Passenger flow statistical method, device, electronic equipment and computer program product

Abstract

The application relates to the technical field of data processing, and provides a passenger flow statistical method and device, electronic equipment and computer program product. The method comprises the following steps: collecting signaling data of users in a target area; determining sample data according to the signaling data of the users, and determining a Gaussian kernel function based on the sample data; inputting the Gaussian kernel function as a parameter into a support vector machine algorithm to determine a classification model; and performing passenger flow statistics according to the classification model. According to the embodiment of the application, the classification model is determined based on the signaling data of the users and the SVM algorithm of the Gaussian mixture semi-supervised model, so that the accuracy of the model is improved, and the accuracy and efficiency of the passenger flow statistics are improved.

Description

Technical Field

[0001] This application relates to the field of data processing technology, specifically to a passenger flow statistics method, apparatus, electronic device, and computer program product. Background Technology

[0002] Regional passenger flow statistics are an important issue. Complex passenger flow locations require effective control of passenger flow to ensure the safety of tourists while improving their experience. Currently, methods for regional passenger flow statistics include: (1) using a pre-sale ticketing system for passenger flow statistics, in the form of online ticketing and manual ticketing; (2) using infrared equipment for passenger flow statistics; (3) using WIFI for passenger flow statistics; and (4) using video surveillance for passenger flow statistics. However, the above methods are inefficient in statistically analyzing regional passenger flow. Summary of the Invention

[0003] This application provides a passenger flow statistics method, apparatus, electronic device, and computer program product to solve the problem of efficiency in regional passenger flow statistics.

[0004] In a first aspect, embodiments of this application provide a passenger flow statistics method, including:

[0005] Collect signaling data from users within the target area;

[0006] Based on the user's signaling data, sample data is determined, and based on the sample data, a Gaussian kernel function is determined.

[0007] The Gaussian kernel function is passed as a parameter to the support vector machine algorithm to determine the classification model;

[0008] Passenger flow statistics are performed based on the classification model.

[0009] In one embodiment, the step of performing passenger flow statistics based on the classification model includes:

[0010] Input the indicator features in the user's signaling data into the classification model, and obtain the weight value and score of each indicator output by the classification model;

[0011] User types are determined based on the weight value and score of each of the aforementioned indicators, and passenger flow statistics are performed based on the user types.

[0012] In one embodiment, determining sample data based on the user's signaling data includes:

[0013] Determine a first ratio of the number of base stations the user is connected to to the total number of base stations in the target area, and the user's movement speed in the target area;

[0014] Based on the first ratio and / or the moving speed, determine the user type corresponding to the user's signaling data;

[0015] The signaling data is obtained by labeling a set proportion of the signaling data based on the user type.

[0016] In one embodiment, the metrics of the user's signaling data include at least one of the following:

[0017] The duration of the user's connection to the base station within the target area;

[0018] Within the target area, the number of user-connected base stations is in a first ratio to the total number of base stations within the target area;

[0019] Within the target area, a second ratio is established between the number of user connection center base stations and the number of connected base stations;

[0020] The user's movement speed within the target area;

[0021] The similarity of the user's trajectory within the target area.

[0022] In one embodiment, determining the trajectory similarity of users within the target area includes:

[0023] Based on the movement matrix principle of Markov chains, the transfer frequency of the user between the two central base stations is determined;

[0024] The trajectory similarity of the user within the target area is determined based on the transfer frequency.

[0025] In one embodiment, after collecting signaling data of users within the target area, the method further includes:

[0026] Determine the location of the user's number and the information collected on the signaling data in order to classify the user's signaling data;

[0027] The classification model for different input data is determined based on the classified signaling data.

[0028] In one embodiment, the step of passing the Gaussian kernel function as a parameter into the support vector machine algorithm to determine the classification model includes:

[0029] The sample data was trained multiple times to obtain multiple training errors;

[0030] The accuracy of the classification model is determined based on the training error.

[0031] Secondly, embodiments of this application provide a passenger flow counting device, comprising:

[0032] The acquisition module is used to collect signaling data from users within the target area.

[0033] A Gaussian kernel function determination module is used to determine sample data based on the user's signaling data and to determine a Gaussian kernel function based on the sample data.

