Civil aviation passenger behavior prediction analysis method, system, medium and equipment
By building a passenger behavior prediction model and optimizing it, the shortcomings in real-time and accuracy of the existing technology are solved, accurate prediction of passenger behavior is achieved, service quality and operational efficiency of airlines are improved, and passenger satisfaction is enhanced. and loyalty.
Patent Information
- Application Number
- CN202411196193.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-29
- Publication Date
- 2024-12-20
AI Technical Summary
The existing civil aviation passenger behavior prediction technology has shortcomings in real-time, accuracy and personalization, and it is difficult to meet the needs of the big data era, especially in improving service quality and enhancing passenger satisfaction.
A method for predicting and analyzing civil aviation passenger behavior is proposed. By obtaining real-time and historical data, a passenger behavior prediction model is constructed, including spatiotemporal analysis model, behavior sequence prediction model and social network model, and the reinforcement learning algorithm optimization model is used to predict passenger behavior. , and then personalized recommendations and operation decision support.
Accurate prediction of passenger behavior has been achieved, the airline's market strategy and scientificity of customer service has been improved, the passenger experience has been enhanced, satisfaction and loyalty has been improved, flight operation management has been optimized, costs have been reduced and benefits have been increased.
Smart Images
Figure CN119168693A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of civil aviation business management, and in particular to a method, system, medium and equipment for predicting and analyzing civil aviation passenger behavior. Background Art
[0002] With the rapid development of the civil aviation industry, passengers' demand for personalized services is growing. In order to improve service quality, enhance passenger satisfaction and loyalty, and optimize operational management, airlines need to have a deeper understanding and prediction of passenger behavior. However, existing technologies mostly rely on traditional statistical analysis methods, which are difficult to cope with the challenges of the big data era, especially in terms of real-time, accuracy and personalization. Summary of the invention
[0003] In order to solve the technical problems existing in the background technology, the present invention proposes a method, system, medium and equipment for predicting and analyzing civil aviation passenger behavior.
[0004] The present invention proposes a method for predicting and analyzing civil aviation passenger behavior, comprising:
[0005] Get real-time passenger data;
[0006] Use the passenger behavior prediction model to predict real-time passenger data and obtain prediction results.
[0007] Preferably, before using the passenger behavior prediction model to predict the real-time passenger data and obtaining the prediction result, the method further includes:
[0008] Build training model;
[0009] Get historical passenger data;
[0010] Analyze the passenger data to obtain historical characteristic data reflecting the passenger behavior characteristics; the historical characteristic data includes: historical flight records, historical behavior logs and historical social network data;
[0011] Divide historical feature data into training set and validation set;
[0012] Use the test set to train the training model and obtain a training model that passes the training;
[0013] The validation set is used to validate the trained model and obtain the passenger behavior prediction model.
[0014] Preferably, after obtaining the historical passenger data, the method further includes:
[0015] Clean historical passenger data.
[0016] Preferably, obtaining historical passenger data specifically includes:
[0017] Obtain historical passenger data according to the preset period.
[0018] Preferably, the training model that has passed the training is verified using the verification set to obtain the passenger behavior prediction model, specifically including:
[0019] Using the validation set to validate multiple evaluation indicators of the training model that has passed the training, the multiple evaluation indicators are used to indicate the effect of the training model;
[0020] The training model in which the multiple assessment indicators all meet the corresponding preset assessment indicators is determined as the passenger behavior prediction model.
[0021] Preferably, the passenger behavior prediction model includes a spatiotemporal analysis model, a passenger behavior sequence prediction model and a passenger social network model; wherein the spatiotemporal analysis model includes a CNN model, the passenger behavior sequence prediction model includes an RNN and LSTM model, and the passenger social network model includes a GNN model.
[0022] Preferably, before using the passenger behavior prediction model to predict the real-time passenger data and obtaining the prediction result, the method further includes:
[0023] Optimize the passenger behavior prediction model.
