CN111833073A - Airline customer segmentation method based on K-Means + + algorithm - Google Patents
Airline customer segmentation method based on K-Means + + algorithm Download PDFInfo
- Publication number
- CN111833073A CN111833073A CN201910852736.4A CN201910852736A CN111833073A CN 111833073 A CN111833073 A CN 111833073A CN 201910852736 A CN201910852736 A CN 201910852736A CN 111833073 A CN111833073 A CN 111833073A
- Authority
- CN
- China
- Prior art keywords
- data
- algorithm
- observation window
- time
- airline
- 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.)
- Withdrawn
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
- G06Q30/00—Commerce
- G06Q30/01—Customer relationship services
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/951—Indexing; Web crawling techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Business, Economics & Management (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Accounting & Taxation (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Finance (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Probability & Statistics with Applications (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a K-Means + + clustering algorithm-based airline customer segmentation method, which comprises the following steps: firstly, the method comprises the following steps: crawling an airline data source, searching and preprocessing data in the first step, performing data modeling processing on the preprocessed data, and providing a more comprehensive method to improve the integrity and accuracy of aviation data acquisition, overcome the defect of high algorithm overhead caused by the influence of initial point selection due to outlier sensitivity, and concentrate limited resources on key crowds.
Description
Technical Field
The invention relates to an airline customer segmentation method based on a K-Means + + algorithm.
Background
In the information explosion era, the technology is rapidly developed, and artificial intelligence brings great changes to the daily life of people and gradually becomes an integral part of the life. Meanwhile, various data have penetrated into the fields of medicine, education, finance, aerospace and the like at present, along with the globalization, digitization and explosive growth of new internet information, the wide application range and variety of the internet data make the big data the most important subject in the present time. Artificial intelligence also enters a brand-new application stage from the traditional theoretical development and becomes an indispensable composition promoting factor in the information era. The data mining technology is the most important technology in the artificial intelligence practice field and is gradually and widely used in various fields due to the spectrum leaning usability.
The problem to be solved by data mining is to extract data with research value hidden in the data by a scientific and effective method in massive data with noise. Today, data mining technology has made great contributions in various fields. For example, in face recognition, a camera or a video camera is only required to collect images or video streams containing a face in advance, and the face can be automatically detected and tracked in an image library, so that a face recognition technology is performed, and the technology is successfully applied to identity detection scenes such as payment, work attendance checking and the like, even case detection and the like; and (4) user behavior analysis, namely establishing a user historical consumption and browsing condition data source, analyzing the past behaviors of the user and similar user behaviors, and providing predictive commodity recommendation service for the user so as to realize mutual win and mutual profit of the consumer and a merchant.
With the fierce competition of domestic and international airlines, the increase of civil and international low-price airlines and the development of other domestic large-scale airlines, the airlines face the huge pressure that the market share is squeezed and reduced year by year, how to hold the marketing strategy, and catch the chance in domestic and foreign markets with changeable wind and clouds, and break through the bottleneck in the enterprise development process; how to utilize advanced technologies such as internet addition and big data, integrate data and resources, and effectively improve the comprehensive competitiveness of enterprises by improving the execution efficiency and effect of marketing strategies is a problem in the current development of airlines.
Nowadays, large airlines have begun to provide more and more favorable marketing schemes for passengers to attract customers, the demands of passengers are more and more diversified, the requirements of the airlines are more and more detailed, and thus the product strategies of the airlines are more and more important. If the policy of the airline company is not targeted, for example, only if the current passenger needs are classified into business and travel, the current user needs are far from being matched, and the policy is introduced in the fierce competitive industry, a more detailed exclusive service needs to be provided for the passenger instead of providing the policies such as air ticket + hotel, air ticket + receiving and sending machine. The airline company excavates the purchasing behavior and consumption mode of the user from the mass historical data, reserves old customers, attracts new customers, provides exclusive resources for customers with different value characteristics, and achieves the maximization of enterprise benefits.
