taking the airlines by the passenger and the total number of times of taking all airlines by the passenger;
constructing a passenger potential value discovery business model based on statistics based on the boarding prior information, the loyalty of passengers to the predetermined airline company, the market share of the predetermined airline company on the airline and the potential airline demand information;
and optimizing the potential airline demand information in the passenger potential value discovery service model based on statistics based on the passenger travel intention distribution matrix theta and airline distribution matrices phi under all travel intentions to obtain the passenger potential value discovery service model based on the passenger travel intention.
A discovery apparatus for potential high-value passengers, comprising:
the ticket booking information acquisition unit is used for acquiring historical ticket booking information of passengers of an airline company;
the document generating unit is used for processing the historical passenger ticket booking information into a passenger airline travel record document in a predefined format;
the intention determining unit is used for processing the travel record document of the passenger airline based on a preset theme model to obtain travel intention distribution information of each passenger;
and the high-value passenger determining unit is used for determining potential high-value passengers of the predetermined airlines based on the travel intention information of the passengers.
Preferably, the document generating unit of the apparatus is specifically configured to:
carrying out data deduplication processing on the passenger ticket booking information;
determining the number of annual passengers of each airline of an airline company;
screening a route P before the number of annual passengers to determine passenger travel intention distribution information;
processing passenger booking information corresponding to a P airline before the number of annual passengers into a passenger airline travel record document with a predefined format;
wherein, the P is an integer larger than 1.
The above apparatus, preferably, the intention determining unit is specifically configured to:
initializing a first hyper-parameter α and a second hyper-parameter β of the preset topic model, randomly extracting the travel intention z of each of the previous P airlines, and constructing an initial state of a Markov chain;
extracting the travel intention of each route in the travel record document of the passenger routes one by one, and finishing one iteration each time the travel intention of one route is extracted;
repeatedly executing the iteration process until the extracted information reaches the target probability distribution;
acquiring a passenger travel intention distribution matrix theta of an airline company and a airline distribution matrix phi under all travel intentions based on an extraction result, wherein the theta and the phi form travel intention distribution information;
wherein the first hyperparameter α is a hyperparameter of the θ, and the second hyperparameter β is a hyperparameter of the φ.
The above apparatus, preferably, the high-value passenger determination unit is specifically configured to:
constructing a passenger potential value discovery business model based on passenger trip intention based on passenger boarding prior information, passenger loyalty, airline market share, passenger potential airline demand information and passenger trip intention information;
and based on the passenger potential value discovery business model, discovering potential high-value passengers of the predetermined airline company from the passengers corresponding to the historical passenger booking information.
The above-mentioned device, preferably, the high-value passenger determination unit constructs a passenger potential value discovery service model based on passenger travel intention based on passenger boarding prior information, passenger loyalty, airline market share, passenger potential airline demand information, and passenger travel intention information, and specifically includes:
calculating passenger boarding prior information based on the passenger boarding total times and the total ticket booking record numbers of all airlines;
calculating the loyalty of the passenger to the predetermined airline company based on the total number of times of taking the predetermined airline company and the historical total number of times of taking the airplane;
calculating the market share of the predetermined airline company on a certain airline based on the total number of flights the predetermined airline company opens on the certain airline and the total number of flights all the airline companies open on the certain airline;
calculating the potential airline demand information of the passenger for the airlines based on the total number of times of taking the airlines by the passenger and the total number of times of taking all airlines by the passenger;
constructing a passenger potential value discovery business model based on statistics based on the boarding prior information, the loyalty of passengers to the predetermined airline company, the market share of the predetermined airline company on the airline and the potential airline demand information;
and optimizing the potential airline demand information in the passenger potential value discovery service model based on statistics based on the passenger travel intention distribution matrix theta and airline distribution matrices phi under all travel intentions to obtain the passenger potential value discovery service model based on the passenger travel intention.
According to the scheme, the method and the device for discovering the potential high-value passenger can discover the travel intention hidden behind the passenger travel based on processing large-scale passenger travel data (historical passenger ticket booking information), obtain the travel intention distribution of each passenger, and further determine the potential high-value passenger of the predetermined airline company based on the travel intention distribution information of each passenger. Therefore, the potential high-value passenger discovering scheme based on the passenger travel intention is provided, the passenger travel intention can reflect the potential demand of the passenger on the airline, correspondingly, the potential value of the passenger can be determined more accurately by mining the travel intention distribution information hidden behind the passenger travel data, the potential high-value passenger with less records of the current passenger can be discovered, and the neglect of the traditional passenger value measuring method to the potential high-value passenger is avoided.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, it is obvious that the drawings in the following description are only embodiments of the present application, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
FIG. 1 is a flow chart of a method for discovering potential high-value passengers according to an embodiment of the present application;
FIG. 2 is a schematic diagram of a data structure of a PNR data set provided by an embodiment of the present application;
FIG. 3 is a schematic diagram of the preprocessing of historical passenger booking information provided by an embodiment of the present application;
FIG. 4 is a schematic format diagram of a passenger-airline travel record document provided by an embodiment of the present application;
FIG. 5 is an exemplary diagram of a travel intention of a traveler provided in an embodiment of the present application;
FIG. 6 is a schematic diagram illustrating an influence of a parameter λ on a passenger potential value discovery service model according to an embodiment of the present application;
fig. 7 is a schematic diagram illustrating correspondence between average travel intention similarity and different values of the number K of travel intentions according to an embodiment of the present application;
fig. 8 is a schematic structural diagram of a device for discovering a potential high-value passenger according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
The application provides a method and a device for discovering potential high-value passengers, which are used for accurately and effectively discovering the potential high-value passengers so as to provide convenience for marketing work of an airline company.
