Vehicle commuting behavior label generation method and system based on travel data
Through the vehicle commuting behavior label generation method based on itinerary data, cluster analysis and preset models are used to identify the car commuting behavior, which solves the problems of untrue data and inaccurate identification results in the existing technology, and achieves more efficient and accurate commuting Behavioral analysis.
Patent Information
- Application Number
- CN202411497261.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-25
- Publication Date
- 2025-02-14
AI Technical Summary
The prior art has problems such as untrue, unstable data and inaccurate identification results when identifying and analyzing car commuting behaviors. In particular, the user behavior restriction and rule setting are found in the methods based on shared electric motorcycles, public bicycles and single machine learning algorithms. Dependency problem.
The vehicle commuter behavior label generation method based on itinerary data is used to identify the vehicle travel data through vehicle identification code, and the vehicle frequent destination and commuter use label are determined using cluster analysis and preset vehicle commuter use model, and the vehicle frequent destination and commuter use label are analyzed and statistics through inverse address Analysis determines commuting time periods and commuting hours.
A more realistic and stable vehicle trip data analysis is achieved, the accuracy and reliability of commuter behavior labels are improved, the dependence on user interference and prior knowledge is reduced, and parameters are dynamically adaptively adjusted to adapt to different travel frequencies.
Smart Images
Figure CN119441911A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of methods for calculating commuting behavior data of automobile users, and in particular to a method and system for generating vehicle commuting behavior labels based on travel data. Background Art
[0002] Commuting behavior occurs when the residence and the workplace are not in the same place. It takes a certain amount of time to go to and from get off work, and generally takes walking or using public transportation to the destination. The existing commuting behavior identification methods include: commuting behavior identification based on shared electric motorcycle borrowing and returning data and public bicycle card swiping data; mining commuting behavior based on set statistical classification rules; mining commuting behavior based on a single machine learning algorithm. However, the above methods have certain disadvantages. For example, the users of shared electric motorcycles and public bicycles are not fixed, and the commuting habits of individual users cannot be reflected; mining commuting behavior based on set rules, the final result is highly correlated with the set rules, and rich and accurate prior knowledge is required to set the rules, and the results obtained are not accurate enough. On the other hand, the commuting behavior of users is mined based on user Internet positioning, but it is highly affected by the user's personal behavior. When the user turns off GPS positioning, the corresponding information cannot be obtained. In order to make up for the limitations and instability of a single rule, the results of multiple rules are aggregated, but rich prior knowledge is still required to complete the rule formulation.
[0003] Among the commuting population, more people choose to commute by car. Applicants are more concerned about the behavioral habits of using cars for commuting. They analyze the vehicle commuting habits from a large amount of travel data, extract relevant information as the commuting habit label of the vehicle, and establish a vehicle commuting behavior label system, thereby providing reference and basis for subsequent vehicle analysis or market applications. Summary of the invention
[0004] To solve at least one aspect of the above problems, the present invention provides a method for generating a vehicle commuting behavior label based on travel data, comprising: identifying the vehicle travel data based on a vehicle identification code to determine the corresponding vehicle, wherein the vehicle travel data includes a power-on signal, a power-off signal, a driving time, a driving mileage, a starting point information, and an end point information; determining the vehicle's frequently visited location label based on the vehicle travel data by a cluster analysis method; determining the commuting vehicle based on the vehicle travel data according to a preset vehicle commuting purpose model, and adding a commuting purpose label; parsing the vehicle's frequently visited location label corresponding to the commuting purpose label using a reverse address resolution interface to determine the home address information and company address information of the commuting vehicle; performing statistical analysis on the driving time of the commuting vehicle corresponding to the vehicle commuting purpose label, and determining the time period with the highest frequency as the commuting time period, and determining the working hours and off-get off work hours corresponding to the commuting time period according to the home address information and company address information of the commuting vehicle.