[0024] Preferably, the passenger behavior prediction model is optimized, specifically including:
[0025] Reinforcement learning algorithm is used to optimize the passenger behavior prediction model.
[0026] Preferably, after predicting the real-time passenger data using the passenger behavior prediction model and obtaining the prediction result, the method further includes:
[0027] Based on the prediction results, one or more of personalized recommendations, marketing strategy formulation, operational decision support, and customer relationship management maintenance are carried out.
[0028] Preferably, after performing one or more of personalized recommendation, marketing strategy formulation, operational decision support and customer relationship management maintenance according to the prediction results, the method further includes:
[0029] Gather feedback on the accuracy of forecast results through customer interactions, service usage, and market feedback;
[0030] Based on the accuracy feedback collected, the passenger behavior prediction model is iteratively updated regularly to adapt to changes in passenger behavior.
[0031] The present invention also proposes a civil aviation passenger behavior prediction and analysis system, comprising:
[0032] A first acquisition module is used to acquire real-time passenger data;
[0033] The prediction module is used to use the passenger behavior prediction model to predict the real-time passenger data and obtain the prediction results.
[0034] In a further embodiment, it also includes:
[0035] Model building module, used to build training models;
[0036] The second acquisition module is used to acquire historical passenger data;
[0037] A feature extraction module is used to analyze passenger data to obtain historical feature data reflecting passenger behavior characteristics; wherein the historical feature data includes: historical flight records, historical behavior logs, and historical social network data;
[0038] A data partitioning module is used to divide historical feature data into a training set and a validation set;
[0039] The training module is used to train the training model using the test set to obtain a training model that passes the training;
[0040] The verification module uses the verification set to verify the trained model and obtain the passenger behavior prediction model.
[0041] In a further embodiment, it also includes:
[0042] The optimization module is used to optimize the passenger behavior prediction model before using the passenger behavior prediction model to predict real-time passenger data and obtaining the prediction result.
[0043] The present invention also proposes a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the civil aviation passenger behavior prediction and analysis method as described in any one of the above.
[0044] The present invention also proposes an electronic device, comprising: a processor and a memory, the memory being used to store one or more programs; when the one or more programs are executed by the processor, the civil aviation passenger behavior prediction and analysis method as described in any one of the above is implemented.
[0045] In the present invention, the proposed civil aviation passenger behavior prediction and analysis method, system, medium and equipment predict real-time passenger data through a passenger behavior prediction model to obtain prediction results, which is conducive to subsequent personalized service recommendations, marketing strategy formulation and operational decision support based on the prediction results. It can provide a scientific basis for the airline's market strategy and customer service, enhance passengers' travel experience, improve satisfaction and loyalty, optimize flight operation management, improve resource allocation efficiency, reduce costs and increase profits. BRIEF DESCRIPTION OF THE DRAWINGS
[0046] Figure 1 The present invention is a flowchart of a method for predicting and analyzing civil aviation passenger behavior. DETAILED DESCRIPTION
[0047] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.
[0048] Reference Figure 1 The present invention proposes a method for predicting and analyzing civil aviation passenger behavior, comprising:
[0049] Get real-time passenger data;
[0050] Use the passenger behavior prediction model to predict real-time passenger data and obtain prediction results.
[0051] The present invention predicts real-time passenger data through a passenger behavior prediction model to obtain prediction results; this is conducive to subsequent personalized service recommendations, marketing strategy formulation and operational decision support based on the prediction results, and can provide a scientific basis for the airline's market strategy and customer service, enhance passengers' travel experience, improve satisfaction and loyalty, optimize flight operation management, improve resource allocation efficiency, reduce costs and increase profits.