Referring to fig. 1, a schematic flowchart of a method for discovering a potential high-value passenger according to an embodiment of the present application is provided, in this embodiment, as shown in fig. 1, the method for discovering a potential high-value passenger includes the following processing steps:
The method and the device for detecting the passenger are well applicable to potential high-value passenger discovery of the civil aviation companies, and therefore in the step 101, the aviation companies can be one or more civil aviation companies correspondingly.
In this step 101, historical Passenger booking information of an airline company may be specifically obtained from PNR (Passenger book) data of the airline company, and the obtained historical Passenger booking information may be used as basic data on which potential high-value Passenger discovery needs to be subsequently performed on a predetermined airline company (an airline company with potential high-value Passenger discovery needs).
Because the PNR data set is large in scale, in a specific implementation, the PNR data in a predetermined time period can be selected to be acquired, and in the embodiment, for example, the PNR passenger booking record in the system for booking tickets in china civil aviation in 1 month to 12 months in 2011 (of course, other time periods can be selected) is specifically selected to perform the acquisition of the PNR data and the subsequent potential high-value passenger discovery process based on the intention. Where, referring to fig. 2, the PNR dataset typically includes the following attribute fields:
1) and (3) numbering the identity card: the identity card number is a number obtained by renumbering the original identity card number of the passenger for protecting the privacy of the passenger, and the specific passenger can not be intuitively known based on the identity card number (certainly, the corresponding passenger can be indirectly known based on a numbering rule);
2) airline number: the number of each airline company is obtained by numbering each airline company to protect the privacy of the airline company, and which airline company is specifically determined cannot be intuitively known based on the number of the airline company (of course, the corresponding airline company can be indirectly known based on a numbering rule);
3) takeoff date, takeoff time: the year, month, day, hour and minute information of the flight taken by the passenger;
4) departure airport, arrival airport: different airports are represented by different three-character codes, and the departure airport and the destination airport are utilized to obtain route information;
5) team name: passengers in group trip make a booking trip together;
6) discount information: reflecting air ticket discount information;
7) flight number: different flights are distinguished.
Statistics shows that a data set of the PNR passenger booking records of the chinese civil aviation booking system from 1 month 2010 to 12 months 2011 contains booking records of 129 airlines and 4129 airlines, and in a specific implementation, one or more airlines with a large booking data volume can be selected for processing, for example, an airline with a maximum booking data volume number of 155 is selected or other airline data with a large data volume can be combined for processing, and the like.
And 102, processing the historical passenger booking information into a passenger airline travel record document in a predefined format.
On the basis of obtaining historical passenger booking information of an airline company, such as the PNR passenger booking record of the airline company with the number of 155 (or other airline companies with large data volume), the historical passenger booking information of the PNR and the like can be further preprocessed, and then on the basis of preprocessing, the historical passenger booking information is processed into a passenger-airline travel record document with a predefined format, so that convenience is provided for subsequent potential high-value passenger discovery processing based on intentions.
Referring to fig. 3, the preprocessing of the historical passenger booking information may include, but is not limited to:
1) data deduplication: removing the repeated PNR data;
2) and (3) data statistics: counting the riding condition data of each airline of an airline company (such as the airline company with the number of 155 or other airline companies with large data quantity), such as counting the number of annual passengers of each airline;
3) and (3) screening routes: the potential high-value passenger originally has few boarding times, is not likely to take a cold flight again, and has too many selected routes, so that a matrix related to intention information in a subsequent processing process is too sparse, therefore, the embodiment screens out the cold routes and some temporarily-opened non-regular routes, and finally selects the non-cold routes and the regular routes of the airline company for intention discovery, specifically, for example, screens 974 routes (the non-cold routes and the regular routes) of the above-mentioned airline company number 155 (or other airline companies with large data amount may be combined for intention discovery, and the like.
After the above-mentioned preprocessing is performed on the historical passenger ticket booking information of the airline company, the embodiment further processes the historical passenger ticket booking information obtained after the preprocessing into a passenger-airline travel record document in a predefined format.
In practical implementation, a distributed computing model may be adopted for large-scale historical passenger ticket booking information processing, and the processing is performed as the format of a passenger-airline travel record document shown in fig. 4, and in the format of the passenger-airline travel record document shown in fig. 3, data is stored in a Key-value form, where the historical passenger ticket booking information of a passenger, such as a PNR record, may be specifically processed in the Key-value form through the following processing procedures:
1) and a mapping stage: for each PNR record, generating a < key, value >, where key is a passenger ID (such as the ID number obtained by numbering the original ID number of the user), and value is a flight line, where the flight line is formed by splicing a takeoff airport and a destination airport, for example, the flight line flying to the haihong bridge in beijing may be denoted as PEKSHA;
2) a rearrangement stage: loading airline ride records of the same key value (passenger ID) into a collection;
3) and (3) a specification stage: for each passenger, his airline ride record, specifically one row in fig. 3, is output.
And 103, processing the travel record document of the passenger airline based on a preset theme model to obtain the travel intention distribution information of each passenger.
The applicant finds that, in real life, a passenger travels based on certain travel intentions, the travel intentions exist objectively and are shared by all passengers, the travel intentions hidden behind the travel of civil aviation passengers can be obtained through large-scale passenger travel data, the travel intention distribution of each passenger can be accurately found, the probability that the passenger takes a historical non-riding route in the future can be predicted, and then the potential route demand of the passenger is found, and the potential route demand of the passenger further shows the potential value of the passenger, so the influence of the travel intentions of the passenger cannot be ignored in the calculation of the passenger value. Therefore, the travel intention of the passenger is introduced when the potential value of the passenger is determined to find the potential high-value passenger.