[0005] Preferably, the step of determining the vehicle location label based on the vehicle travel data by using a cluster analysis method also includes: clustering the vehicle start and end points based on the starting point information and the end point information of the vehicle travel data by using an unsupervised density clustering method, and deleting discrete points; performing secondary clustering on the vehicle travel data after deleting the discrete points based on the KMeans algorithm, and determining the number of clusters by using the silhouette coefficient method; sorting the clusters of each vehicle start and end point, and adding labels of the locations frequently visited by the vehicle.
[0006] Preferably, the step of adding a commuting label based on user travel data according to a preset vehicle commuting use model also includes: extracting feature values based on the vehicle travel data to generate commuting use samples, the feature values including the vehicle's travel rate, average daily trip number, and average daily mileage; using a random forest classifier to perform training based on the extracted commuting use samples to generate a vehicle commuting use model; using the vehicle commuting use model to identify commuting vehicles based on vehicle travel data and add commuting use labels.
[0007] Preferably, it also includes: extracting features based on vehicle travel data to generate household car usage samples, using a random forest classifier to train the extracted household car usage samples to generate a household car usage model, identifying household cars based on the vehicle travel data through the household car usage model and adding household car usage labels; extracting features based on vehicle travel data to generate online car-hailing usage samples, using a random forest classifier to train the extracted online car-hailing usage samples to generate an online car-hailing usage model, identifying online car-hailing based on the vehicle travel data through the online car-hailing usage model and adding online car-hailing usage labels.
[0008] On the other hand, a vehicle commuting behavior label generation system based on travel data is provided, including: a data acquisition module, the data acquisition module is used to connect to the user's vehicle and receive the vehicle travel data, the data acquisition module identifies the vehicle travel data based on the vehicle identification code and determines the corresponding vehicle, the vehicle travel data includes a power-on signal, a power-off signal, a driving time, a driving mileage, a starting point information and an end point information; a location label adding module, the location label adding module uses a cluster analysis method based on the vehicle travel data to determine the vehicle's frequently visited location label; a usage label adding module adds a usage label according to a preset vehicle. A commuting purpose model determines a commuting vehicle based on vehicle travel data and adds a commuting purpose label; a commuting address parsing module, which uses a reverse address parsing interface to parse the vehicle frequented location label of the commuting vehicle corresponding to the commuting purpose label, and determines the home address information and company address information of the commuting vehicle; a commuting time calculation module, which performs statistical analysis on the driving time of the commuting vehicle corresponding to the vehicle commuting purpose label, and determines the time period with the highest frequency as the commuting time period, and determines the working time and off-get off work time corresponding to the commuting time period according to the home address information and company address information of the commuting vehicle.
[0009] Preferably, the step of determining the vehicle location label by clustering analysis based on the user travel data by the location label adding module also includes: clustering the vehicle starting and ending points based on the starting point information and the ending point information of the vehicle travel data by using an unsupervised density clustering method, and deleting discrete points; performing secondary clustering on the vehicle travel data after deleting the discrete points based on the KMeans algorithm, and determining the number of clusters by using the silhouette coefficient method; sorting the clusters of each vehicle starting and ending point, and adding vehicle frequented location labels.
[0010] Preferably, the step of adding commuting labels based on user travel data according to a preset vehicle commuting use model by the usage label adding module also includes: extracting feature values based on vehicle travel data to generate commuting use samples, wherein the feature values include the vehicle's travel rate, average daily trip number, and average daily mileage; training a random forest classifier based on the extracted commuting use samples to generate a vehicle commuting use model; and identifying commuting vehicles based on vehicle travel data using the vehicle commuting use model and adding commuting use labels.
[0011] Preferably, the usage label adding module also includes: extracting feature values based on vehicle travel data to generate household car usage samples and online car-hailing usage samples, using a random forest classifier to train based on the extracted household car usage samples and online car-hailing usage samples to generate a household car usage model and an online car-hailing usage model, identifying household cars based on vehicle travel data through the household car usage model and the online car-hailing usage model and adding household car usage labels and online car-hailing usage labels.