[0034] The classification model determination module is used to pass the Gaussian kernel function as a parameter into the support vector machine algorithm to determine the classification model.

[0035] The passenger flow statistics module is used to perform passenger flow statistics based on the classification model.

[0036] Thirdly, embodiments of this application provide an electronic device, including a processor and a memory storing a computer program, wherein the processor executes the program to implement the passenger flow statistics method described in the first aspect.

[0037] Fourthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the passenger flow statistics method described in the first aspect.

[0038] The passenger flow statistics method, apparatus, electronic device, and computer program product provided in this application collect signaling data from users within a target area; determine sample data based on the user signaling data; determine a Gaussian kernel function based on the sample data; input the Gaussian kernel function as a parameter into a support vector machine algorithm to determine a classification model; and perform passenger flow statistics based on the classification model. This application embodiment determines the classification model using an SVM algorithm based on user signaling data and a Gaussian mixture semi-supervised model, thereby improving the model's accuracy and consequently enhancing the accuracy and efficiency of passenger flow statistics. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is a flowchart illustrating the passenger flow statistics method provided in the embodiments of this application;

[0041] Figure 2 This is a schematic diagram of the structure provided in this application embodiment for mapping the classification plane of the original features to a higher-dimensional hyperplane based on the kernel function;

[0042] Figure 4This is a schematic diagram of the passenger flow statistics device provided in the embodiments of this application;

[0043] Figure 3 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0045] Figure 1 This is a flowchart illustrating the passenger flow statistics method provided in an embodiment of this application. (Refer to...) Figure 1 This application provides a passenger flow statistics method, which may include:

[0046] Step 100: Collect signaling data from users within the target area;

[0047] It should be noted that the target area refers to the area where passenger flow statistics need to be performed; the user's signaling data refers to the communication data between the end user and the transmitting base station or micro-station.

[0048] Among the relevant data sources within the target area, user signaling data is relatively more complete than other data types, making it more convincing in determining whether a user is a tourist. The collected base station data includes the following fields, as shown in Table 1:

[0049] Table 1

[0050] Field Name Field content Area Area Name Location Latitude and longitude of the base station Cell-id serial number

[0051] When collecting user signaling data, it's important to consider that using signaling data from non-holiday periods to construct the training set might result in an imbalanced sample size, with the proportion of users classified as regional tourists and non-regional tourists approaching zero. This imbalance could negatively impact the prediction accuracy of the established model. Therefore, when building the training set, it's necessary to collect signaling data not only from users connected to regional base stations during a non-holiday period but also from all users connected to regional base stations during a specific holiday period.

[0052] Collecting signaling data from users connecting to regional base stations during a non-holiday period is to increase the sample size of the "non-regional tourist" category in the training set; while collecting signaling data from all users connecting to regional base stations during a specific holiday period is to increase the sample size of the "regional tourist" category in the training set. This is because a machine learning model trained on a balanced training set with a large sample size can better predict users of unknown categories, resulting in more reliable results.

[0053] The signaling data of the users is associated with the base stations in the target area, and the proportion of all users connected to the base stations is calculated. The proportions are sorted from largest to smallest, and the top 1000 users are selected. Data is extracted from the first to the last connection of each user to the base station in the target area, in 10-minute slices.

[0054] The collected raw data has the following fields, see Table 2 for reference:

[0055] Table 2

[0056]

[0057]

[0058] Step 200: Determine sample data based on the user's signaling data, and determine the Gaussian kernel function based on the sample data;

[0059] After collecting user signaling data, the criteria for determining user type are established. These criteria are then used to determine the user type (including regional visitors and non-regional visitors) corresponding to the user's signaling data. Based on the user type, a predetermined proportion of the signaling data is labeled to obtain sample data. The predetermined proportion can be 5:5 or 6:4 between labeled and unlabeled data, and this proportion can be determined based on model training requirements.

[0060] For example, since the collected user signaling data is not labeled as to whether a user is a regional tourist, and manually determining whether a user is a regional tourist is costly and inaccurate, this application embodiment constructs an indicator for determining whether a user is a regional tourist as input to the model. The model uses user type determination conditions to determine whether a user is a regional tourist, thereby manually labeling a portion of the dataset.

