User travel path identification method and device, electronic equipment and storage medium

Abstract

The application provides a user travel path identification method and device based on big data, electronic equipment and computer readable storage medium, relates to the technical field of big data analysis, and comprises the following steps: preprocessing big data of a target path to eliminate abnormal data and obtaining effective big data; the shortest time corresponding to the target path is extracted; the total travel time of a given path is obtained by combining a path time model and performing operation; a to-be-matched path set is determined based on target card swiping data and the total travel time; the selection probability of each candidate path in the to-be-matched path set is calculated based on a path selection model, and the candidate path with the highest selection probability is determined as the actual travel path corresponding to the target card swiping data. The application improves the accuracy of the path selection model, is beneficial to accurate analysis of passenger travel behavior and passenger flow of a specific section, provides decision support for predicting travel and optimizing traffic resource allocation, and has high scalability.

Description

Technical Field

[0001] This invention relates to the field of big data analytics, and in particular to a method, apparatus, electronic device, and computer-readable storage medium for identifying user travel routes based on big data. Background Technology

[0002] With rapid urbanization, major cities are accelerating investment in public transportation, especially rail transit systems. Urban rail transit, with its advantages of speed and large capacity, has become the primary choice for an increasing number of travelers and a key focus for transportation authorities. Understanding passenger routes and subway ridership helps transportation departments to analyze subway passenger flow in detail, assess subway comfort and congestion, and optimize traffic allocation and planning. This is of positive significance for further attracting subway passengers and promoting sustainable development.

[0003] Currently, the main method for obtaining subway passenger flow data is through estimation using traffic flow distribution models built based on survey data. However, building an accurate model requires extensive questionnaire surveys, which consumes a lot of human and material resources, and ultimately only yields a relatively small dataset. At the same time, the application scope of the constructed model is limited; although the passenger flow estimation accuracy is high at the surveyed stations, it is difficult to achieve flow estimation across the entire subway network.

[0004] The emergence of big data from card swipes provides a valuable opportunity to analyze passenger flow across the entire subway network. Currently, urban rail transit uses automatic ticketing systems that record each time a passenger enters the turnstile using a magnetic card or contactless IC card. However, these systems only record the origin and destination stations and the time of the swipe, leaving the passenger's actual travel trajectory and path within the subway system incomplete. This makes it difficult to assess precise passenger flow and congestion levels. Therefore, identifying passenger travel paths based on card swipe data can enable low-cost analysis of passenger behavior and traffic statistics, contributing to refined analysis and decision-making regarding subway travel and future construction optimization. However, currently, there are no methods for identifying user travel paths using card swipe data. Summary of the Invention

[0005] In view of the above problems, the present invention is proposed to provide a method, apparatus, electronic device and computer-readable storage medium for identifying user travel routes based on big data to overcome or at least partially solve the above problems.

[0006] Firstly, a method for identifying user travel routes based on big data is provided, the method comprising:

[0007] The big data of the target path is preprocessed to remove abnormal data and obtain effective big data;

[0008] The Pareto optimal route algorithm is used to extract the effective big data to obtain the shortest time corresponding to the target path;

[0009] Based on the shortest time, the total travel time for a given route from any starting point to any ending point is calculated using the path time model.

[0010] Based on the target card swipe data and the total travel time, a set of routes to be matched is determined;

[0011] Based on the path selection model, the probability of each candidate path being selected is calculated in the set of candidate paths to be matched, thus obtaining the probability of each candidate path being selected.

[0012] The candidate path with the highest selection probability is determined as the actual travel path corresponding to the target card swipe data.

[0013] Optionally, the big data of the target path can be preprocessed to remove outlier data and obtain effective big data, including:

[0014] The big data of the target path is analyzed to obtain the total travel time. Here, a big data point represents a travel time between a pair of stations, and a pair of stations corresponds to multiple travel times.

[0015] Determine the distribution of multiple travel times corresponding to a station pair, and remove the travel times of the pre-preset proportion and the post-preset proportion in the distribution to obtain multiple valid travel times;

[0016] The effective travel times mentioned above are used as the effective big data for the target path.

[0017] Optionally, the effective big data is extracted using the Pareto optimal travel algorithm to obtain the shortest time corresponding to the target path, including:

[0018] Determine the arrival and departure times corresponding to each valid travel time in the effective big data;

[0019] Based on the arrival and departure times, select the valid travel time with the latest arrival time and the earliest departure time.

[0020] The effective travel time with the latest arrival time and the earliest departure time is determined as the shortest time.

[0021] Optionally, based on the shortest time, a path time model is used to calculate the total travel time for a given path from any starting point to any ending point, including:

[0022] Based on the shortest time, the path time model is calibrated to obtain the accurate values ​​of the model parameters in the path time model;

[0023] Based on the precise values ​​of the model parameters, the total travel time for a given route from any starting point to any ending point is calculated using the path time model.

[0024] Optionally, based on the shortest time, the path time model is calibrated to obtain accurate values ​​of the model parameters in the path time model, including:

[0025] Based on the traffic system network model, determine all possible travel routes between station pairs in the target path;

[0026] Based on all possible travel routes and in conjunction with the traffic system network model, determine the number of transfers and travel time for each travel route;

[0027] Using the shortest time, the number of transfers, and the travel time as partial model parameters, and substituting them into the path time model for calculation, a set of transfer delay times and walking times to and from stations are obtained.

[0028] Based on the aforementioned traffic system network model, all station pairs are identified;

[0029] For each of the aforementioned site pairs, we have:

[0030] Using the shortest time, number of transfers, and travel time corresponding to the station as some model parameters, and substituting them into the path time model for calculation, a set of transfer delay time and walking time to and from the station are obtained.

[0031] Based on multiple sets of transfer delay times and station entry / exit walking times, parameter calibration is performed to obtain accurate values ​​for the transfer delay time parameter and station entry / exit walking time parameter in the path time model.

[0032] Optionally, the path time model is characterized by the following expression:

[0033]

[0034] In the above formula, This refers to either the shortest time or the total travel time. This indicates the number of transfers. This indicates the travel time. This indicates the transfer delay time. This indicates the walking time for entering and exiting the station.