From the aspect of passenger travel, the travel intention is the motivation of a passenger for selecting a certain airline to travel, and more precisely, the reason of the passenger for selecting the airline is that the passenger has the travel intention, but the travel intention is that the passenger exists in the mind and the sea and is invisible.
After introducing the factor of the travel intention of the passenger, one travel of the passenger can be expressed as two aspects: the passenger goes out based on a certain intention with a certain probability and selects a certain travel route based on the travel intention.
In practical application scenes, specific motivations of passengers cannot be known in advance, and therefore the method and the device for obtaining the travel intentions of the passengers excavate and obtain travel intention distribution of the passengers through historical travel routes of the passengers under the condition of giving the number of the travel intentions, obtain travel intentions corresponding to each route in all historical travel routes of the passengers, and correspondingly calculate and obtain a passenger travel intention distribution matrix theta and route distribution matrices under all travel intentionsAnd the passenger travel intention distribution matrix theta and all travel intention lower route distribution matricesThe travel intention distribution information described in this step 103 is constructed.
Specifically, the key of potential high-value passenger discovery based on travel intentions is to obtain travel intentions corresponding to each airline in all passenger historical travel airlines, so as to obtain a passenger travel intention distribution matrix theta and all travel intention airline distribution matrices under all travel intentions
In the embodiment, the distribution matrix theta of the travel intention of the passenger and the distribution matrix of all lower routes of the travel intention are solved by adopting the idea of the theme modelThe travel intention distribution of the passengers is conformed to(i.e. a vector consisting of K dimensions α, α is described in detail below) as a parameterWhereinEach dimension value of α, extracting the travel intention of the passenger by polynomial probability, and the route distribution under different travel intentions is matched with(i.e. vector consisting of V dimension β, β see in particular the description below) as parameter V dimension Dirichlet distributionWhereinEach dimension value of β. all historical travel route records of each passenger form a passenger-route document of the passenger, the passenger-route documents of all passengers form a whole corpus corresponding to the passenger-route travel record documents, and each historical travel route record of each passenger occupies one position in the corpus.
Considering posterior probability p (z | r) of the travel route to the travel intention, the embodiment adopts a Gibbs sampling algorithm to solve the travel intention of each route in the historical travel route of the passenger. The Gibbs sampling algorithm is a fast and efficient mcmc (Markov Chain Monte carlo) sampling algorithm that aims to construct a Markov Chain that converges on a certain target probability distribution and to extract samples from the Chain that approximate the probability distribution values. All passenger historical travel route records form travel route record documentThe travel intention of each historical travel route record of the document forms a vectorRepresenting the number of times the route r ( r 1,2, … V) is assigned the travel intention k in all the passenger historical travel route records,representing passenger uiThe number of the historical travel route records belonging to the travel intention k in the historical travel route records. Then:
wherein the content of the first and second substances,to give the hyperparameter α, the joint probability of the route and intent,and alsoEquation (1) can therefore be written:
and solving the travel intention of each route record in the historical travel route record of the passenger by adopting a Gibbs sampling algorithm. Sampling the history record r of the d-th route in the document every timedThe travel intention zdWhile keeping the values of the other components unchanged. To facilitate the derivation of the Gibbs solution formula, the notation:representing the travel route record document after the d route history is removed,and representing a travel intention vector after removing the travel intention corresponding to the d-th route historical record.Representing passenger uiIn the airline document (except r)d) The number of route records belonging to the travel intention k,in a representation corpus (except r)d) The route r belongs to the number of occurrences of the travel intent k.
Is shown to be knownInference rdCorresponding to the probability that the travel intent is y. Then, it can be inferred from equation (2):
in conclusion, the passenger travel intention matrix theta and all travel intention lower route distribution matrices are solvedThe steps of (A) can be summarized as follows:
initializing a first hyper-parameter α and a second hyper-parameter β of the preset theme model, randomly extracting a travel intention z of each route, and constructing an initial state of a Markov (Markov) chain;
step two: extracting the travel intention of a historical travel route in the passenger-route travel record document;
step three: extracting the travel intention of each historical travel route in the passenger-route travel record document one by one, and finishing one iteration each time the travel intention of one historical travel route is extracted;
step four: repeatedly executing the iteration process until the extracted information reaches the target probability distribution;
and fifthly, acquiring a passenger travel intention distribution matrix theta of the preset airline company and a airline distribution matrix phi under all travel intentions based on an extraction result, wherein the theta and the phi form travel intention distribution information, the first hyper-parameter α is a hyper-parameter of the theta, and the second hyper-parameter β is a hyper-parameter of the phi.
Wherein θ and φ can be obtained based on the following calculation:
wherein, thetaui,yRepresenting passenger uiAs to the probability of the intention y,and each row in the phi matrix represents the probability of adopting the corresponding each route under a certain intention. Referring to fig. 5, for the illustration of travel intention of the passenger provided in this embodiment, for the illustration of travel intention in fig. 5, a certain row in the θ matrix may represent a probability of a certain passenger for each intention of the travel intention 1, the travel intention 2, and the travel intention 3, and a certain row in the Φ matrix may represent a probability of a corresponding route corresponding to the travel intention 1, the travel intention 2, or the travel intention 3, such as a probability of "beijing → shenyang" route being specifically corresponding to the travel intention 1 being 0.2, and a probability of "shenyang → beijing" route being 0.15 … …, etc.
And 104, determining potential high-value passengers of the predetermined airline company based on the travel intention distribution information of each passenger.
On the basis of determining the travel intention distribution information of the passengers, the method and the device for determining the potential high-value passengers of a certain predetermined airline company with the potential high-value passenger finding demand are determined on the basis of the travel intention distribution information of the passengers.