[0052] In order to improve the accuracy of passenger behavior prediction and provide a scientific basis for the airline's marketing strategy and customer service, in this embodiment, before using the passenger behavior prediction model to predict real-time passenger data and obtaining the prediction result, the following steps are also included:
[0053] Build a training model;
[0054] Get historical passenger data;
[0055] Analyze the passenger data to obtain historical characteristic data reflecting the passenger behavior characteristics; the historical characteristic data includes: historical flight records, historical behavior logs and historical social network data;
[0056] Divide historical feature data into training set and validation set;
[0057] Use the test set to train the training model and obtain a training model that passes the training;
[0058] The validation set is used to validate the trained model and obtain the passenger behavior prediction model.
[0059] Among them, historical passenger data includes: personal information, historical flight records, service usage, loyalty program participation, online behavior logs, customer feedback, airport traffic monitoring and social network data.
[0060] It should be noted that the acquisition of historical passenger data and real-time passenger data in this embodiment is based on data sharing agreements established with the airline's PSS, CRM, airport information system (AGIS), global distribution system (GDS), and partners such as hotels and car rental service providers.
[0061] In this embodiment, after obtaining the historical passenger data, the following steps are also included:
[0062] Clean historical passenger data to standardize data formats, handle missing values, identify and handle abnormal or erroneous data, and ensure data quality.
[0063] Among them, historical flight records include but are not limited to [passenger ID, departure place (Airport_O), destination (Airport_D), departure date and time (Depart_DT), arrival date and time (Arrive_DT), flight number (Flight_ID), cabin class (Seat_Class), fare (Ticket_Price)], airport operation data, such as [airport ID, airport location (Geo_Location), passenger volume (Passenger_Volume), flight takeoff and landing volume (Flight_Volume)], and macroeconomic indicators, such as regional GDP, tourism seasonality index, etc.
[0064] The historical behavior log includes [passenger ID, behavior type (Action_Type), timestamp (Timestamp), behavior duration (Duration), behavior result (Action_Result)] as well as the passenger's personal preferences and feedback, such as seat preferences, meal selections, and frequent flyer program interaction records.
[0065] Among them, historical social network data includes [traveler ID, connected traveler ID (Connected_ID), relationship type (Relationship_Type), interaction frequency (Interaction_Frequency), influence score (Influence_Score)] and group behavior data on social media, such as topic participation, community activity records, etc.
[0066] In a further embodiment, obtaining historical passenger data specifically includes:
[0067] Historical passenger data is obtained according to a preset period, so that the training model can be retrained and verified according to the historical passenger data at the preset period, thereby realizing a dynamic update mechanism of the passenger behavior prediction model.
[0068] Of course, in other embodiments, the training model may also be retrained when significant changes occur in the historical passenger data.
[0069] In order to accurately obtain historical characteristic data reflecting the behavior characteristics of passengers, in a further embodiment, the passenger data is analyzed to obtain historical characteristic data reflecting the behavior characteristics of passengers, specifically including:
[0070] Passenger data is analyzed based on their travel habits, preferences, consumption behaviors, social network data, behavioral indicators, and timeliness characteristics to obtain historical characteristic data reflecting passenger behavior characteristics.
[0071] Among them, behavioral indicators include average booking time, frequent flight routes, preferred cabin class, and frequency of service use; timeliness characteristics include seasonal travel patterns, holiday travel peaks, and emergency bookings.
[0072] In this embodiment, the passenger behavior prediction model includes a spatiotemporal analysis model, a passenger behavior sequence prediction model and a passenger social network model; wherein the spatiotemporal analysis model includes a CNN model, the passenger behavior sequence prediction model includes an RNN and LSTM model, and the passenger social network model includes a GNN model.
[0073] In specific implementation, the CNN model in the spatiotemporal analysis model is used to extract features from the spatiotemporal data based on historical flight records, so as to train the CNN model in the spatiotemporal analysis model to identify flight patterns; the RNN and LSTM in the passenger behavior sequence prediction model are used to model the passenger behavior sequence based on historical behavior logs, and a passenger behavior sequence prediction model is obtained to predict the future behavior of passengers; and the GNN in the passenger social network model is used to analyze the historical social network data of passengers to identify key influencers and group behavior trends.