[0035] Optionally, based on the precise values ​​of the model parameters, the path time model is used to calculate the total travel time for a given path from any starting point to any ending point, including:

[0036] Based on the given path, and combined with the traffic system network model, multiple travel times corresponding to the given path are obtained;

[0037] Based on the given path and the traffic system network model, determine the number of transfers corresponding to the given path;

[0038] The precise values ​​of the transfer delay time parameter and the walking time parameter for entering and exiting the station, as well as the multiple travel times and multiple transfers corresponding to the given path, are substituted into the expression of the path time model for calculation to obtain the multiple total travel times for the given path.

[0039] Optionally, based on the target card swipe data and the total travel time, a set of routes to be matched is determined, including:

[0040] Analyze the target card swipe data to obtain the actual station pair and actual travel time corresponding to the target card swipe data;

[0041] Among the multiple total travel times corresponding to the actual station pairs, determine the effective total travel time within a preset time range that differs from the actual travel time;

[0042] The multiple paths corresponding to the total effective travel time are determined as the set of paths to be matched.

[0043] Optionally, the path selection model is an MNL model;

[0044] Based on the path selection model, the probability of selection for each candidate path in the set of paths to be matched is calculated to obtain the probability of selection for each candidate path, including:

[0045] Based on the stochastic utility formula in the MNL model, the utility of each candidate path in the set of paths to be matched is calculated.

[0046] Based on the utility of each alternative path, the probability of selection for each alternative path is calculated using the selection probability formula.

[0047] Optionally, the random utility formula is as follows:

[0048] ;

[0049] In the above formula, This represents the utility of any alternative path. Indicates the definite utility portion, Represents the random error term;

[0050] Based on the utility maximization criterion, choose any alternative path. Its utility needs to be satisfied Larger than the set of paths to be matched any other path in The utility ,Right now Therefore, any alternative path The probability of being selected is:

[0051]

[0052] Assumption If the path follows an independent and identically distributed Gumbel distribution, then any alternative path... The probability of being selected is calculated as follows:

[0053]

[0054] The above formula is the formula for the probability of being selected.

[0055] Optionally, the determination of utility portion This is a generalized cost function, which includes deterministic factors that influence the actual choice of travel route;

[0056] The determination of utility portion The expression is:

[0057]

[0058] In the above formula, This refers to either the shortest time or the total travel time. This represents the penalty index for cumulative transfer times. and These represent the coefficients of different variables;

[0059] The cumulative transfer number penalty index The expression is:

[0060]

[0061] In the above formula, This indicates the number of transfers. Indicates site and sites The straight-line distance.

[0062] Optionally, the variable coefficients and Obtain it through the following methods:

[0063] Based on existing travel routes and travel times, maximum likelihood estimation is performed, and parameter values ​​that satisfy statistical significance are selected as the variable coefficients. and .

[0064] Secondly, a user travel path identification device based on big data is provided, the user travel path identification device comprising:

[0065] The preprocessing module is used to preprocess the big data of the target path, remove abnormal data, and obtain effective big data;

[0066] The extraction module is used to extract the effective big data using the Pareto optimal travel algorithm to obtain the shortest time corresponding to the target path;

[0067] The total travel time module is used to calculate the total travel time for a given route from any starting point to any ending point based on the shortest time and the route time model.

[0068] The matching set module is used to determine the matching path set based on the target card swipe data and the total travel time;

[0069] The probability calculation module is used to calculate the selection probability of each candidate path in the set of candidate paths based on the path selection model, so as to obtain the selection probability of each candidate path.

[0070] The path determination module is used to determine the candidate path with the highest selection probability as the actual travel path corresponding to the target card swipe data.

[0071] Optionally, the preprocessing module is specifically used for:

[0072] The big data of the target path is analyzed to obtain the total travel time. Here, a big data point represents a travel time between a pair of stations, and a pair of stations corresponds to multiple travel times.

[0073] Determine the distribution of multiple travel times corresponding to a station pair, and remove the travel times of the pre-preset proportion and the post-preset proportion in the distribution to obtain multiple valid travel times;

[0074] The effective travel times mentioned above are used as the effective big data for the target path.

[0075] Optionally, the extraction module is specifically used for:

[0076] Determine the arrival and departure times corresponding to each valid travel time in the effective big data;

[0077] Based on the arrival and departure times, select the valid travel time with the latest arrival time and the earliest departure time.

[0078] The effective travel time with the latest arrival time and the earliest departure time is determined as the shortest time.

[0079] Optionally, the total travel time module includes:

[0080] A calibration unit is used to calibrate the model parameters of the path time model based on the shortest time to obtain accurate values ​​of the model parameters in the path time model.

[0081] The total travel time unit is used to calculate the total travel time of a given route from any starting point to any ending point using the precise values ​​of the model parameters and the route time model.

[0082] Optionally, the calibration unit is specifically used for:

[0083] Based on the traffic system network model, determine all possible travel routes between station pairs in the target path;

[0084] Based on all possible travel routes and in conjunction with the traffic system network model, determine the number of transfers and travel time for each travel route;

[0085] Using the shortest time, the number of transfers, and the travel time as partial model parameters, and substituting them into the path time model for calculation, a set of transfer delay times and walking times to and from stations are obtained.

[0086] Based on the aforementioned traffic system network model, all station pairs are identified;

[0087] For each of the aforementioned site pairs, we have:

[0088] Using the shortest time, number of transfers, and travel time corresponding to the station as some model parameters, and substituting them into the path time model for calculation, a set of transfer delay time and walking time to and from the station are obtained.

[0089] Based on multiple sets of transfer delay times and station entry / exit walking times, parameter calibration is performed to obtain accurate values ​​for the transfer delay time parameter and station entry / exit walking time parameter in the path time model.

[0090] Optionally, the total travel time unit is specifically used for:

[0091] Based on the given path, and combined with the traffic system network model, multiple travel times corresponding to the given path are obtained;

[0092] Based on the given path and the traffic system network model, determine the number of transfers corresponding to the given path;

[0093] The precise values ​​of the transfer delay time parameter and the walking time parameter for entering and exiting the station, as well as the multiple travel times and multiple transfers corresponding to the given path, are substituted into the expression of the path time model for calculation to obtain the multiple total travel times for the given path.