Specifically, in the embodiment, a statistical-based passenger potential value discovery service model is first constructed through a bayesian formula, and the passenger potential value includes two parts, namely the current value and the value corresponding to the potential airline demand. Wherein, the current value is influenced by the prior information of the passenger on the airplane and the loyalty of the passenger to the airline company; the potential value is influenced by the potential airline needs of the passenger and the market share of the airline airlines.
In view of the above, the present embodiment first calculates the passenger historical boarding prior information, passenger loyalty, airline market share, and passenger potential airline requirements, and then uses these information to construct a statistical-based passenger potential value discovery business model.
The passenger historical boarding prior information, the passenger loyalty, the airline market share and the passenger potential airline demand information can be calculated in the following modes:
passenger historical passenger ride prior information:
statistical passenger uiObtaining passenger boarding prior information P (u) based on the total times of boarding and the total booking record number of all airlinesi):
Passenger loyalty:
statistical passenger uiTotal number of taking said predetermined airline c, and passenger uiObtaining the total times of passenger u based on the historical total times of passenger ridingiLoyalty P (c | u) to airline ci):
Airline market share:
counting the total number of flights opened on a certain route r by the predetermined airline company c and the total number of flights opened on the route r by all the airlines at present, and calculating the market share P (c | r) of the airline company c on the route r on the basis of the total number:
potential airline requirements of passengers:
counting passengers u for passengers with historical passenger recordsiTotal number of times of taking route r, and passenger uiTaking all routes, and calculating the number of passengers uiPotential demand for flight line r P (r | u)i) Passenger airline demand P (r | u) without historical travel recordi) Temporarily noted as 0:
On the basis, a passenger potential value discovery business model based on statistics can be further constructed based on the passenger historical boarding prior information, the passenger loyalty, the airline market share and the passenger potential airline requirements.
Suppose there are M passengers, uiRepresenting any passenger (i ═ 1,2 … M), c represents a particular airline, R represents an airline, R represents a flight route, R represents a passenger namecIs the set of all airlines for airline c,for the set of historical travel routes for the passenger ui, probability p (u)i| c) represents passenger uiFor the value of airline c, the physical meaning is given to airline c, in terms of passenger uiPreference for airline c and for route R ∈ RcPassenger uiPossibility to select airline c. For p (u)iC) modeling, and expanding by a Bayesian formula, wherein the method comprises the following steps:
however, the passenger's historical airline information does not completely and objectively reflect the passenger's potential airline needs, and in real life, a passenger travels based on a certain travel intention, the travel intention exists objectively and is shared by all passengers, the probability that the passenger will take the historical airline before can be predicted by accurately finding the travel intention distribution of each passenger, and then the passenger's potential airline needs can be found, so that the passenger's potential airline needs P (r | u |) in the above equation (6) can be subjected to the travel intention distribution information of all passengersi) And (5) carrying out improvement.
If the historical travel data of the passengers comprises M passengers, the passenger uiIs provided withAnd (4) recording historical routes. Airline collection R for airline ccIn the middle of the flight, there are V routes (r is 1,2 … V), z represents the passenger travel intention, and the passenger travel is based on K travel intentions. Note p (z | u)i) For passenger uiThe probability of travel intent z is selected, p (r | z) is the probability of the flight route r appearing based on travel intent z. Thus, in the statistics-based potential high-value passenger discovery probability model, the passenger potential airline demand p (r | u)i) The calculation of (d) can be expressed as follows:
because passenger boarding behaviors are inherited and continuous, if the frequency characteristics of passenger historical boarding are ignored, the accuracy of calculation of potential airline requirements of passengers is seriously influenced. Therefore, whenEpsilon represents the number of times of the passenger's historical riding route r; when in useε is 1. Through the dynamic change of epsilon, not only the historical riding route of the passenger is addedFor calculating passenger pairingThe need for (1) provides a method.
M passenger travel intention distributions form an M multiplied by K order matrix theta, thetaRepresenting passenger uiThe distribution of travel intentions of all travel intentions forms a K multiplied by V order matrixAs a result of this, the number of the,
the passenger value represented by equation (6) includes: passenger current value p (u)i)p(c|ui) Value corresponding to passenger potential airline requirementsTwo parts. Lambda is a coefficient for balancing the current value and the value corresponding to the potential airline demand, and potential high-value passengers are passenger groups with higher values under the influence of a value comprehensive factor corresponding to the current value and the potential airline demand.
At the moment, different travelers have different travel intention distributions, so that different demands of passengers riding a certain airline for the airline can be distinguished. For all airlines c R ∈ RcPassing through passenger uiDistribution of travel intentionAnd distribution of routes under different travel intentionsThe method can calculate the corresponding airline R for the passenger based on different travel intentions according to the formula (9)cPotential requirements of (2). For passenger uiOf a historical trip, i.e. no routes r', i.e.Since p (c | r') ≠ 0,such that:
p(ui)p(r′|ui)p(c|r′)≠0 (10)
thus, obtaining the value component corresponding to the passenger's potential airline requirements in equation (6) has:
as can be seen from the formula (11), the passenger potential value discovery service model introducing the travel intention optimizes the calculation of the value corresponding to the passenger potential airline demand: through the selection of the coefficient lambda, on one hand, the inheritance and the continuity of passenger boarding behaviors are respected, and on the other hand, the boarding potential of an airline which is not ridden in passenger riding history is predicted. The formula (6) can more accurately form a passenger potential value discovery business model based on the travel intention.
In order to enable the passenger potential value discovery service model to discover potential high-value passengers, the value of lambda needs to be adjusted. So that the passenger potential value discovery business model can filter out current high-value passengers who are already in a decline period.