[0074] The above models are explained as follows:
[0075] The spatiotemporal analysis model is defined as a spatiotemporal feature extraction function Φ(·), which is used to extract the spatiotemporal features of passenger behavior from the original data. Its convolution operation can be expressed as:
[0076] F = CNN(Φ(D_spacetime), K), where F is the feature map, D_spacetime is the preprocessed and feature-engineered spatiotemporal dataset, and K is a set of convolution kernels used to capture the spatial distribution and temporal regularity of passenger flight patterns;
[0077] The passenger behavior sequence prediction model can be expressed as: Prediction = LSTM (Sequence_D, θ), where Sequence_D is the passenger's behavior sequence, θ is the model parameter, and Prediction is the prediction of future behavior;
[0078] The passenger social network model can be implemented through S=GNN(G,X), where G is the social network graph, X is the node feature matrix, and S is the social network embedding after GNN processing, which can capture the passenger's position and influence in the social network.
[0079] In this embodiment, the training model that has passed the training is verified using the verification set to obtain a passenger behavior prediction model, which specifically includes:
[0080] Using the validation set to validate multiple evaluation indicators of the training model that has passed the training, the multiple evaluation indicators are used to indicate the effect of the training model;
[0081] The training model in which the multiple assessment indicators all meet the corresponding preset assessment indicators is determined as the passenger behavior prediction model.
[0082] Among them, the evaluation indicators include accuracy, recall, F1 score and ROC-AUC (Receiver Operating Characteristic Area Under the Curve) curve.
[0083] Accuracy is the most intuitive performance indicator, indicating the proportion of samples correctly predicted by the model to the total number of samples. The calculation formula for accuracy is:
[0084] Where TP (True Positive) is the number of samples correctly predicted as positive; TN (True Negative) is the number of samples correctly predicted as negative; FP (False Positive) is the number of samples incorrectly predicted as positive; FN (False Negative) is the number of samples incorrectly predicted as negative.
[0085] The recall rate indicates the proportion of positive samples captured by the passenger behavior prediction model to all actual positive samples, reflecting the completeness of the model. The calculation formula for the recall rate is:
[0086] The F1 score is the harmonic mean of precision and recall, which strikes a balance between the two and is more effective in cases of class imbalance. The calculation formula for the F1 score is Among them, Precision represents accuracy, and the calculation formula of accuracy is
[0087]
[0088] The ROC curve is a curve drawn by plotting all possible values between the true positive rate (recall rate) and the false positive rate (1-specificity) when the classification threshold of the model changes from 0 to 1. AUC represents the area under the ROC curve, which measures the model's ability to classify samples.
[0089] The calculation of AUC usually involves numerical integration of the area under the ROC curve, which can be calculated using existing numerical methods or software libraries. For binary classification problems, the TPR (recall rate) and FPR (false positive rate = 1-specificity) at different thresholds can be calculated by changing the classification threshold, and then these points can be used to draw the ROC curve. For multi-classification problems, a one-vs-all method can be used to draw the ROC curve for each category, and then the AUC of each curve can be calculated, and finally the average or other statistical methods can be used to comprehensively evaluate the model performance.
[0090] In this embodiment, after obtaining the real-time passenger data, the following steps are also included:
[0091] Stream data processing technology is used to process real-time passenger data in real time, so as to facilitate subsequent prediction using passenger behavior prediction models.
[0092] Specifically, the stream data processing technology includes Apache Kafka and / or Apache Flink stream processing technology.
[0093] In this embodiment, before using the passenger behavior prediction model to predict the real-time passenger data to obtain the prediction result, the following steps are also included:
[0094] The passenger behavior prediction model is optimized to further improve the prediction accuracy of the passenger behavior prediction model.
[0095] In a further embodiment, the passenger behavior prediction model is optimized, specifically including:
[0096] Reinforcement learning algorithm is used to optimize the passenger behavior prediction model.