[0094] Optionally, the matching set module is specifically used for:

[0095] Analyze the target card swipe data to obtain the actual station pair and actual travel time corresponding to the target card swipe data;

[0096] Among the multiple total travel times corresponding to the actual station pairs, determine the effective total travel time within a preset time range that differs from the actual travel time;

[0097] The multiple paths corresponding to the total effective travel time are determined as the set of paths to be matched.

[0098] Optionally, the path selection model is an MNL model; the probability calculation module is specifically used for:

[0099] Based on the stochastic utility formula in the MNL model, the utility of each candidate path in the set of paths to be matched is calculated.

[0100] Based on the utility of each alternative path, the probability of selection for each alternative path is calculated using the selection probability formula.

[0101] Thirdly, an electronic device is provided, comprising:

[0102] One or more processors; and

[0103] One or more machine-readable media storing instructions thereon, when executed by the one or more processors, cause the electronic device to perform the big data-based user travel path identification method as described in any of the first aspects.

[0104] Fourthly, a computer-readable storage medium is provided, characterized in that the computer program stored therein causes a processor to execute the user travel path identification method based on big data as described in any of the first aspects.

[0105] This application has the following advantages:

[0106] In this invention, the big data of the target path is first preprocessed to remove abnormal data and obtain effective big data; then, the effective big data is extracted using the Pareto optimal travel algorithm to obtain the shortest time corresponding to the target path; then, based on the shortest time, the path time model is combined to calculate the total travel time of a given path from any starting point to any ending point.

[0107] Based on the target card swipe data and the total travel time, a set of routes to be matched is determined. Based on the route selection model, the probability of selection for each candidate route in the set of routes to be matched is calculated to obtain the probability of selection for each candidate route. Finally, the candidate route with the highest selection probability is determined as the actual travel route corresponding to the target card swipe data.

[0108] This method addresses the problem that current card-swipe data only includes origin-destination (OD) data and lacks user travel routes. It overcomes the shortcomings of current subway passenger flow allocation models, which primarily rely on small datasets of survey data and have limited applicability to specific stations. The route selection model constructed in this invention can be applied to travel between any stations in the network. This significantly reduces the manpower and resources required for extensive field surveys and visits, a common practice in traditional methods. Furthermore, it improves the accuracy of the route selection model, ensuring accuracy even with relatively limited survey data.

[0109] The route selection model, built based on card swipe data and traffic network data, identifies the travel routes of passengers based on a single card swipe record, rather than assigning a probability of total passenger flow between origin-destination (OD) stations as in traditional methods. This facilitates more precise analysis of passenger travel behavior and passenger flow on specific sections of the subway system, providing methodological support for transportation departments to analyze passenger behavior, passenger volume, and carriage congestion, and offering decision support for travel prediction and optimization of traffic resource allocation.

[0110] The method proposed in this invention covers the key elements considered in subway passenger route selection, has high scalability, and can be applied to different urban subway systems. It can also be applied to most urban rail transit systems equipped with automatic fare checking systems, and can be further extended to bus systems equipped with card-based fare checking systems for boarding and alighting. Attached Figure Description

[0111] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0112] Figure 1 This is a flowchart of a user travel route identification method based on big data according to an embodiment of the present invention;

[0113] Figure 2 This is an exemplary schematic diagram of abnormal data removal in an embodiment of the present invention;

[0114] Figure 3This is an exemplary schematic diagram of extracting the shortest time using the Pareto optimal travel algorithm in an embodiment of the present invention;

[0115] Figure 4 This is a schematic diagram of an exemplary set of paths to be matched in an embodiment of the present invention;

[0116] Figure 5 This is an overview diagram of a user travel route identification method according to an embodiment of the present invention;

[0117] Figure 6 This is a block diagram of a user travel route identification device based on big data according to an embodiment of the present invention. Detailed Implementation

[0118] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present invention, and are only some, not all, embodiments of the present invention, and are not intended to limit the present invention.

[0119] Reference Figure 1 The flowchart illustrates a user travel path identification method based on big data according to an embodiment of the present invention. This user travel path identification method based on big data includes:

[0120] Step 101: Preprocess the big data of the target path, remove abnormal data, and obtain effective big data.

[0121] In this embodiment of the invention, any transportation system will definitely include multiple stations, and there may be multiple paths between any two stations. Taking a subway system as an example: it may include multiple subway lines, such as Line 1, Line 2, Line 3, etc. Each subway line will include many stations. For any two stations A and B, assuming subway user XX enters from station A and exits from station B, subway user XX may first take Line 1 from station A to station C, then transfer to Line 3 to station D, and then transfer to Line 5 to station B; subway user XX may also first take Line 1 from station A to station E, then transfer to Line 2 to station B. In this embodiment of the invention, the path from entering station A to exiting station B is defined as a target path, equivalent to a station pair, or the path from exiting station A to entering station B is also a target path, equivalent to a station pair.

[0122] For any given target route, a massive amount of card-swipe data is generated daily. This data is stored, and when needed, it first requires preprocessing to remove outlier data and obtain valid data. Outlier data refers to time data that clearly does not conform to objective facts. For example, for a station A and B, ignoring transfer delays, walking time to and from the station, and waiting time for the subway, the subway travel time is calculated to be 30 minutes based solely on the distance between A and B and the subway speed. However, if subway user XX's card-swipe data for stations A and B corresponds to a time less than or significantly less than 30 minutes, then this card-swipe data is considered outlier. Of course, it's understandable that if subway user XX's card-swipe data for stations A and B is excessively long—for example, considering all possible times, the longest travel time between stations A and B should be 2 hours, but subway user XX's card-swipe data for stations A and B corresponds to a time greater than or significantly greater than 2 hours—then this card-swipe data is also considered outlier.

[0123] Outlier data can affect the accuracy of subsequent calculations; therefore, it is necessary to extract outlier data to obtain valid big data. In one possible embodiment, the method for obtaining valid big data is as follows:

[0124] First, the big data of the target route is analyzed to obtain all travel times. Each big data point represents a travel time between a pair of stations, and a pair of stations corresponds to multiple travel times. Then, the distribution of multiple travel times between a pair of stations is determined, and travel times with a pre-preset proportion and a post-preset proportion in the distribution are removed to obtain multiple effective travel times. Finally, the multiple effective travel times are used as the effective big data of the target route.