When lambda is larger, the calculation of the passenger value is greatly influenced by the current value of the passenger, and otherwise, when lambda is smaller, the calculation of the passenger value is greatly influenced by the value corresponding to the potential airline requirement of the passenger.
Referring to the schematic diagram of the influence of the parameter λ on the passenger potential value discovery service model shown in fig. 6, it can be known through multiple experiments by selecting different λ coefficients, that when the parameter λ is 0.2, the similarity coefficient between the passenger potential value discovery service model and the real high-value passenger collection Jaccard is the largest, so in this embodiment, it is preferable that λ is 0.2.
The passenger potential value based on the travel intentions discovers the business model, the hyper-parameter α and the number of the travel intentions K need to be set, the empirical value α of α can be 50.0/K, the β of 3501 represents the number of the travel intentions of the passenger, and the number cannot be directly observed.
Wherein, different trip intentions Zi、ZjSimilarity corre (Z) between themi,Zj) The calculation formula is as follows:
corre(Zi,Zj) The smaller, the travel intention Zi、ZjThe more independent the interval.
The average similarity avg _ corr (structure) between travel intentions is calculated as follows:
when the average similarity avg _ core (structure) among the travel intents is minimum, the corresponding K value is optimal.
Referring to fig. 7, it can be known that the average similarity of the passenger travel intentions is the smallest when the number K of the passenger travel intentions is 2 by performing the average similarity experiment of the passenger travel intentions for different K values.
Through the processing process, a passenger potential value discovery business model based on the travel intention can be obtained, on the basis, the potential value of each passenger to the preset airline company is calculated from the passengers corresponding to the historical passenger booking information according to the historical passenger boarding prior information, the passenger loyalty, the airline market share, the passenger potential airline demand of the passenger and the introduced passenger probability distribution information based on the model, and finally the potential high-value passenger is determined based on the calculated potential value of each passenger.
In this embodiment, based on that N is 10000 (that is, when the potential value of passengers is predicted according to equation (6), the TOP N passengers are selected according to descending order of their values for the prediction results), and the number k of travel intentions is 2, an experiment is performed, and the result shows that 11% more potential high-value passengers can be found by the discovery method based on the travel intentions than by the frequency method. Table 1 shows the comparison results of the number of flights of some of the discovered potential high-value passengers between 2010 and 2011.
TABLE 1
As can be seen from table 1, the potential high-value passenger discovering manner based on travel intentions can discover potential high-value passengers who have fewer passengers in 2010 but have significantly increased passengers in 2011. In addition, for the passengers who took airline passengers with the number of times of flight of 155 in 2010, in the discovery model based on the travel intention, the real value of the passengers is 0 currently for the airline with the number of 155, however, for all booking data in 2010, the passengers take flight records of other airlines, and the value corresponding to the potential airline needs of the passengers can be obtained by mining the travel intention. Therefore, the potential high-value passenger discovering mode based on the travel intention integrates the current real value of the passenger and the value corresponding to the potential airline demand, and can discover potential high-value passengers with lower current value, which cannot be discovered by traditional methods based on real data statistics, such as a frequency method.
According to the scheme, the method for discovering the potential high-value passenger provided by the embodiment of the application discovers the travel intention hidden behind the passenger travel based on processing of large-scale passenger travel data (historical passenger ticket booking information), obtains the travel intention distribution of each passenger, and further determines the potential high-value passenger of the predetermined airline company based on the travel intention distribution information of each passenger. Therefore, the potential high-value passenger discovering scheme based on the passenger travel intention is provided, the passenger travel intention can reflect the potential demand of the passenger on the airline, correspondingly, the potential value of the passenger can be determined more accurately by mining the travel intention distribution information hidden behind the passenger travel data, the potential high-value passenger with less records of the current passenger can be discovered, and the neglect of the traditional passenger value measuring method to the potential high-value passenger is avoided.
Corresponding to the above-mentioned discovery method for potential high-value passengers, the present application also discloses a discovery apparatus for potential high-value passengers, which is shown in fig. 8 as a schematic structural diagram, and may include:
a ticket booking information acquisition unit 801 for acquiring historical passenger ticket booking information of an airline company;
a document generating unit 802, configured to process the historical passenger booking information into a passenger airline travel record document in a predefined format;
an intention determining unit 803, configured to process the passenger airline travel record document based on a predetermined theme model, to obtain travel intention distribution information of each passenger;
a high-value passenger determination unit 804 for determining potential high-value passengers of the predetermined airline company based on the travel intention information of the respective passengers.
In an optional implementation manner of the embodiment of the present application, the document generating unit 702 is specifically configured to:
carrying out data deduplication processing on the passenger ticket booking information;
determining the number of annual passengers of each airline of an airline company;
and screening the routes P before the number of the annual passengers to determine the travel intention distribution information of the passengers.
Processing passenger booking information corresponding to a P airline before the number of annual passengers into a passenger airline travel record document with a predefined format;
wherein, the P is an integer larger than 1.
In an optional implementation manner of the embodiment of the present application, the intention determining unit 803 is specifically configured to:
initializing a first hyper-parameter α and a second hyper-parameter β of the preset topic model, randomly extracting a travel intention z of each route, and constructing an initial state of a Markov chain;
extracting the travel intention of each historical travel route in the travel record document of the passenger route one by one, and finishing one iteration each time the travel intention of one historical travel route is extracted;
repeatedly executing the iteration process until the extracted information reaches the target probability distribution;
acquiring a passenger travel intention distribution matrix theta of an airline company and a airline distribution matrix phi under all travel intentions based on an extraction result, wherein the theta and the phi form travel intention distribution information;
wherein the first hyperparameter α is a hyperparameter of the θ, and the second hyperparameter β is a hyperparameter of the φ.