[0097] It is important to know that reinforcement learning algorithm is a machine learning method that learns how to make decisions by interacting with the environment. In the context of passenger behavior prediction, reinforcement learning can help passenger behavior prediction models learn how to make optimal predictions among different passenger behavior patterns.
[0098] In one specific embodiment, a reinforcement learning algorithm is used to optimize the passenger behavior prediction model, specifically including: using a Q-Learning algorithm to optimize the passenger behavior prediction model.
[0099] In another specific embodiment, a reinforcement learning algorithm is used to optimize the passenger behavior prediction model, specifically including:
[0100] The deep reinforcement learning (DRL) algorithm is used to optimize the passenger behavior prediction model in order to maximize its long-term benefits.
[0101] In this embodiment, after predicting the real-time passenger data using the passenger behavior prediction model and obtaining the prediction result, the following steps are further included:
[0102] Based on the prediction results, one or more of personalized recommendations, marketing strategy formulation, operational decision support, and customer relationship management maintenance are carried out.
[0103] Among them, personalized recommendations are made based on the prediction results, including:
[0104] Based on the prediction results, personalized flight and additional service recommendations are pushed to passengers through the airline’s marketing platform.
[0105] Among them, marketing strategies are formulated based on the prediction results, including:
[0106] Based on the forecast results, analyze passengers’ price sensitivity and develop dynamic pricing strategies for different passenger groups.
[0107] Among them, operational decision support is provided based on the forecast results, including:
[0108] Based on the prediction results, optimize flight operational decisions such as seat allocation, crew arrangement, and onboard inventory.
[0109] Among them, customer relationship management and maintenance are carried out according to the prediction results, including:
[0110] Based on the prediction results, passengers with different customer relationship levels are identified and customized services are provided for them.
[0111] In this embodiment, after performing one or more of personalized recommendation, marketing strategy formulation, operational decision support, and customer relationship management maintenance according to the prediction results, the following is further included:
[0112] Gather feedback on the accuracy of forecast results through customer interactions, service usage, and market feedback;
[0113] Based on the accuracy feedback collected, the passenger behavior prediction model is iteratively updated regularly to adapt to changes in passenger behavior.
[0114] Of course, a continuous improvement mechanism is also established in this embodiment to continuously optimize each link of data collection, feature extraction, model construction and result application.
[0115] The present invention also proposes a civil aviation passenger behavior prediction and analysis system, comprising:
[0116] A first acquisition module is used to acquire real-time passenger data;
[0117] The prediction module is used to use the passenger behavior prediction model to predict the real-time passenger data and obtain the prediction results.
[0118] In a further embodiment, it also includes:
[0119] Model building module, used to build training models;
[0120] The second acquisition module is used to acquire historical passenger data;
[0121] A feature extraction module is used to analyze passenger data to obtain historical feature data reflecting passenger behavior characteristics; wherein the historical feature data includes: historical flight records, historical behavior logs, and historical social network data;
[0122] A data partitioning module is used to divide historical feature data into a training set and a validation set;
[0123] The training module is used to train the training model using the test set to obtain a training model that passes the training;
[0124] The verification module uses the verification set to verify the trained model and obtain the passenger behavior prediction model.
[0125] In a further embodiment, it also includes:
[0126] The optimization module is used to optimize the passenger behavior prediction model before using the passenger behavior prediction model to predict real-time passenger data and obtaining the prediction result.
[0127] The present invention also proposes a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the civil aviation passenger behavior prediction and analysis method as described in any one of the above.
[0128] The present invention also proposes an electronic device, comprising: a processor and a memory, the memory being used to store one or more programs; when the one or more programs are executed by the processor, the civil aviation passenger behavior prediction and analysis method as described in any one of the above is implemented.
[0129] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.
Claims
1. A method for predicting and analyzing civil aviation passenger behavior, characterized in that: include: Get real-time passenger data; Use the passenger behavior prediction model to predict real-time passenger data and obtain prediction results.