[0125] Since the target path refers to the path between a pair of stations, a single big data entry represents a travel time between those stations. For example, if subway user XX enters station A in the morning and exits station B, this creates one big data entry; if they enter station B again in the afternoon and exit station A, this creates another big data entry. Therefore, a single pair of stations can correspond to multiple travel times.

[0126] For multiple travel times, under normal subway user travel conditions, the differences between multiple travel times should be small, thus the distribution should be relatively concentrated. However, the occurrence times of outliers are more dispersed and relatively fewer, and they are certainly distributed in the earlier and later parts of the concentrated travel times. Therefore, by removing the travel times from the pre-set proportions before and after the pre-set proportions in the distribution, we can obtain multiple effective travel times. Assuming that both the pre-set proportions before and after are 10%, then removing the first 10% and the last 10% of travel times in the distribution, the remaining travel times are the multiple effective travel times. These multiple effective travel times are used as the effective big data for the target path. Figure 2 The diagram illustrates an exemplary outlier removal process. It can be seen that the vast majority of travel times are concentrated (represented by the black area in the diagram), while outlier data is scattered and relatively few, distributed in the first and last 10% of the concentrated area.

[0127] Step 102: Extract the effective big data using the Pareto optimal route algorithm to obtain the shortest time corresponding to the target path.

[0128] After obtaining the effective big data, in order to calibrate some parameters in the path-time model and obtain a more accurate travel route, the effective big data is first extracted using the Pareto optimal travel algorithm to obtain the shortest time corresponding to the target path. In one possible embodiment, this step specifically includes:

[0129] Step S1: Determine the arrival and departure times corresponding to each valid travel time in the valid big data;

[0130] Step S2: Based on the arrival and departure times, select the valid travel time with the latest arrival time and the earliest departure time;

[0131] Step S3: Determine the shortest travel time as the effective travel time between the latest arrival time and the earliest departure time.

[0132] A Pareto-optimal journey is defined as a trip with a given origin and destination that involves a later departure time, an earlier arrival time, and fewer transfers. For example... Figure 3 As shown, there are four card swipe records: t1, t2, t3, and t4. The superscript "dep" indicates entry, and the superscript "arr" indicates exit. It can be seen that the time corresponding to t2 is the Pareto optimal time, i.e., the shortest time. We then filter out the entry and exit times corresponding to all card swipe records for a station pair, assuming that the travel route follows the subway network. arrive of The route with the fewest transfers among the shortest travel times. Values ​​are set through practical experience (e.g.) ).

[0133] Based on this principle, the arrival and departure times corresponding to each valid travel time are determined from the available big data. Then, based on the arrival and departure times, the valid travel time with the latest arrival time and the earliest departure time is selected. Finally, this valid travel time with the latest arrival time and the earliest departure time is determined as the shortest time. It is understandable that the shortest time is definitely a definite time value, but it may not be unique in number; that is, if... Figure 2If t2 corresponds to 50 minutes, then there may be multiple times such as t5, t6, etc., each corresponding to 50 minutes.

[0134] Step 103: Based on the shortest time, and combined with the path time model, calculate the total travel time for a given path from any starting point to any ending point.

[0135] In this invention, the path-time model is a novel approach proposed by the inventors based on a transportation system network model. Taking a subway system as an example, complex network theory is used to extract and abstract the structure of the subway system, constructing a network topology graph for route analysis. The subway system is abstracted as a network with stations as nodes and connecting lines between adjacent stations as edges. The weight of each edge is set as the travel time on that route, which can be determined using data such as train timetables. Based on the constructed subway network model, each travel route can be extracted and encoded using the stations and lines it passes through. The sum of the weights of all edges on a route represents the travel time for that route.

[0136] To calculate a line from station in a subway network model arrive The given path The time distance can be broken down into three parts: travel time and travel time. Transfer time and walking time to and from the platform The transfer time is actually the product of the transfer delay time and the number of transfers. Therefore, for a given route... Its total travel time That is, the sum of the three parts of time. Generally, the travel time... It can be determined based on data such as train timetables, and can be calculated by summing the in-vehicle time (attribute) of each edge in the network model of the travel path; transfer time is determined by the number of transfers. and transfer delay time Calculate the product.

[0137] For any given travel route, if the number of transfers is fixed, then only the transfer delay time needs to be known. Walking time to and from the station You can then get the total travel time for the route. In order to obtain a more accurate passenger delay time... Walking time to and from the station In one possible embodiment, the path time model is first calibrated based on the shortest time to obtain the accurate values ​​of the model parameters in the path time model.

[0138] A preferred calibration method is as follows:

[0139] Step T1: Based on the traffic system network model, determine all possible travel routes between station pairs in the target path.

[0140] For any given travel path, defined as the target path, the route between station pairs A and B may include multiple travel options, such as direct routes without transfers, routes requiring one transfer, routes requiring two transfers, etc. While a direct route eliminates the need for transfers, it might be a loop route with many stops along the way. Conversely, while routes requiring two transfers incur some delays, they may involve fewer stops and therefore may not necessarily take longer than a direct route. For accurate model parameters, it is necessary to determine all possible travel routes between station pairs.

[0141] Step T2: Based on all possible travel routes and the transportation system network model, determine the number of transfers and travel time for each travel route;

[0142] Step T3: Using the shortest time, number of transfers, and travel time as partial model parameters, substitute them into the path time model to obtain a set of transfer delay times and walking times to and from stations.

[0143] For all possible travel routes, by combining the transportation system network model, the number of transfers and travel time for each route can be determined. Then, using the shortest time, number of transfers, and travel time as partial model parameters, and substituting them into the path time model for calculation, a set of transfer delay times and walking times to and from stations can be obtained. The expression for the path time model is as follows:

[0144]

[0145] In the above formula, Indicates the shortest or total travel time. Indicates the number of transfers. Indicates travel time. Indicates the delay time for transfers. Indicates the walking time for entering and exiting the station.