In an optional implementation manner of the embodiment of the present application, the high-value passenger determination unit 804 is specifically configured to:
constructing a passenger potential value discovery business model based on passenger trip intention based on passenger boarding prior information, passenger loyalty, airline market share, passenger potential airline demand information and passenger trip intention information;
and based on the passenger potential value discovery business model, discovering potential high-value passengers of the predetermined airline company from the passengers corresponding to the historical passenger booking information.
In an optional implementation manner of the embodiment of the present application, the high-value passenger determining unit 804 constructs a passenger potential value discovery service model based on passenger travel intention based on passenger boarding prior information, passenger loyalty, airline market share, passenger potential airline demand information, and passenger travel intention information, and specifically includes:
calculating passenger boarding prior information based on the passenger boarding total times and the total ticket booking record numbers of all airlines;
calculating the loyalty of the passenger to the predetermined airline company based on the total number of times of taking the predetermined airline company and the historical total number of times of taking the airplane;
calculating the market share of the predetermined airline company on a certain airline based on the total number of flights the predetermined airline company opens on the certain airline and the total number of flights all the airline companies open on the certain airline;
calculating the potential airline demand information of the passenger for the airlines based on the total number of times of taking the airlines by the passenger and the total number of times of taking all airlines by the passenger;
constructing a passenger potential value discovery business model based on statistics based on the boarding prior information, the loyalty of passengers to the predetermined airline company, the market share of the predetermined airline company on the airline and the potential airline demand information;
and optimizing the potential airline demand information in the passenger potential value discovery service model based on statistics based on the passenger travel intention distribution matrix theta and airline distribution matrices phi under all travel intentions to obtain the passenger potential value discovery service model based on the passenger travel intention.
The device for discovering the potential high-value passenger disclosed in the embodiment of the present application is relatively simple in description since it corresponds to the method for discovering the potential high-value passenger disclosed in the above embodiment, and for the relevant similarities, please refer to the description of the method for discovering the potential high-value passenger in the above embodiment, and details thereof are omitted here.
It should be noted that, in the present specification, the embodiments are all described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments may be referred to each other.
For convenience of description, the above system or apparatus is described as being divided into various modules or units by function, respectively. Of course, the functionality of the units may be implemented in one or more software and/or hardware when implementing the present application.
From the above description of the embodiments, it is clear to those skilled in the art that the present application can be implemented by software plus necessary general hardware platform. Based on such understanding, the technical solutions of the present application may be essentially or partially implemented in the form of a software product, which may be stored in a storage medium, such as a ROM/RAM, a magnetic disk, an optical disk, etc., and includes several instructions for enabling 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 of the present application.
Finally, it is further noted that, herein, relational terms such as first, second, third, fourth, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The foregoing is only a preferred embodiment of the present application and it should be noted that those skilled in the art can make several improvements and modifications without departing from the principle of the present application, and these improvements and modifications should also be considered as the protection scope of the present application.
Claims (10)
1. A method for discovering a potential high-value passenger, comprising:
obtaining historical passenger ticket booking information of an airline company;
processing the historical passenger booking information into a passenger airline travel record document in a predefined format;
processing the travel record document of the passenger airline based on a preset theme model to obtain travel intention distribution information of each passenger;
potential high-value passengers of a predetermined airline are determined based on travel intention distribution information of the respective passengers.
2. The method according to claim 1, wherein the processing the passenger booking information into a passenger airline travel record document of a predefined format comprises:
carrying out data deduplication processing on the passenger ticket booking information;
determining the number of annual passengers of each airline of an airline company;
screening a route P before the number of annual passengers to determine passenger travel intention distribution information;
processing passenger booking information corresponding to a P airline before the number of annual passengers into a passenger airline travel record document with a predefined format;
wherein, the P is an integer larger than 1.
3. The method according to claim 2, wherein the processing the passenger airline travel record document based on the topic model to obtain the passenger travel intention distribution information comprises:
initializing a first hyper-parameter of the predetermined topic modelαA second hyper-parameter β, randomly extracting the travel intention z of each of the previous P routes, and constructing the initial state of the Markov chain;
extracting the travel intention of each route in the travel record document of the passenger routes one by one, and finishing one iteration each time the travel intention of one route is extracted;
repeatedly executing the iteration process until the extracted information reaches the target probability distribution;
acquiring a passenger travel intention distribution matrix theta of an airline company and a airline distribution matrix phi under all travel intentions based on an extraction result, wherein the theta and the phi form travel intention distribution information;
wherein the first hyper-parameterαThe second hyperparameter β is the hyperparameter of θ and the hyperparameter of φ.
4. The method of claim 3, wherein determining a potentially high value passenger for a predetermined airline based on travel intent information of the passenger comprises:
constructing a passenger potential value discovery business model based on passenger trip intention based on passenger boarding prior information, passenger loyalty, airline market share, passenger potential airline demand information and passenger trip intention information;
and based on the passenger potential value discovery business model, discovering potential high-value passengers of the predetermined airline company from the passengers corresponding to the historical passenger booking information.
5. The method according to claim 4, wherein the constructing of the passenger potential value discovery business model based on the passenger travel intention based on the passenger boarding prior information, the passenger loyalty, the airline market share, the passenger potential airline demand information, and the passenger travel intention information comprises:
calculating passenger boarding prior information based on the passenger boarding total times and the total ticket booking record numbers of all airlines;
calculating the loyalty of the passenger to the predetermined airline company based on the total number of times of taking the predetermined airline company and the historical total number of times of taking the airplane;
calculating the market share of the predetermined airline company on a certain airline based on the total number of flights the predetermined airline company opens on the certain airline and the total number of flights all the airline companies open on the certain airline;
calculating the potential airline demand information of the passenger for the airlines based on the total number of times of taking the airlines by the passenger and the total number of times of taking all airlines by the passenger;
constructing a passenger potential value discovery business model based on statistics based on the boarding prior information, the loyalty of passengers to the predetermined airline company, the market share of the predetermined airline company on the airline and the potential airline demand information;
and optimizing the potential airline demand information in the passenger potential value discovery service model based on statistics based on the passenger travel intention distribution matrix theta and airline distribution matrices phi under all travel intentions to obtain the passenger potential value discovery service model based on the passenger travel intention.