Disclosure of Invention
The purpose of the invention is as follows: in order to overcome the defects of the prior art, the invention provides an airline customer segmentation method based on a K-Means + + clustering algorithm,
2) the technical scheme is as follows: the invention provides a K-Means + + clustering algorithm-based airline customer segmentation method, which comprises the following steps: the method comprises the following steps: crawling an airline data source, and forming historical data from detailed information records of all passengers recorded in the airplane in the last two years, wherein the historical data comprises: the method comprises the following steps of (1) extracting detailed boarding passenger information in a data source in a month near to, and forming extracted incremental data by extracting the detailed boarding passenger information in the data source;
step two: data exploration and pretreatment; performing data exploration and pretreatment on the two data sources extracted in the step one, wherein the data exploration and pretreatment comprise missing value and abnormal value analysis, data cleaning, attribute stipulation and data transformation;
step three: carrying out data modeling processing on the preprocessed data; grouping the customers by using a K-Means + + algorithm; first, initial k cluster centers are determined: randomly selecting a sample as a cluster center c0Selecting a new centroid ciSelecting a sample X ∈ X probability ofRepeating the steps until k clustering centers are selected;
logging in a TipDM-HB data mining platform, importing the processed data source into a new scheme, calling a K-Means + + algorithm in the TipDM-HB data platform, and performing characteristic analysis, wherein the imported data column attributes are 6 attributes set in the second step; the imported data column attribute is "R, C, F, M, L", and is set as: the distance calculation method is euclidean distance, the convergence coefficient is 0.5, the threshold value is 0.5, the maximum iteration number is 5, the grouping number is 5, and the output is shown in fig. 4, wherein: l ═ LOAD _ TIME-FFP _ DATE; r ═ LAST _ TO _ END, (month), F ═ FLIGHT _ COUNT, (time), M ═ SEG _ KM _ SUM, (KM), C ═ AVG _ COUNT (KM);
step four: and aiming at value analysis result feedback, customizing corresponding service strategies for different value customers, and preferentially throwing resources to potential users with high F, C, M, L, low R and high C and low R, F, M, so as to perform differentiated management and improve the loyalty and satisfaction of the customers.
Step two, the data exploration and pretreatment are carried out on the two extracted data sources, and the method comprises the following steps:
3) using a regression filling method to continuously perform missing value filling processing on the two parts of data, and using an auxiliary variable XkEstablishing a regression model in a linear relation with the target variable Z so as to estimate the missing value of the target variable; the estimated value of the ith missing value is expressed as:wherein beta is0、βkIs a parameter, is a residual parameter, and follows a zero-mean normal distribution; cleaning data, in a Python compiling environment, deleting records with a fare less than or equal to 0, a discount rate less than or equal to 0 and a total flying mileage less than or equal to 0 to obtain new data, screening attributes in the new data based on an LRFMC model, selecting an observation window with the length of 2 years, and selecting 6 attributes: LOAD _ TIME, FFP _ DATE, LAST _ TO _ END, flush _ COUNT, SEG _ KM _ SUM, AVG _ COUNT, respectively: the end time of the observation window, the start time of the observation window, the time from the last flight taking time to the end of the observation window, the flight times of the observation window, the total flight kilometers of the observation window and the average discount rate belong to the processed dataAnd (4) a sexual convention.
2) The data is subjected to dispersion standardization, namely the original data is distributed in the interval [ min, max ]]To [ min ', max']The formula is as follows:
the observation window time period is selected to be two years.
The airline data source is crawled using Python or octopus crawler software.
Has the advantages that: the airline customer segmentation method based on the K-Means + + clustering algorithm has the following advantages:
1. obviously, a regression filling method filling missing value and dispersion standardization are added in the data processing to carry out data transformation, so that the completeness and the accuracy of data are improved;
compared with the traditional k-Means algorithm, the Means + + clustering algorithm is a more efficient clustering algorithm and can overcome the defect of high algorithm overhead caused by the influence of initial point selection due to sensitive outliers;
the FMC model can divide the attributes of the disordered data into high and low values, so that an airline can provide more targeted personalized customer customization services conveniently, and limited resources can be concentrated in key crowds. Overcomes the defects of the prior art
Drawings
FIG. 1 is a general flow chart of data mining modeling;
FIG. 2 attribute data sets;
FIG. 3 is a customer base profile analysis diagram;
FIG. 4 customer category feature analysis.
Detailed Description
The technical scheme of the invention is further explained in detail by combining the attached drawings:
fig. 1 is a general flow chart of data mining modeling, which is divided into four steps: crawling data sources, data exploration and preprocessing, data modeling and customer value degree identification, and the steps of the following technical scheme are basically the same.