[0146] It should be noted that the expression may not be true for all possible travel routes. For example, for a certain travel route, only the travel time... It has already exceeded or is close to the shortest time. Or the calculated transfer delay time Walking time to and from the station This clearly does not conform to objective laws. In such cases, discard this data to avoid affecting transfer delays. Walking time to and from the station The accuracy of these two model parameters.

[0147] After receiving a set of transfer delay times Walking time to and from the station However, this data cannot be considered accurate and requires calibration with a large amount of data. Therefore, based on the required traffic system network model, all station pairs are determined; for each station pair, the following applies:

[0148] Using the shortest travel time, number of transfers, and travel time corresponding to a particular station as partial model parameters, these are substituted into the path time model for calculation, resulting in a set of transfer delay times and walking times for entering and exiting the station. This yields multiple sets of transfer delay times and walking times for entering and exiting the station. Based on these multiple sets of transfer delay times and walking times, parameter calibration is performed to obtain accurate values ​​for the transfer delay time parameters and walking times for entering and exiting the station in the path time model. This satisfies the transfer delay time parameters and walking times for entering and exiting the station used in the path time model, conforming to the normal travel patterns of most subway users. However, some extreme cases, such as subway users running quickly, are exceptions and not within the scope of the path time model.

[0149] After obtaining precise values ​​for transfer delay time and station entry / exit walking time, the total travel time for a given route from any starting point to any destination can be calculated using a path time model. That is:

[0150] Based on a given route, and using a traffic system network model, multiple travel times corresponding to the given route are calculated. Based on the given route and the traffic system network model, multiple transfer counts corresponding to the given route are determined. The precise values ​​of transfer delay time parameters and station entry / exit walking time parameters, along with the multiple travel times and multiple transfer counts corresponding to the given route, are substituted into the expressions of the aforementioned route time model for calculation, resulting in multiple total travel times for the given route. This total travel time There are several reasons for this. For a given route, the total travel time will definitely be different depending on the specific route taken by the traveler. Even if the traveler takes the same route, the total travel time will also be different depending on the traveler's speed, but the difference will definitely not be significant.

[0151] Step 104: Based on the target card swipe data and the total travel time, determine the set of routes to be matched.

[0152] The preceding steps yield the total travel time for a given route from any starting point to any destination, thus creating a precise database. Subsequently, for any card swipe data, this database can be used to obtain possible actual travel routes. First, based on the target card swipe data (equivalent to any card swipe data), combined with the total travel time, a set of routes to be matched needs to be determined. In one possible embodiment, the specific method is as follows:

[0153] The target card swipe data is parsed to obtain the corresponding actual station pair and actual travel time, i.e., to determine which station the card entered from, which station the card exited from, and the total travel time. Then, among the multiple total travel times corresponding to the actual station pair, the valid total travel time within a preset time range is determined. Since the database already contains all the total travel times corresponding to this actual station pair, a selection needs to be made. If the time difference is too large, the two trips must have used different routes. Based on this consideration, assuming a preset time of 10 minutes, valid total travel times within a 10-minute range from the actual travel time are determined. There are multiple valid total travel times, and finally, the multiple paths corresponding to all valid total travel times are determined as the set of paths to be matched. For example, refer to... Figure 4 The diagram illustrates an exemplary set of paths to be matched, from a website. i Arrive at the station j Between these, three paths, 1, 2, and 3, were obtained to be matched. The time difference between the total travel time corresponding to these three different travel paths and the actual travel time corresponding to the target card swipe data is within 10 minutes.

[0154] Step 105: Based on the path selection model, calculate the probability of selection for each candidate path in the set of candidate paths to be matched, and obtain the probability of selection for each candidate path.

[0155] Even after obtaining the set of paths to be matched, it is still impossible to determine which specific path the target card swipe data chose to travel. To solve this problem, the inventors designed a path selection model and a selection probability formula. Based on the path selection model, the selection probability of each candidate path in the set of candidate paths to be matched is calculated to obtain the selection probability of each candidate path.

[0156] In one possible implementation, the route selection model can be an MNL (Multinomial Logit) model, which estimates the probability of selecting alternative routes and is used for traffic flow allocation. Then, based on the stochastic utility formula in the MNL model, the utility of each alternative route in the set of routes to be matched is calculated. Since, according to stochastic utility theory, the influence of each factor on the choice can be quantified by utility values, the sum of the utilities of all factors represents the cost for the traveler to choose that route. Each time a traveler chooses a route, they select from the set of alternative routes. Medium utility The largest route The utility of each path is a random variable, consisting of a deterministic utility component. and random error term Based on this theory, the formula for random utility is as follows:

[0157] ;

[0158] In the above formula, This represents the utility of any alternative path. Indicates the definite utility portion, This represents the random error term.

[0159] Based on the utility maximization criterion, choose any alternative path. Its utility needs to be satisfied Larger than the set of paths to be matched any other path in The utility ,Right now Therefore, any alternative path The probability of being selected is:

[0160]

[0161] Assumption If the path follows an independent and identically distributed Gumbel distribution, then any alternative path... The probability of being selected is calculated as follows:

[0162]

[0163] The above formula is the formula for the probability of being selected.

[0164] Due to the determination of utility component This is a generalized cost function, which includes deterministic factors that influence the actual choice of travel route;

[0165] Therefore, the utility component is determined. The expression is:

[0166]

[0167] In the above formula, Indicates the shortest or total travel time. This represents the penalty index for cumulative transfer times. and These represent the coefficients of different variables;

[0168] In this study, it is considered that as the number of transfers increases, the impedance to route selection by travelers increases non-linearly, with the rate of increase decreasing, and the impact of this cumulative impedance is inversely proportional to the travel distance. Therefore, a non-linear function is used to calculate the penalty index for cumulative transfers. The expression is:

[0169]

[0170] In the above formula, Indicates the number of transfers. Indicates site and sites The straight-line distance. The exponential form in the numerator reflects the nonlinear impact of the number of transfers on route selection cost. As the denominator, it distinguishes the impact of the number of transfers on route selection for long-distance and short-distance travel; that is, the number of transfers has a greater impact on short-distance travel than on long-distance travel.

[0171] The coefficients of the variables in the above formula and Obtain it through the following methods:

[0172] Based on existing travel routes and travel times, maximum likelihood estimation is performed, and parameter values ​​that satisfy statistical significance are selected as variable coefficients. and .