[0012] The method and system for generating vehicle commuting behavior labels based on travel data of the embodiments of the present invention have the following beneficial effects: directly using more real and stable vehicle travel data, the vehicle data can be automatically acquired based on vehicle characteristics such as starting and shutting down, which is more real and stable and less affected by user interference; the data is dynamic data, and the commuting operation data is calculated in combination with mileage information, and the calculation is more accurate; for labels that are easy to formulate rules, the rules are directly used for calculation, such as commuting time, etc.; for scenarios where rules are difficult to formulate or are not accurate enough, algorithm calculation is used, such as the places frequently visited by the vehicle and the purpose of the vehicle, etc.; when using the algorithm model to calculate the label, a variety of model methods are combined; in order to meet the needs of different travel frequency data, the parameters are not set manually directly, but the parameters are learned in a dynamic and adaptive manner. BRIEF DESCRIPTION OF THE DRAWINGS
[0013] In order to better understand the above and other purposes, features, advantages and functions of the present invention, reference may be made to the embodiments shown in the accompanying drawings. The same reference numerals in the accompanying drawings refer to the same components. It should be understood by those skilled in the art that the accompanying drawings are intended to schematically illustrate the preferred embodiments of the present invention and have no limiting effect on the scope of the present invention, and the components in the drawings are not drawn to scale.
[0014] Figure 1 A schematic flow chart of a method for generating vehicle commuting behavior labels based on travel data according to an embodiment of the present invention is shown. DETAILED DESCRIPTION
[0015] The following is a description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, it should be recognized by those of ordinary skill in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.
[0016] As used herein, the term "including" and its variations mean open inclusion, i.e., "including but not limited to". Unless otherwise stated, the term "or" means "and / or". The term "based on" means "based at least in part on". The terms "an example embodiment" and "an embodiment" mean "at least one example embodiment". The term "another embodiment" means "at least one additional embodiment". The terms "first", "second", etc. may refer to different or the same objects. Other explicit and implicit definitions may also be included below.
[0017] In order to at least partially solve one or more of the above problems and other potential problems, an embodiment of the present disclosure proposes a method for generating a vehicle commuting behavior label based on trip data, including:
[0018] Step S1, identifying the vehicle travel data based on the vehicle identification code to determine the corresponding vehicle, the vehicle travel data includes a power-on signal, a power-off signal, driving time, driving mileage, starting point information and end point information.
[0019] Specifically, the vehicle identification code is a vehicle VIN code, which is used to uniquely identify the vehicle. Since the connected user vehicles and the acquired vehicle travel data are very large, the vehicle travel data belonging to the same vehicle can be identified by the vehicle identification code. In this method, the user vehicle is connected through the cloud server, and the vehicle travel data of the user vehicle is uploaded to the cloud server at a preset time period.
[0020] Step S2, determining the vehicle's frequently visited location tags using a cluster analysis method based on the vehicle travel data.
[0021] Specifically, the vehicle travel data corresponding to each vehicle is determined according to the vehicle identification code, and then the vehicle travel data of each vehicle is clustered to determine the corresponding vehicle frequented location labels. The starting point information and the end point information in the vehicle travel data include longitude and latitude. The starting and ending points of the vehicle are clustered through cluster analysis to obtain multiple vehicle frequented location clusters, and vehicle frequented location labels are added to each vehicle frequented location cluster.
[0022] In some embodiments, step S2 further includes:
[0023] Step S201 : based on the starting point information and the end point information of the vehicle travel data, the vehicle starting and ending points are clustered by using an unsupervised density clustering method, and the discrete points are deleted.
[0024] Specifically, clustering is performed based on the latitude and longitude of the starting point information and the end point information in the vehicle travel data. Since vehicles may have scattered stops, the unsupervised density clustering method (Density-Based Spatial Clustering of Applications with Noise, referred to as DBSCAN) is used to cluster the vehicle start and end points, and the noise points are removed by deleting discrete points.
[0025] Step S202, performing secondary clustering on the vehicle travel data after the discrete points are deleted based on the KMeans algorithm, and using the silhouette coefficient method to determine the number of clusters.