The method comprises the following steps: the use of Python first requires that passenger member information be found at the airline customer data information website, on which basis two-click entries of hyperlinks are made. Extracting all useful data in the links under the a label by using a loop statement; according to the passenger detailed information link, entering a next-level website, and finding a second-level link a label required by us; parsing the read webpage codes by using a designated parser html.parser; reading the specific label of the analyzed webpage through the soup, and reading all tr; and returning the data to a point, and circularly acquiring the data. The octopus is a professional crawler software, websites needing to be crawled are added into a software creating task, circulation is set, and mass data resources can be crawled by clicking a required crawling information project.
After the data source of the airline company is crawled by using crawler professional software such as Python or octopus, the time period of an observation window is selected to be two years, historical data is formed by recording detailed information of all passengers recorded by riding an airplane, and the detailed information of the passengers riding the airplane in the last month in the data source is extracted to form extracted incremental data.
Step two: data exploration and preprocessing are important parts in the invention. And (4) performing data exploration and preprocessing on the two data sources extracted in the step one. The method comprises missing value and abnormal value analysis, data cleaning, attribute specification and data transformation.
1) The reasons for data loss are manifold, and it is mainly possible that some information cannot be acquired temporarily, so that a part of attribute values are left vacant; it is also possible that information is artificially lost; or that some property or properties of the object are not available; or the cost of acquiring such information is too great to acquire the data. However, the null value can make the important deterministic component in the research more difficult to grasp due to the loss of a large amount of useful information, so that the mining process is confused, and the output result is unreliable. The missing value and abnormal value analysis also comprises the following steps except for manually filling out the method which is too time-consuming: special value filling, average value filling, hot card filling, K nearest neighbor method, regression filling method and the like, wherein the regression filling method is low in cost, high in efficiency and high in precision and is suitable for the scene.
The traditional RFM model identifies the value of a client, and the application indexes are mainly three: time interval of consumption (Recency), Frequency of consumption (Frequency), and amount of consumption (money). And the consumption amount represents the total consumption amount of the client in a period of time. However, since the airline fare distance and the class of the slot have an influence on the fare and the same amount of money has a different value for the airline company, the LRFMC model is set, C is the average value of the discount coefficient corresponding to the slot, and M is the accumulated mileage for a certain period of time. In the same way, the member system is considered, the index L is increased as the conference time length, and in addition, the original consumption time interval R and the consumption frequency F are added, and the purpose of accurate marketing is achieved by the total five indexes.
Using a regression filling method to continuously perform missing value filling processing on the two parts of data, and using an auxiliary variable XkAnd establishing a regression model through a linear relation with the target variable Z, thereby estimating the missing value of the target variable. The estimate of the ith missing value can be expressed as:wherein beta is0、βkIs a parameter, is a residual parameter, and follows a zero-mean normal distribution. Cleaning data, in a Python compiling environment, deleting records with a fare less than or equal to 0, a discount rate less than or equal to 0 and a total flying mileage less than or equal to 0 to obtain new data, screening attributes in the new data based on an LRFMC model, selecting an observation window with the length of 2 years, and selecting 6 attributes: LOAD _ TIME, FFP _ DATE, LAST _ TO _ END, flush _ COUNT, SEG _ KM _ SUM, AVG _ COUNT, respectively: and performing attribute specification on the processed data according to the ending time of the observation window, the starting time of the observation window, the time from the last flight taking time to the ending time of the observation window, the flight times of the observation window, the total flight kilometers of the observation window and the average discount rate.
2) The normalization of the data is to scale the data according to a certain rule so that it falls into a small specific interval. Therefore, unit limitation of the data is removed, the data is converted into a dimensionless pure numerical value, and indexes of different units or orders of magnitude can be compared and weighted conveniently. The normalization methods include 0-1 normalization and Z normalization, Min-Max normalization, normalization and dispersion normalization.
The data is subjected to dispersion standardization, namely the original data is distributed in the interval [ min, max ]]To [ min ', max']The linear conversion of the original data can be realized, and the formula is as follows:the attribute data set is as in figure 2.