[0173] The utility of each candidate path is obtained by using the above random utility formula. Then, by combining the selected probability formula, the selected probability of each candidate path can be obtained.

[0174] Step 106: Determine the candidate path with the highest probability of selection as the actual travel path for the corresponding target card swipe data.

[0175] After obtaining the selection probability of each candidate path in the set of paths to be matched, the candidate path with the highest selection probability is directly determined as the actual travel path of the corresponding target card swipe data, thereby accurately determining the actual travel path of any card swipe data.

[0176] The above methods can be referred to Figure 5The overview diagram shown illustrates the following: First, preparation is carried out at the basic data layer. Card swipe data is acquired and preprocessed. The subway network module is constructed using train timetables and traffic system structure diagrams. At the same time, survey data of various travel routes also need to be acquired to provide parameter estimation capabilities for the parameters in the discrete choice model (i.e., the path selection model, stochastic utility formula, and probability of selection formula in step 105).

[0177] Network path time calculation, which is to obtain the total travel time using the path travel time model, requires the extraction of Pareto optimal route algorithm to estimate some parameters in the path travel time model. The process of parameter estimation is also the process of parameter calibration.

[0178] Finally, in the actual travel route matching process, valid card swipe data is extracted from the preprocessed card swipe data, and routes to be matched are filtered based on the total travel time, thus obtaining the set of routes to be matched. Then, a discrete selection model is used to match these routes, ultimately yielding the actual travel route.

[0179] By employing the methods described in steps 101-106 above, this invention addresses the problem that current card-swipe data only includes origin-destination (OD) data and lacks user travel routes. It overcomes the shortcomings of current subway passenger flow allocation models, which primarily rely on small datasets of survey data and have limited applicability to specific stations. The route selection model constructed in this invention can be applied to travel between any stations in the network. This significantly reduces the manpower and resources required for extensive field surveys and visits in traditional methods. Furthermore, it improves the accuracy of the route selection model, ensuring accuracy even with relatively limited survey data.

[0180] The route selection model, built based on card swipe data and traffic network data, identifies the travel routes of passengers based on a single card swipe record, rather than assigning a probability of total passenger flow between origin-destination (OD) stations as in traditional methods. This facilitates more precise analysis of passenger travel behavior and passenger flow on specific sections of the subway system, providing methodological support for transportation departments to analyze passenger behavior, passenger volume, and carriage congestion, and offering decision support for travel prediction and optimization of traffic resource allocation.

[0181] Based on the above-described user travel path identification method based on big data, this invention also provides a user travel path identification device based on big data, referring to... Figure 6 The user travel route identification device based on big data includes:

[0182] The preprocessing module 610 is used to preprocess the big data of the target path, remove abnormal data, and obtain effective big data.

[0183] Extraction module 620 is used to extract the effective big data using the Pareto optimal travel algorithm to obtain the shortest time corresponding to the target path;

[0184] The total travel time module 630 is used to calculate the total travel time of a given route from any starting point to any ending point based on the shortest time and the route time model.

[0185] The matching set module 640 is used to determine the matching path set based on the target card swipe data and the total travel time;

[0186] The probability calculation module 650 is used to calculate the selection probability of each candidate path in the set of candidate paths to be matched based on the path selection model, so as to obtain the selection probability of each candidate path.

[0187] The path determination module 660 is used to determine the candidate path with the highest selection probability as the actual travel path corresponding to the target card swipe data.

[0188] Optionally, the preprocessing module 610 is specifically used for:

[0189] The big data of the target path is analyzed to obtain the total travel time. Here, a big data point represents a travel time between a pair of stations, and a pair of stations corresponds to multiple travel times.

[0190] Determine the distribution of multiple travel times corresponding to a station pair, and remove the travel times of the pre-preset proportion and the post-preset proportion in the distribution to obtain multiple valid travel times;

[0191] The effective travel times mentioned above are used as the effective big data for the target path.

[0192] Optionally, the extraction module 620 is specifically used for:

[0193] Determine the arrival and departure times corresponding to each valid travel time in the effective big data;

[0194] Based on the arrival and departure times, select the valid travel time with the latest arrival time and the earliest departure time.

[0195] The effective travel time with the latest arrival time and the earliest departure time is determined as the shortest time.

[0196] Optionally, the total travel time module 630 includes:

[0197] A calibration unit is used to calibrate the model parameters of the path time model based on the shortest time to obtain accurate values ​​of the model parameters in the path time model.

[0198] The total travel time unit is used to calculate the total travel time of a given route from any starting point to any ending point using the precise values ​​of the model parameters and the route time model.

[0199] Optionally, the calibration unit is specifically used for:

[0200] Based on the traffic system network model, determine all possible travel routes between station pairs in the target path;

[0201] Based on all possible travel routes and in conjunction with the traffic system network model, determine the number of transfers and travel time for each travel route;

[0202] Using the shortest time, the number of transfers, and the travel time as partial model parameters, and substituting them into the path time model for calculation, a set of transfer delay times and walking times to and from stations are obtained.

[0203] Based on the aforementioned traffic system network model, all station pairs are identified;

[0204] For each of the aforementioned site pairs, we have:

[0205] Using the shortest time, number of transfers, and travel time corresponding to the station as some model parameters, and substituting them into the path time model for calculation, a set of transfer delay time and walking time to and from the station are obtained.

[0206] Based on multiple sets of transfer delay times and station entry / exit walking times, parameter calibration is performed to obtain accurate values ​​for the transfer delay time parameter and station entry / exit walking time parameter in the path time model.

[0207] Optionally, the total travel time unit is specifically used for:

[0208] Based on the given path, and combined with the traffic system network model, multiple travel times corresponding to the given path are obtained;

[0209] Based on the given path and the traffic system network model, determine the number of transfers corresponding to the given path;

[0210] The precise values ​​of the transfer delay time parameter and the walking time parameter for entering and exiting the station, as well as the multiple travel times and multiple transfers corresponding to the given path, are substituted into the expression of the path time model for calculation to obtain the multiple total travel times for the given path.