[0061] After collecting the user's signaling data, the user type is determined based on the following criteria: regional visitor and non-regional visitor.

[0062] If the ratio of the number of base stations a user connects to throughout the entire process to the total number of base stations in the area is greater than or equal to 0.7, then the user is considered a regional visitor.

[0063] If 0.7 >= the number of base stations the user connected to throughout the journey / the total number of base stations in the area >= 0.3, and the user's movement speed <= 0.1 m / s, then the user is considered a resident or worker in the area, and thus a visitor. This is because in reality, residents or workers in the area also increase the area's capacity density.

[0064] If either of the above two criteria is met, the user is considered a regional tourist; otherwise, if neither of the above two criteria is met, the user is classified as a non-regional tourist.

[0065] It should be noted that the ratio of regional tourist sample size to non-regional tourist sample size will affect the training results of the model. Therefore, it is necessary to calculate the ratio of regional tourist sample size to non-regional tourist sample size. If the ratio is between 0.4 and 0.6, the data is considered to be good and there is no sample imbalance problem. Otherwise, if it is outside this range, the data imbalance needs to be addressed, such as by using the SMOTE algorithm to generate virtual samples for the category with a smaller proportion to achieve balance.

[0066] After dividing the sample data into labeled and unlabeled samples, a semi-supervised support vector machine classification model based on Gaussian mixture model kernel is established. This enables the model to learn the clustering features of labeled samples and improve the binary classification output results through training on unlabeled samples.

[0067] Furthermore, after determining the training sample data, the Gaussian kernel function is determined based on the sample data.

[0068] For example, a Gaussian Mixture Model (GMM) is used to learn the selected sample dataset {x1, ..., x5}, {x1′, ..., x5′}, where x1, ..., x5 represent "tourists" and x′1, ..., x′5 represent "non-tourists". The data in the sample dataset are indicator data, including duration, first proportion, second proportion, movement speed, and trajectory similarity. The initial learning process focuses on the basic characteristics and classification features of the data. The basic characteristics refer to the indicator features, and the classification features refer to the differences and critical values ​​between the "tourists" and "non-tourists" categories under these indicator features.

[0069] For each sample point x i ∈X, x i ={x i (1) ,x i (2) ,x i (3) ,x i (4) ,x i (5)To prevent the dimensions of the data from affecting the model results, all indicators are normalized using the MINMAX method, as shown in the following formula:

[0070]

[0071] After normalizing the samples, they are fed into the model for learning. Based on the principle of minimizing the error of the fitting function, a Gaussian kernel function for each class is fitted. The learned Gaussian kernel function for each class is then used as a parameter to feed into the support vector machine classification algorithm. In this way, not only can the classification information of the labeled samples be fully learned, but the learned classification model parameters can also be improved using unlabeled samples, resulting in higher model classification accuracy.

[0072] The steps to determine the Gaussian kernel function based on sample data are as follows:

[0073] The output of the classification model is actually a binary classification problem, meaning that the sample dataset X can be fitted and learned using a binary Gaussian mixture distribution model, represented in the following form:

[0074]

[0075] Where p(v) represents the probability that the sample belongs to class v, and q(X|θ) represents the Gaussian probability density function with parameter θ.

[0076] For labeled variable x∈X, unlabeled variable x ′ ∈X, the joint Gaussian kernel of these two variables is:

[0077]

[0078] Right now:

[0079]

[0080] Where V represents a finite set of latent variables, and the number of latent variables is 2; k z (z,z ′ () represents two variables: z = [x, v], z ′ =[x ′ ,v ′ The joint Gaussian distribution probability density function of ], i.e., the Gaussian kernel function.

[0081] Using the sample dataset X, the EM algorithm is employed to obtain the local optimum estimate of the Gaussian mixture model for the parameter θ, where θ = {μ, Σ}. Therefore, for any x ∈ X, we have:

[0082]

[0083] Therefore:

[0084]

[0085] After obtaining the estimate, the Gaussian kernel function k can be determined. z (z,z ′ ):

[0086]

[0087] Step 300: Pass the Gaussian kernel function as a parameter into the support vector machine algorithm to determine the classification model;

[0088] After determining the Gaussian kernel function, the Gaussian kernel function is passed as a parameter into the support vector machine algorithm to determine the classification model.