Step three: detailed description of the K-Means + + Algorithm
And carrying out data modeling processing on the preprocessed data. The customers are grouped using the K-Means + + algorithm. First, initial k cluster centers are determined: randomly selecting a sample as a cluster center c0Selecting a new centroid ciSelecting a sample X ∈ X probability ofThis is repeated until k cluster centers are selected. Logging in a TipDM-HB data mining platform, importing the processed data source into a new scheme, calling a K-Means + + algorithm in the TipDM-HB data platform, and performing characteristic analysis, wherein the imported data column attributes are 6 attributes set in the second step; the imported data column attribute is "R, C, F, M, L", and is set as: the distance calculation method is euclidean distance, the convergence coefficient is 0.5, the threshold value is 0.5, the maximum iteration number is 5, the grouping number is 5, and the output is shown in fig. 4, wherein: l ═ LOAD _ TIME-FFP _ DATE; r ═ LAST _ TO _ END, (month), F ═ FLIGHT _ COUNT, (time), M ═ SEG _ KM _ SUM, (KM), C ═ AVG _ COUNT (KM);
step four: aiming at the value analysis result feedback of the figure 4, corresponding service strategies are customized for different value customers, and resources are preferentially put to the potential users with high F, C, M, L and low R and the potential users with high C and low R, F, M, so that differentiated management is carried out, and the loyalty and satisfaction of the customers are improved.
The above description is only an embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can understand that the modifications or substitutions within the technical scope of the present invention are included in the scope of the present invention, and therefore, the scope of the present invention should be subject to the protection scope of the claims.
Claims (4)
1. An airline customer segmentation method based on a K-Means + + algorithm is characterized by comprising the following steps: the method comprises the following steps: the method comprises the following steps: crawling an airline data source, and forming historical data from detailed information records of all passengers recorded in the airplane in the last two years, wherein the historical data comprises: the method comprises the steps of collecting information of passengers on the airplane, and obtaining the information of passengers on the airplane in detail in the last month in a data source;
step two: data exploration and pretreatment; performing data exploration and pretreatment on the two data sources extracted in the step one, wherein the data exploration and pretreatment comprise missing value and abnormal value analysis, data cleaning, attribute stipulation and data transformation;
step three: carrying out data modeling processing on the preprocessed data; grouping the customers by using a K-Means + + algorithm; first, initial k cluster centers are determined: randomly selecting a sample as a cluster center c0Selecting a new centroid ciSelecting a sample X ∈ X probability ofRepeating the steps until k clustering centers are selected;
logging in a TipDM-HB data mining platform, importing the processed data source into a new scheme, calling a K-Means + + algorithm in the TipDM-HB data platform, and setting the imported data column attributes as 6 attributes set in the second step; the imported data column attribute is "R, C, F, M, L", and is set as: the distance calculation method is euclidean distance, the convergence coefficient is 0.5, the threshold value is 0.5, the maximum iteration number is 5, the grouping number is 5, and the output is shown in fig. 4, wherein: l ═ LOAD _ TIME-FFP _ DATE; r ═ LAST _ TO _ END, (month), F ═ FLIGHT _ COUNT, (time), M ═ SEG _ KM _ SUM, (KM), C ═ AVG _ COUNT (KM);
step four: and aiming at value analysis result feedback, customizing corresponding service strategies for different value customers, and preferentially throwing resources to potential users with high F, C, M, L, low R and high C and low R, F, M, so as to perform differentiated management and improve the loyalty and satisfaction of the customers.
2. The airline customer segmentation method based on the K-Means + + algorithm as claimed in claim 1, wherein: step two, the data exploration and pretreatment are carried out on the two extracted data sources, and the method comprises the following steps:
1) using a regression filling method to continuously perform missing value filling processing on the two parts of data, and using an auxiliary variable XkEstablishing a regression model in a linear relation with the target variable Z so as to estimate the missing value of the target variable; the estimated value of the ith missing value is expressed as:wherein beta is0、βkIs a parameter, is a residual parameter, and follows a zero-mean normal distribution; cleaning data, in a Python compiling environment, deleting records with a fare less than or equal to 0, a discount rate less than or equal to 0 and a total flying mileage less than or equal to 0 to obtain new data, screening attributes in the new data based on an LRFMC model, selecting an observation window with the length of 2 years, and selecting 6 attributes: LOAD _ TIME, FFP _ DATE, LAST _ TO _ END, flush _ COUNT, SEG _ KM _ SUM, AVG _ COUNT, respectively: and performing attribute specification on the processed data according to the ending time of the observation window, the starting time of the observation window, the time from the last flight taking time to the ending time of the observation window, the flight times of the observation window, the total flight kilometers of the observation window and the average discount rate.