[0211] Optionally, the matching set module 640 is specifically used for:

[0212] Analyze the target card swipe data to obtain the actual station pair and actual travel time corresponding to the target card swipe data;

[0213] Among the multiple total travel times corresponding to the actual station pairs, determine the effective total travel time within a preset time range that differs from the actual travel time;

[0214] The multiple paths corresponding to the total effective travel time are determined as the set of paths to be matched.

[0215] Optionally, the path selection model is an MNL model; the probability calculation module 650 is specifically used for:

[0216] Based on the stochastic utility formula in the MNL model, the utility of each candidate path in the set of paths to be matched is calculated.

[0217] Based on the utility of each alternative path, the probability of selection for each alternative path is calculated using the selection probability formula.

[0218] Based on the above-described user travel path identification method based on big data, this invention also provides an electronic device, including:

[0219] One or more processors; and

[0220] One or more machine-readable media storing instructions thereon, when executed by the one or more processors, cause the electronic device to perform the big data user travel path identification method as described in any one of steps 101 to 106.

[0221] Based on the above-described user travel path identification method based on big data, this embodiment of the invention also provides a computer-readable storage medium, wherein the stored computer program causes a processor to execute the user travel path identification method based on big data as described in any one of steps 101 to 106.

[0222] Through the above embodiments, the present invention first preprocesses the big data of the target path to remove abnormal data and obtain effective big data; then, it extracts the effective big data using the Pareto optimal travel algorithm to obtain the shortest time corresponding to the target path; and then, based on the shortest time, it performs calculations in combination with the path time model to obtain the total travel time of a given path from any starting point to any ending point.

[0223] Based on the target card swipe data and the total travel time, a set of routes to be matched is determined. Based on the route selection model, the probability of selection for each candidate route in the set of routes to be matched is calculated to obtain the probability of selection for each candidate route. Finally, the candidate route with the highest selection probability is determined as the actual travel route corresponding to the target card swipe data.

[0224] This method addresses the problem that current card-swipe data only includes origin-destination (OD) data and lacks user travel routes. It overcomes the shortcomings of current subway passenger flow allocation models, which primarily rely on small datasets of survey data and have limited applicability to specific stations. The route selection model constructed in this invention can be applied to travel between any stations in the network. This significantly reduces the manpower and resources required for extensive field surveys and visits, a common practice in traditional methods. Furthermore, it improves the accuracy of the route selection model, ensuring accuracy even with relatively limited survey data.

[0225] The route selection model, built based on card swipe data and traffic network data, identifies the travel routes of passengers based on a single card swipe record, rather than assigning a probability of total passenger flow between origin-destination (OD) stations as in traditional methods. This facilitates more precise analysis of passenger travel behavior and passenger flow on specific sections of the subway system, providing methodological support for transportation departments to analyze passenger behavior, passenger volume, and carriage congestion, and offering decision support for travel prediction and optimization of traffic resource allocation.

[0226] The method proposed in this invention covers the key elements considered in subway passenger route selection, has high scalability, and can be applied to different urban subway systems. It can also be applied to most urban rail transit systems equipped with automatic fare checking systems, and can be further extended to bus systems equipped with card-based fare checking systems for boarding and alighting.

[0227] Although preferred embodiments of the present invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present invention.

[0228] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0229] The technical solutions provided by the embodiments of the present invention have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for identifying user travel routes based on big data, characterized in that, The user travel route identification method includes: The big data of the target path is preprocessed to remove abnormal data and obtain effective big data; The Pareto optimal route algorithm is used to extract the effective big data to obtain the shortest time corresponding to the target path; Based on the shortest time, the total travel time for a given route from any starting point to any ending point is calculated using the path time model. Based on the target card swipe data and the total travel time, a set of routes to be matched is determined; Based on the path selection model, the probability of each candidate path being selected is calculated in the set of candidate paths to be matched, thus obtaining the probability of each candidate path being selected. The candidate path with the highest selection probability is determined as the actual travel path corresponding to the target card swipe data; Among them, based on the shortest time, and combined with the path time model, the total travel time for a given path from any starting point to any ending point is calculated, including: Based on the shortest time, the path time model is calibrated to obtain the accurate values ​​of the model parameters in the path time model; Based on the precise values ​​of the model parameters, the total travel time for a given route from any starting point to any ending point is calculated using the path time model. Specifically, based on the shortest time, the path time model is calibrated to obtain accurate values ​​of the model parameters, including: Based on the traffic system network model, determine all possible travel routes between station pairs in the target path; Based on all possible travel routes and in conjunction with the traffic system network model, determine the number of transfers and travel time for each travel route; Using the shortest time, the number of transfers, and the travel time as partial model parameters, and substituting them into the path time model for calculation, a set of transfer delay times and walking times to and from stations are obtained. Based on the aforementioned traffic system network model, all station pairs are identified; For each of the aforementioned site pairs, we have: Using the shortest time, number of transfers, and travel time corresponding to the station as some model parameters, and substituting them into the path time model for calculation, a set of transfer delay time and walking time to and from the station are obtained. Based on multiple sets of transfer delay times and station entry / exit walking times, parameter calibration is performed to obtain accurate values ​​of the transfer delay time parameter and station entry / exit walking time parameter in the path time model. The expression for the path time model is as follows: In the above formula, This refers to either the shortest time or the total travel time. This indicates the number of transfers. This indicates the travel time. This indicates the transfer delay time. This indicates the walking time for entering and exiting the station; The path selection model is the MNL model. Based on the path selection model, the probability of selection for each candidate path in the set of paths to be matched is calculated to obtain the probability of selection for each candidate path, including: Based on the stochastic utility formula in the MNL model, the utility of each candidate path in the set of paths to be matched is calculated. Based on the utility of each alternative path, the probability of each alternative path being selected is calculated using the selection probability formula. The formula for random utility is as follows: ； In the above formula, This represents the utility of any alternative path. Indicates the definite utility portion, Represents the random error term; Based on the utility maximization criterion, choose any alternative path. Its utility needs to be satisfied Larger than the set of paths to be matched any other path in The utility ,Right now Therefore, any alternative path The probability of being selected is: Assumption If the path follows an independent and identically distributed Gumbel distribution, then any alternative path... The probability of being selected is calculated as follows: The above formula is the formula for the probability of being selected; Among them, the determination of utility portion This is a generalized cost function, which includes deterministic factors that influence the actual choice of travel route; The determination of utility portion The expression is: In the above formula, This refers to either the shortest time or the total travel time. This represents the penalty index for cumulative transfer times. and These represent the coefficients of different variables; The cumulative transfer number penalty index The expression is: In the above formula, This indicates the number of transfers. Indicates site and sites The straight-line distance.