[0026] Specifically, the unsupervised density clustering method may classify long distances into one cluster. Therefore, the vehicle travel data after deleting discrete points is clustered again based on the KMeans algorithm to further reduce the size of the cluster. The silhouette coefficient method is used to evaluate the clustering results and determine the optimal number of clusters.
[0027] In some embodiments, it also includes setting an intra-class distance threshold. When the average intra-class distance in each cluster obtained by secondary clustering based on the KMeans algorithm is greater than the intra-class distance threshold, the cluster is deleted to limit the intra-cluster distance.
[0028] Step S203, sorting the starting and ending point clusters of each vehicle, and adding labels of the locations frequently visited by the vehicles.
[0029] Specifically, each vehicle starting and ending point cluster corresponding to the optimal cluster number is sorted, for example, based on the number of starting and ending points in each vehicle starting and ending point cluster, a vehicle frequented location label is added to each vehicle starting and ending point cluster in sequence according to the number. In some embodiments, the vehicle frequented location label also includes clustering order information.
[0030] In other embodiments, the clustering analysis method may also be a single KMeans algorithm, an unsupervised purpose-based clustering algorithm, or a combination thereof with other clustering algorithms.
[0031] Step S3, determining a commuting vehicle based on the vehicle travel data according to a preset vehicle commuting use model, and adding a commuting use label.
[0032] Specifically, the preset vehicle commuting use model is a model generated by training based on the travel characteristics of commuting vehicles. The vehicle travel data of each vehicle is identified through the vehicle commuting use model to determine the commuting vehicle, and a commuting use label is added to the identified commuting vehicle.
[0033] In some embodiments, step S3 further includes:
[0034] Step S301, extracting feature values based on vehicle travel data to generate commuting usage samples, wherein the feature values include the vehicle's travel rate, average daily number of trips, and average daily mileage.
[0035] Specifically, the online buried data will judge a single trip based on the power-on and power-off signals in the vehicle travel data, and record the time, mileage, fuel volume, starting point and end point of each trip. This information is obtained from the vehicle travel data database, and distributed computing is performed with the help of Flink to calculate the characteristic values of each vehicle, including the vehicle's travel rate, average daily number of trips, average daily mileage, etc. Select multiple commuting vehicles, and use the samples composed of the characteristic values of multiple commuting vehicles as commuting use samples.
[0036] Step S302: A random forest classifier is used to generate a vehicle commuting usage model through training based on the extracted commuting usage samples.
[0037] Specifically, the random forest classifier is also very tolerant to outliers and noise, and has better prediction and classification performance than the decision tree. The random forest classifier is used to generate a vehicle commuting usage model based on commuting usage samples.
[0038] Step S303: using a vehicle commuting purpose model to identify commuting vehicles based on vehicle travel data and adding commuting purpose labels.
[0039] Specifically, feature values are extracted from the acquired vehicle travel data, and then a vehicle travel purpose model is used to identify the vehicle based on the vehicle feature values, and a commuting purpose label is set for the identified commuting vehicle.
[0040] In some embodiments, step S3 also includes: extracting features based on vehicle travel data to generate household car usage samples, using a random forest classifier to train the extracted household car usage samples to generate a household car usage model, identifying household cars based on the vehicle travel data through the household car usage model and adding household car usage labels; extracting features based on vehicle travel data to generate online car-hailing usage samples, using a random forest classifier to train the extracted online car-hailing usage samples to generate an online car-hailing usage model, identifying online car-hailing based on the vehicle travel data through the online car-hailing usage model and adding online car-hailing usage labels.
[0041] Specifically, the vehicle trip data of family cars is manually selected, and the trip rate, average daily trip number, average daily mileage and other feature values are extracted to generate family car usage samples, and then a family car usage model is generated based on the family car usage samples, and the family car usage model is further used to add a family car usage label to the identified family cars. The same is true for online car-hailing.
[0042] Step S4, using a reverse address resolution interface to resolve the vehicle frequented location label of the commuting vehicle corresponding to the commuting purpose label, and determine the home address information and company address information of the commuting vehicle.