6. A device for discovering potentially high-value passengers, comprising:
the ticket booking information acquisition unit is used for acquiring historical ticket booking information of passengers of an airline company;
the document generating unit is used for processing the historical passenger ticket booking information into a passenger airline travel record document in a predefined format;
the intention determining unit is used for processing the travel record document of the passenger airline based on a preset theme model to obtain travel intention distribution information of each passenger;
and the high-value passenger determining unit is used for determining potential high-value passengers of the predetermined airlines based on the travel intention information of the passengers.
7. The apparatus according to claim 6, wherein the document generation unit is specifically configured to:
carrying out data deduplication processing on the passenger ticket booking information;
determining the number of annual passengers of each airline of an airline company;
screening a route P before the number of annual passengers to determine passenger travel intention distribution information;
processing passenger booking information corresponding to a P airline before the number of annual passengers into a passenger airline travel record document with a predefined format;
wherein, the P is an integer larger than 1.
8. The apparatus of claim 7, the intent determination unit to:
initializing a first hyper-parameter of the predetermined topic modelαA second hyper-parameter β, randomly extracting the travel intention z of each of the previous P routes, and constructing the initial state of the Markov chain;
extracting the travel intention of each route in the travel record document of the passenger routes one by one, and finishing one iteration each time the travel intention of one route is extracted;
repeatedly executing the iteration process until the extracted information reaches the target probability distribution;
acquiring a passenger travel intention distribution matrix theta of an airline company and a airline distribution matrix phi under all travel intentions based on an extraction result, wherein the theta and the phi form travel intention distribution information;
wherein the first hyper-parameterαThe second hyperparameter β is the hyperparameter of θ and the hyperparameter of φ.
9. The apparatus according to claim 8, said high value passenger determination unit being specifically configured to:
constructing a passenger potential value discovery business model based on passenger trip intention based on passenger boarding prior information, passenger loyalty, airline market share, passenger potential airline demand information and passenger trip intention information;
and based on the passenger potential value discovery business model, discovering potential high-value passengers of the predetermined airline company from the passengers corresponding to the historical passenger booking information.
10. The apparatus according to claim 9, wherein the high-value passenger determination unit constructs a passenger potential value discovery service model based on passenger travel intention based on passenger boarding prior information, passenger loyalty, airline market share, passenger potential airline demand information, and passenger travel intention information, and specifically includes:
calculating passenger boarding prior information based on the passenger boarding total times and the total ticket booking record numbers of all airlines;
calculating the loyalty of the passenger to the predetermined airline company based on the total number of times of taking the predetermined airline company and the historical total number of times of taking the airplane;
calculating the market share of the predetermined airline company on a certain airline based on the total number of flights the predetermined airline company opens on the certain airline and the total number of flights all the airline companies open on the certain airline;
calculating the potential airline demand information of the passenger for the airlines based on the total number of times of taking the airlines by the passenger and the total number of times of taking all airlines by the passenger;
constructing a passenger potential value discovery business model based on statistics based on the boarding prior information, the loyalty of passengers to the predetermined airline company, the market share of the predetermined airline company on the airline and the potential airline demand information;
and optimizing the potential airline demand information in the passenger potential value discovery service model based on statistics based on the passenger travel intention distribution matrix theta and airline distribution matrices phi under all travel intentions to obtain the passenger potential value discovery service model based on the passenger travel intention.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911002633.5A CN110751523A (en) | 2019-10-21 | 2019-10-21 | Method and device for discovering potential high-value passengers |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201911002633.5A CN110751523A (en) | 2019-10-21 | 2019-10-21 | Method and device for discovering potential high-value passengers |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110751523A true CN110751523A (en) | 2020-02-04 |
Family
ID=69279243
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201911002633.5A Pending CN110751523A (en) | 2019-10-21 | 2019-10-21 | Method and device for discovering potential high-value passengers |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110751523A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112052898A (en) * | 2020-09-03 | 2020-12-08 | 五邑大学 | Method and system for constructing potential classification model of intercity high-speed rail passenger |
CN112163787A (en) * | 2020-10-19 | 2021-01-01 | 科技谷(厦门)信息技术有限公司 | Passenger relative relationship prediction method based on big data |
CN112163716A (en) * | 2020-10-19 | 2021-01-01 | 科技谷(厦门)信息技术有限公司 | Passenger absolute relation prediction method based on big data |
CN112163786A (en) * | 2020-10-19 | 2021-01-01 | 科技谷(厦门)信息技术有限公司 | Civil aviation passenger personal influence assessment method based on big data and pagerank algorithm |