2. The user travel route identification method according to claim 1, characterized in that, The big data of the target path is preprocessed to remove outliers, resulting in effective big data, including: The big data of the target path is analyzed to obtain the total travel time. Here, a big data point represents a travel time between a pair of stations, and a pair of stations corresponds to multiple travel times. Determine the distribution of multiple travel times corresponding to a station pair, and remove the travel times of the pre-preset proportion and the post-preset proportion in the distribution to obtain multiple valid travel times; The effective travel times mentioned above are used as the effective big data for the target path.

3. The user travel route identification method according to claim 2, characterized in that, The Pareto optimal route algorithm is used to extract the shortest time corresponding to the target path from the effective big data, including: Determine the arrival and departure times corresponding to each valid travel time in the effective big data; Based on the arrival and departure times, select the valid travel time with the latest arrival time and the earliest departure time. The effective travel time with the latest arrival time and the earliest departure time is determined as the shortest time.

4. The user travel route identification method according to claim 1, characterized in that, Based on the precise values ​​of the model parameters, the total travel time for a given route from any starting point to any ending point is calculated using the path time model, including: Based on the given path, and combined with the traffic system network model, multiple travel times corresponding to the given path are obtained; Based on the given path and the traffic system network model, determine the number of transfers corresponding to the given path; The precise values ​​of the transfer delay time parameter and the walking time parameter for entering and exiting the station, as well as the multiple travel times and multiple transfers corresponding to the given path, are substituted into the expression of the path time model for calculation to obtain the multiple total travel times for the given path.

5. The user travel route identification method according to claim 1, characterized in that, Based on the target card swipe data and the total travel time, a set of routes to be matched is determined, including: Analyze the target card swipe data to obtain the actual station pair and actual travel time corresponding to the target card swipe data; Among the multiple total travel times corresponding to the actual station pairs, determine the effective total travel time within a preset time range that differs from the actual travel time; The multiple paths corresponding to the total effective travel time are determined as the set of paths to be matched.

6. The user travel route identification method according to claim 1, characterized in that, The variable coefficients and Obtain it through the following methods: Based on existing travel routes and travel times, maximum likelihood estimation is performed, and parameter values ​​that satisfy statistical significance are selected as the variable coefficients. and .

7. A user travel route identification device based on big data, characterized in that, The user travel route identification device includes: The preprocessing module is used to preprocess the big data of the target path, remove abnormal data, and obtain effective big data; The extraction module is used to extract the effective big data using the Pareto optimal travel algorithm to obtain the shortest time corresponding to the target path; The total travel time module is used to calculate the total travel time for a given route from any starting point to any ending point based on the shortest time and the route time model. The matching set module is used to determine the matching path set based on the target card swipe data and the total travel time; The probability calculation module is used to calculate the selection probability of each candidate path in the set of candidate paths based on the path selection model, so as to obtain the selection probability of each candidate path. The path determination module is used to determine the candidate path with the highest selection probability as the actual travel path corresponding to the target card swipe data. The total travel time module includes: A calibration unit is used to calibrate the model parameters of the path time model based on the shortest time to obtain accurate values ​​of the model parameters in the path time model. The total travel time unit is used to calculate the total travel time of a given route from any starting point to any ending point using the precise values ​​of the model parameters and the route time model. Specifically, the calibration unit is used for: Based on the traffic system network model, determine all possible travel routes between station pairs in the target path; Based on all possible travel routes and in conjunction with the traffic system network model, determine the number of transfers and travel time for each travel route; Using the shortest time, the number of transfers, and the travel time as partial model parameters, and substituting them into the path time model for calculation, a set of transfer delay times and walking times to and from stations are obtained. Based on the aforementioned traffic system network model, all station pairs are identified; For each of the aforementioned site pairs, we have: Using the shortest time, number of transfers, and travel time corresponding to the station as some model parameters, and substituting them into the path time model for calculation, a set of transfer delay time and walking time to and from the station are obtained. Based on multiple sets of transfer delay times and station entry / exit walking times, parameter calibration is performed to obtain accurate values ​​of the transfer delay time parameter and station entry / exit walking time parameter in the path time model. The expression for the path time model is as follows: In the above formula, This refers to either the shortest time or the total travel time. This indicates the number of transfers. This indicates the travel time. This indicates the transfer delay time. This indicates the walking time for entering and exiting the station; The path selection model is the MNL model. Based on the path selection model, the probability of selection for each candidate path in the set of paths to be matched is calculated to obtain the probability of selection for each candidate path, including: Based on the stochastic utility formula in the MNL model, the utility of each candidate path in the set of paths to be matched is calculated. Based on the utility of each alternative path, the probability of each alternative path being selected is calculated using the selection probability formula. The formula for random utility is as follows: ； In the above formula, This represents the utility of any alternative path. Indicates the definite utility portion, Represents the random error term; Based on the utility maximization criterion, choose any alternative path. Its utility needs to be satisfied Larger than the set of paths to be matched any other path in The utility ,Right now Therefore, any alternative path The probability of being selected is: Assumption If the path follows an independent and identically distributed Gumbel distribution, then any alternative path... The probability of being selected is calculated as follows: The above formula is the formula for the probability of being selected; Among them, the determination of utility portion This is a generalized cost function, which includes deterministic factors that influence the actual choice of travel route; The determination of utility portion The expression is: In the above formula, This refers to either the shortest time or the total travel time. This represents the penalty index for cumulative transfer times. and These represent the coefficients of different variables; The cumulative transfer number penalty index The expression is: In the above formula, This indicates the number of transfers. Indicates site and sites The straight-line distance.

8. An electronic device, characterized in that, include: One or more processors; and One or more machine-readable media storing instructions thereon, when executed by the one or more processors, cause the electronic device to perform the big data-based user travel path identification method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The stored computer program enables the processor to execute the big data-based user travel path identification method as described in any one of claims 1 to 6.

Images