[0043] Specifically, based on the obtained vehicle's frequently visited location tags and combined with the open source reverse address resolution interface, the location types and POI areas corresponding to all locations are analyzed to obtain the vehicle's home address information. At the same time, combined with the vehicle's commuting purpose tags, the company address information of commuter-type vehicles is analyzed to obtain.
[0044] In another example, based on the analysis and inference of home and company addresses without labels, directly using reverse address resolution to obtain the POI type of the location will result in a more accurate home address while reducing the computational cost.
[0045] Step S5, performing statistical analysis on the driving time of the commuting vehicle corresponding to the vehicle commuting purpose label, and determining the time period with the highest frequency as the commuting time period, and determining the working hours and off-get off work hours corresponding to the commuting time period based on the home address information and company address information of the commuting vehicle.
[0046] Specifically, based on the statistical analysis method, the commuting vehicle travel time periods are counted, and the time periods with the highest frequency are selected as the commuting time periods. At the same time, the commuting time of the vehicle is obtained by combining the information such as the home and company addresses. For example, the commuting time can be determined based on the time corresponding to the commuting vehicle's home address and the time when it is at the company address during the commuting time period.
[0047] The above-mentioned vehicle commuting behavior labels include vehicle commuting purpose labels, home address information, company address information, and commuting time. Commuting vehicles are identified and marked through vehicle commuting behavior labels to form a vehicle commuting behavior label system, so as to facilitate targeted improvements and optimizations to vehicles and / or vehicle networking systems based on vehicle commuting needs.
[0048] On the other hand, a vehicle commuting behavior label generation system based on travel data is provided, including: a data acquisition module, the data acquisition module is used to connect to the user's vehicle and receive the vehicle travel data, the data acquisition module identifies the vehicle travel data based on the vehicle identification code, and determines the corresponding vehicle, the vehicle travel data includes a power-on signal, a power-off signal, a driving time, a driving mileage, a starting point information and an end point information; a location label adding module, the location label adding module determines the vehicle's frequent location label based on the vehicle travel data by a cluster analysis method; a purpose label adding module, the purpose label adding module determines the commuting vehicle based on the vehicle travel data according to a preset vehicle commuting purpose model, and adds a commuting purpose label; a commuting address parsing module, the commuting address parsing module uses a reverse address parsing interface to parse the vehicle's frequent location label corresponding to the commuting purpose label of the commuting vehicle, and determines the home address information and company address information of the commuting vehicle; a commuting time calculation module, the commuting time calculation module performs statistical analysis on the driving time of the commuting vehicle corresponding to the vehicle commuting purpose label, and determines the time period with the highest frequency as the commuting time period, and determines the working time and off-work time corresponding to the commuting time period according to the home address information and company address information of the commuting vehicle.
[0049] Specifically, the data acquisition module is set in the cloud server, and the data acquisition module includes a communication unit, a processor and a memory. The data acquisition module receives the vehicle travel data through a communication connection with the user vehicle. The processor determines the user vehicle corresponding to the vehicle travel data through the vehicle identification code. The memory is used to store the vehicle travel data.
[0050] The location tag adding module, the purpose tag module, the commuting address parsing module, and the commuting time calculation module are multiple processor units set in the same terminal. The location tag adding module and the purpose tag module are respectively connected to the data acquisition module to receive vehicle travel data. The commuting address parsing module is respectively connected to the location tag adding module and the purpose tag module, and obtains the location tag information of the vehicle through the location tag adding module, and obtains the purpose tag information of the vehicle through the purpose tag module. The commuting address parsing module is used to determine the home address information and company address information corresponding to each commuting vehicle.
[0051] The commuting time calculation module obtains the vehicle travel data corresponding to the communication vehicle tag by connecting to the commuting purpose tag module, further determines the commuting period based on the time statistics, and obtains the commuting purpose tag vehicle information and its corresponding home address information and company address information through the purpose tag module and the commuting address parsing module, and determines the commuting time by combining the time corresponding to the commuting vehicle home address and company address. A vehicle commuting behavior tag system is generated that includes vehicle commuting purpose tags, home address information, company address information, commuting time and other vehicle commuting behavior tags.