CN112163785A (en) * | 2020-10-19 | 2021-01-01 | 科技谷(厦门)信息技术有限公司 | Civil aviation passenger personal influence assessment method based on big data and neural network |
CN113900961A (en) * | 2021-12-08 | 2022-01-07 | 深圳市活力天汇科技股份有限公司 | Sample generation method, device, equipment and medium for automatic testing |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CA2833853A1 (en) * | 2013-03-06 | 2014-09-06 | Accenture Global Services Limited | Automatic preference-based waitlist and clearance for accommodations |
CN106779214A (en) * | 2016-12-15 | 2017-05-31 | 南开大学 | A kind of multifactor fusion civil aviation passenger travel forecasting approaches based on topic model |
CN106779872A (en) * | 2017-01-11 | 2017-05-31 | 广东工业大学 | A kind of passenger's divided method and device |
CN107886372A (en) * | 2017-12-06 | 2018-04-06 | 中国南方航空股份有限公司 | Customer value discovering method |
CN108596678A (en) * | 2018-05-02 | 2018-09-28 | 陈思恩 | A kind of airline passenger value calculation method |
-
2019
- 2019-10-21 CN CN201911002633.5A patent/CN110751523A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CA2833853A1 (en) * | 2013-03-06 | 2014-09-06 | Accenture Global Services Limited | Automatic preference-based waitlist and clearance for accommodations |
CN106779214A (en) * | 2016-12-15 | 2017-05-31 | 南开大学 | A kind of multifactor fusion civil aviation passenger travel forecasting approaches based on topic model |
CN106779872A (en) * | 2017-01-11 | 2017-05-31 | 广东工业大学 | A kind of passenger's divided method and device |
CN107886372A (en) * | 2017-12-06 | 2018-04-06 | 中国南方航空股份有限公司 | Customer value discovering method |
CN108596678A (en) * | 2018-05-02 | 2018-09-28 | 陈思恩 | A kind of airline passenger value calculation method |
Non-Patent Citations (3)
Title |
---|
卢敏等: "基于旅客出行意图的潜在高价值航线挖掘", 《铁路计算机应用》 * |
张继水: "基于出行意图发现的高价值旅客挖掘研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
王中强等: "基于改进马尔可夫链的航线预测算法", 《计算机应用》 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112052898A (en) * | 2020-09-03 | 2020-12-08 | 五邑大学 | Method and system for constructing potential classification model of intercity high-speed rail passenger |
CN112052898B (en) * | 2020-09-03 | 2024-01-05 | 五邑大学 | Construction method and system for potential classification model of intercity high-speed rail passenger |
CN112163787A (en) * | 2020-10-19 | 2021-01-01 | 科技谷(厦门)信息技术有限公司 | Passenger relative relationship prediction method based on big data |
CN112163716A (en) * | 2020-10-19 | 2021-01-01 | 科技谷(厦门)信息技术有限公司 | Passenger absolute relation prediction method based on big data |
CN112163786A (en) * | 2020-10-19 | 2021-01-01 | 科技谷(厦门)信息技术有限公司 | Civil aviation passenger personal influence assessment method based on big data and pagerank algorithm |
CN112163785A (en) * | 2020-10-19 | 2021-01-01 | 科技谷(厦门)信息技术有限公司 | Civil aviation passenger personal influence assessment method based on big data and neural network |
CN112163787B (en) * | 2020-10-19 | 2024-05-24 | 科技谷(厦门)信息技术有限公司 | Passenger relative relation prediction method based on big data |
CN112163786B (en) * | 2020-10-19 | 2024-05-28 | 科技谷(厦门)信息技术有限公司 | Civil aviation passenger personal influence assessment method based on big data and pagerank algorithm |
CN113900961A (en) * | 2021-12-08 | 2022-01-07 | 深圳市活力天汇科技股份有限公司 | Sample generation method, device, equipment and medium for automatic testing |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110751523A (en) | Method and device for discovering potential high-value passengers | |
Ray et al. | An ensemble-based hotel recommender system using sentiment analysis and aspect categorization of hotel reviews | |
Lucini et al. | Text mining approach to explore dimensions of airline customer satisfaction using online customer reviews | |
Tubishat et al. | Implicit aspect extraction in sentiment analysis: Review, taxonomy, oppportunities, and open challenges | |
Cui et al. | Personalized travel route recommendation using collaborative filtering based on GPS trajectories | |
Cai et al. | What are popular: exploring twitter features for event detection, tracking and visualization | |
Çavdar et al. | Airline customer lifetime value estimation using data analytics supported by social network information | |
KR20200007713A (en) | Method and Apparatus for determining a topic based on sentiment analysis | |
Bauman et al. | Discovering Contextual Information from User Reviews for Recommendation Purposes. | |
CN111090731A (en) | Electric power public opinion abstract extraction optimization method and system based on topic clustering | |
Liu et al. | Personalized air travel prediction: A multi-factor perspective | |
Xiao et al. | Coupled matrix factorization and topic modeling for aspect mining | |
Jiang et al. | Forecast and analysis of aircraft passenger satisfaction based on RF-RFE-LR model | |
Sun et al. | Multi-source information fusion for personalized restaurant recommendation | |
Trupthi et al. | Possibilistic fuzzy C-means topic modelling for twitter sentiment analysis | |
Chandra et al. | Collective representation learning on spatiotemporal heterogeneous information networks | |
Sabo et al. | Clustering of Brazilian legal judgments about failures in air transport service: an evaluation of different approaches | |
CN110751403A (en) | Credit scoring method and device | |
Pavlick et al. | Identifying 1950s american jazz musicians: Fine-grained isa extraction via modifier composition | |
Wu et al. | How Airbnb tells you will enjoy sunset sailing in Barcelona? Recommendation in a two-sided travel marketplace | |
Ellouze et al. | A comparative study on the extraction of dependency links between different personality traits | |
CN112926701B (en) | GCN semi-supervision-based classification method, system and equipment for airline passengers | |
Parbat et al. | Understanding the customer perception using machine learning while booking flight tickets | |
CN115409630A (en) | Insurance product accurate recommendation method based on mixed recommendation algorithm | |
Wang et al. | The Design of Advertising Text Keyword Recommendation for Internet Search Engines |
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 |
Application publication date: 20200204 |
|
RJ01 | Rejection of invention patent application after publication |