2. The civil aviation passenger behavior prediction and analysis method according to claim 1, characterized in that: Before using the passenger behavior prediction model to predict real-time passenger data and obtain the prediction results, it also includes: Build training model; Get historical passenger data; Analyze the passenger data to obtain historical characteristic data reflecting the passenger behavior characteristics; the historical characteristic data includes: historical flight records, historical behavior logs and historical social network data; Divide historical feature data into training set and validation set; Use the test set to train the training model and obtain a training model that passes the training; Use the validation set to validate the trained model and obtain the passenger behavior prediction model; Preferably, obtaining historical passenger data specifically includes: Obtain historical passenger data according to the preset period.
3. The civil aviation passenger behavior prediction and analysis method according to claim 2, characterized in that: Use the validation set to validate the trained model and obtain the passenger behavior prediction model, including: Using the validation set to validate multiple evaluation indicators of the training model that has passed the training, the multiple evaluation indicators are used to indicate the effect of the training model; The training model in which the multiple assessment indicators all meet the corresponding preset assessment indicators is determined as the passenger behavior prediction model.
4. The method for predicting and analyzing civil aviation passenger behavior according to claim 1, characterized in that: The passenger behavior prediction model includes a spatiotemporal analysis model, a passenger behavior sequence prediction model and a passenger social network model; among them, the spatiotemporal analysis model includes a CNN model, the passenger behavior sequence prediction model includes an RNN and LSTM model, and the passenger social network model includes a GNN model.
5. The method for predicting and analyzing civil aviation passenger behavior according to claim 1, characterized in that: Before using the passenger behavior prediction model to predict real-time passenger data and obtain the prediction results, it also includes: Optimize passenger behavior prediction models; Preferably, the passenger behavior prediction model is optimized, specifically including: Reinforcement learning algorithm is used to optimize the passenger behavior prediction model.
6. The method for predicting and analyzing civil aviation passenger behavior according to claim 1, characterized in that: After using the passenger behavior prediction model to predict real-time passenger data and obtain the prediction results, it also includes: Based on the prediction results, one or more of personalized recommendations, marketing strategy formulation, operational decision support, and customer relationship management maintenance are performed; Preferably, after performing one or more of personalized recommendation, marketing strategy formulation, operational decision support and customer relationship management maintenance according to the prediction results, the method further includes: Gather feedback on the accuracy of forecast results through customer interactions, service usage, and market feedback; Based on the accuracy feedback collected, the passenger behavior prediction model is iteratively updated regularly to adapt to changes in passenger behavior.
7. A civil aviation passenger behavior prediction and analysis system, characterized in that: include: A first acquisition module is used to acquire real-time passenger data; The prediction module is used to use the passenger behavior prediction model to predict the real-time passenger data and obtain the prediction results.
8. The civil aviation passenger behavior prediction and analysis system according to claim 7, characterized in that: Also includes: Model building module, used to build training models; The second acquisition module is used to acquire historical passenger data; A feature extraction module is used to analyze passenger data to obtain historical feature data reflecting passenger behavior characteristics; wherein the historical feature data includes: historical flight records, historical behavior logs, and historical social network data; A data partitioning module is used to divide historical feature data into a training set and a validation set; The training module is used to train the training model using the test set to obtain a training model that passes the training; The verification module uses the verification set to verify the trained model and obtain the passenger behavior prediction model. Preferably, it also includes: The optimization module is used to optimize the passenger behavior prediction model before using the passenger behavior prediction model to predict real-time passenger data and obtaining the prediction result.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the civil aviation passenger behavior prediction and analysis method as described in any one of claims 1 to 7 is implemented.
10. An electronic device, characterized in that: include: A processor and a memory, the memory being used to store one or more programs; When one or more programs are executed by a processor, the civil aviation passenger behavior prediction and analysis method as described in any one of claims 1 to 7 is implemented.
Citation Information
Cited By
Large-scale competition civil aviation transportation flow prediction model and method thereof
CN120046804A
High-speed rail sliding plug door safety control method and device integrated with TOF camera
CN121853885A