[0052] In some embodiments, the step of determining the vehicle location label by the location label adding module through cluster analysis based on the user travel data also includes: clustering the vehicle starting and ending points based on the starting point information and the ending point information of the vehicle travel data using an unsupervised density clustering method, and deleting discrete points; performing secondary clustering on the vehicle travel data after deleting the discrete points based on the KMeans algorithm, and determining the number of clusters using the silhouette coefficient method; sorting the clusters of each vehicle starting and ending point, and adding labels for the locations frequently visited by the vehicle.
[0053] Specifically, the location label adding module also includes a first clustering unit, a second clustering unit and a label unit. The first clustering unit is used to execute an unsupervised density clustering method to cluster the vehicle starting and ending points and delete discrete points; the second clustering unit receives the vehicle travel data with discrete points deleted through the first clustering unit, and uses the KMeans algorithm for secondary clustering, and uses the silhouette coefficient method to determine the number of clusters, and outputs the vehicle starting and ending point clusters; the label unit responds to the vehicle starting and ending point clusters output by the second clustering unit to add vehicle frequently visited location labels.
[0054] In some embodiments, the step of adding commuting labels based on the user travel data according to the preset vehicle commuting use model by the use label adding module also includes: extracting features based on the vehicle travel data to generate commuting use samples, the feature values including the vehicle's travel rate, average daily trips, and average daily mileage; training based on the extracted commuting use samples to generate a vehicle commuting use model using a random forest classifier; identifying commuting vehicles based on the vehicle travel data using the vehicle commuting use model and adding commuting use labels. The use label adding module also includes: extracting features based on the vehicle travel data to generate household car use samples and online car-hailing use samples, training based on the extracted household car use samples and online car-hailing use samples using a random forest classifier to generate household car use models and online car-hailing use models, identifying household cars based on the vehicle travel data using the household car use model and online car-hailing use model and adding household car use labels and online car-hailing use labels.
[0055] Specifically, the usage label adding module includes a receiving unit, a model generating unit and a label adding unit. The receiving unit is used to receive commuting usage samples, household car usage samples and online car-hailing usage samples; the model generating unit is used to train and generate corresponding vehicle commuting usage models, household car usage models and online car-hailing usage models based on the received samples using a random forest classifier; the label unit is used to identify the vehicle age and travel data according to each trained model and add corresponding usage labels.
[0056] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or technical improvements in the market, or to enable other persons of ordinary skill in the art to understand the present invention.
Claims
1. A method for generating vehicle commuting behavior labels based on travel data, characterized in that: include: Identify the vehicle travel data based on the vehicle identification code to determine the corresponding vehicle, wherein the vehicle travel data includes a power-on signal, a power-off signal, a driving time, a driving mileage, a starting point information, and an end point information; Based on the vehicle travel data, cluster analysis method is used to determine the labels of the places frequently visited by the vehicle; Determine commuting vehicles based on vehicle travel data according to a preset vehicle commuting use model, and add commuting use labels; The reverse address resolution interface is used to resolve the vehicle frequented location label of the commuting vehicle corresponding to the commuting purpose label, and the home address information and company address information of the commuting vehicle are determined; A statistical analysis is performed on the driving time of the commuting vehicles corresponding to the vehicle commuting purpose labels, and the time period with the highest frequency is determined as the commuting time period. The working hours and off-get off work times corresponding to the commuting time period are determined based on the home address information and company address information of the commuting vehicles.
2. The method according to claim 1, characterized in that The step of determining the vehicle location tag using a cluster analysis method based on the vehicle travel data also includes: Based on the starting point and end point information of vehicle travel data, the vehicle starting and ending points are clustered using an unsupervised density clustering method, and discrete points are deleted; The vehicle travel data after the discrete points are deleted is clustered again based on the KMeans algorithm, and the silhouette coefficient method is used to determine the number of clusters; Sort the starting and ending point clusters of each vehicle, and add labels for the locations that the vehicles frequently visit.