3. the airline customer segmentation method based on the K-Means + + algorithm as claimed in claim 1, wherein: the observation window time period is selected to be two years.
4. The airline customer segmentation method based on the K-Means + + algorithm as claimed in claim 1, wherein: the airline data source is crawled using Python or octopus crawler software.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910852736.4A CN111833073A (en) | 2019-09-10 | 2019-09-10 | Airline customer segmentation method based on K-Means + + algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910852736.4A CN111833073A (en) | 2019-09-10 | 2019-09-10 | Airline customer segmentation method based on K-Means + + algorithm |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111833073A true CN111833073A (en) | 2020-10-27 |
Family
ID=72912569
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910852736.4A Withdrawn CN111833073A (en) | 2019-09-10 | 2019-09-10 | Airline customer segmentation method based on K-Means + + algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111833073A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWI776742B (en) * | 2021-11-29 | 2022-09-01 | 愛酷智能科技股份有限公司 | System for analyzing user behavior in information exchange platform |
CN116823346A (en) * | 2023-07-17 | 2023-09-29 | 广州百奕信息科技有限公司 | Civil aviation digital asset public comprehensive service method based on blockchain technology |
-
2019
- 2019-09-10 CN CN201910852736.4A patent/CN111833073A/en not_active Withdrawn
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWI776742B (en) * | 2021-11-29 | 2022-09-01 | 愛酷智能科技股份有限公司 | System for analyzing user behavior in information exchange platform |
CN116823346A (en) * | 2023-07-17 | 2023-09-29 | 广州百奕信息科技有限公司 | Civil aviation digital asset public comprehensive service method based on blockchain technology |
CN116823346B (en) * | 2023-07-17 | 2024-04-05 | 广州百奕信息科技有限公司 | Civil aviation digital asset public comprehensive service method based on blockchain technology |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110223168B (en) | Label propagation anti-fraud detection method and system based on enterprise relationship map | |
US11392840B2 (en) | System and method for generating recommendations | |
US20210042767A1 (en) | Digital content prioritization to accelerate hyper-targeting | |
CN105488697A (en) | Potential customer mining method based on customer behavior characteristics | |
CN112632405B (en) | Recommendation method, recommendation device, recommendation equipment and storage medium | |
Sumiran | An overview of data mining techniques and their application in industrial engineering | |
CN114997916A (en) | Prediction method, system, electronic device and storage medium of potential user | |
CN111833073A (en) | Airline customer segmentation method based on K-Means + + algorithm | |
CN113204603A (en) | Method and device for marking categories of financial data assets | |
CN116757779A (en) | Recommendation method based on user portrait | |
CN112070126A (en) | Internet of things data mining method | |
Singh et al. | Analysis of machine learning techniques for airfare prediction | |
CN107609921A (en) | A kind of data processing method and server | |
Girish et al. | Mining the Web Data for Classifying and Predicting Users’ Requests | |
Das et al. | A Review of Data Warehousing Using Feature Engineering | |
EP3493082A1 (en) | A method of exploring databases of time-stamped data in order to discover dependencies between the data and predict future trends | |
KR20190088813A (en) | Marketing Analysis Service Platform and method for Offline Store | |
Shen et al. | Research on e-commerce application based on web mining | |
CN114155057A (en) | Commodity recommendation system for electronic commerce platform | |
CN113408207A (en) | Data mining method based on social network analysis technology | |
CN114118610A (en) | Product sales prediction method and system based on relevance big data | |
Yan et al. | Customer segmentation using real transactional data in e-commerce platform: A case of online fashion bags shop | |
Vikram et al. | Data mining tools and techniques: a review | |
Yang et al. | Consumers’ Purchase Behavior Preference in E-Commerce Platform Based on Data Mining Algorithm | |
Xiaona | Personalized recommendation model for mobile E-commerce users |
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 | ||
WW01 | Invention patent application withdrawn after publication |
Application publication date: 20201027 |
|
WW01 | Invention patent application withdrawn after publication |