3. The method according to claim 2, characterized in that The step of adding a commuting label based on the user's travel data according to the preset vehicle commuting usage model also includes: Extracting feature values based on vehicle travel data to generate commuting usage samples, wherein the feature values include a vehicle travel rate, an average number of daily trips, and an average daily mileage; A random forest classifier is used to generate a vehicle commuting usage model based on the extracted commuting usage samples for training; A vehicle commuting usage model is used to identify commuting vehicles based on vehicle trip data and add commuting usage labels.
4. The method according to claim 3, characterized in that Also includes: Extract feature values based on vehicle travel data to generate household car usage samples, use a random forest classifier to train based on the extracted household car usage samples to generate a household car usage model, and use the household car usage model to identify household cars based on vehicle travel data and add household car usage labels; Feature value extraction is performed based on vehicle travel data to generate a sample of online car-hailing usage. A random forest classifier is used to train the extracted online car-hailing usage samples to generate an online car-hailing usage model. The online car-hailing usage model is used to identify online car-hailing vehicles based on vehicle travel data and add online car-hailing usage labels.
5. A vehicle commuting behavior label generation system based on travel data, characterized in that: include: A data acquisition module, the data acquisition module is used to connect to the user's vehicle and receive vehicle travel data. The data acquisition module identifies the vehicle travel data based on the vehicle identification code and determines the corresponding vehicle. The vehicle travel data includes a power-on signal, a power-off signal, driving time, driving mileage, starting point information, and end point information; A location tag adding module, wherein the location tag adding module determines the location tags frequently visited by the vehicle by using a cluster analysis method based on the vehicle travel data; a usage label adding module, wherein the usage label adding module determines a commuting vehicle based on the vehicle travel data according to a preset vehicle commuting usage model, and adds a commuting usage label; A commuting address resolution module, wherein the commuting address resolution module uses a reverse address resolution interface to resolve the vehicle frequented location label of the commuting vehicle corresponding to the commuting purpose label, and determines the home address information and company address information of the commuting vehicle; A commuting time calculation module, wherein the commuting time calculation module performs statistical analysis on the driving time of the commuting vehicle corresponding to the vehicle commuting purpose label, determines the time period with the highest frequency as the commuting time period, and determines the working time and off-get off work time corresponding to the commuting time period according to the home address information and company address information of the commuting vehicle.
6. The system according to claim 5, characterized in that The step of the location tag adding module performing cluster analysis based on the user travel data to determine the vehicle location tag also includes: Based on the starting point and end point information of vehicle travel data, the vehicle starting and ending points are clustered using an unsupervised density clustering method, and discrete points are deleted; The vehicle travel data after the discrete points are deleted is clustered again based on the KMeans algorithm, and the silhouette coefficient method is used to determine the number of clusters; Sort the starting and ending point clusters of each vehicle, and add labels for the locations that the vehicles frequently visit.
7. The system according to claim 6, characterized in that The step of adding a commuting label by the usage label adding module based on the preset vehicle commuting usage model and the user's travel data further includes: Extracting feature values based on vehicle travel data to generate commuting usage samples, wherein the feature values include a vehicle travel rate, an average number of daily trips, and an average daily mileage; A random forest classifier is used to generate a vehicle commuting usage model based on the extracted commuting usage samples for training; A vehicle commuting usage model is used to identify commuting vehicles based on vehicle trip data and add commuting usage labels.
8. The system according to claim 7, characterized in that The usage label adding module also includes: extracting feature values based on vehicle travel data to generate household car usage samples and online car-hailing usage samples, using a random forest classifier to train based on the extracted household car usage samples and online car-hailing usage samples to generate a household car usage model and an online car-hailing usage model, identifying household cars based on vehicle travel data through the household car usage model and the online car-hailing usage model and adding household car usage labels and online car-hailing usage labels.