[0089] For example, the Support Vector Machine (SVM) algorithm maps the original feature classification plane to a higher-dimensional hyperplane through a kernel function, resulting in a linearly separable high-dimensional plane and achieving better classification results. (Reference) Figure 2 , Figure 2 In this context, φ represents the kernel function, which must be a positive definite kernel function. In the SVM binary classification algorithm, the aforementioned Gaussian kernel function (i.e., the semi-supervised mixture Gaussian kernel function) is introduced as the mapping function of the support vector machine and passed as parameters. The data and dataset X are represented as {X... l ,X u}, where X l For the dataset labeled using the methods described above, X u This represents the unlabeled dataset. For labeled samples, the discriminant function of the SVM is:

[0090]

[0091] Where, α i Let x represent the Lagrange multipliers, which are used to solve for the local optimum. i The corresponding weights are given by b, which is the calculated threshold. Then, unlabeled data is added to construct a new optimal function:

[0092]

[0093] Where Ω represents the regularization function, which is a constraint on the space; C represents the penalty function, which reflects the degree of matching of f in the labeled sample set; y i For sample x i Corresponding tags; Let represent the new space containing labeled and unlabeled samples, and let the geometric distance of the space be expressed as:

[0094]

[0095] By incorporating information from unlabeled samples into the classifier, the model is able to fully explore the overall characteristics of the data, ultimately resulting in a binary classification model for regional tourists based on regional user base station data.

[0096] Furthermore, the accuracy of the classification model is verified using ten-fold cross-validation. For example, a dataset is constructed using five selected indicators. The SVM parameters are set as follows: a Gaussian kernel function is chosen, with parameters estimated using the EM algorithm; the penalty factor C is empirically set to 1000; and the number of iterations is chosen to be 60. This yields the sample classification result of the input data, indicating whether the user is a tourist from the region. Finally, ten-fold cross-validation is used. The samples are divided into 10 parts, each containing labeled and unlabeled samples. One part is used as the test set, and the other as the training set. The model is trained ten times without repetition, and the mean training error is used as the criterion for judging the accuracy of the model's fit.

[0097] Step 400: Perform passenger flow statistics based on the classification model.

[0098] It should be noted that if the signaling data features of users are directly used as the input of the model, some user features may not be identified, resulting in the model being insensitive to different user data. Therefore, it is necessary to extract indicator features from the user's signaling data and then input the extracted features into the classification model.

[0099] After determining the classification model, the indicator features in the user's signaling data are input into the classification model to obtain the weight value and score of each indicator output by the classification model. Then, the user type is determined based on the weight value and score of each indicator to perform passenger flow statistics based on the user type.

[0100] For example, five metrics are used to evaluate each user x: duration, first ratio, second ratio, movement speed, and trajectory similarity. i (i.e., sample points), constituting the training metric set {x} i (1) ,x i (2) ,x i (3) ,x i (4) ,x i (5) The system inputs the set of indicators into the classification model, which outputs the weights and scores of each indicator. Based on the weights and scores of each indicator, the importance of the indicator can be determined and whether the user is a tourist can be predicted, thereby statistically analyzing tourist traffic in the target area.

[0101] The passenger flow statistics method provided in this application collects signaling data from users within a target area; determines sample data based on the user signaling data, and determines a Gaussian kernel function based on the sample data. The sample data includes labeled and unlabeled sample data; the Gaussian kernel function is passed as a parameter into a support vector machine algorithm to determine a classification model; and passenger flow statistics are performed based on the classification model. This application uses signaling data as the data source to statistically analyze real-time passenger flow in a region, providing wide coverage and more accurate statistics. Using a classification model for passenger flow statistics is cost-effective, and utilizing user signaling data collected from mobile base stations saves on equipment purchase costs and improves passenger flow statistics efficiency.

[0102] Furthermore, the SVM algorithm based on a Gaussian mixture semi-supervised model determines whether a user is a tourist within the area using user location and time data within the base station range. This algorithm then predicts regional passenger flow in real time, offering lower costs, handling larger datasets, and faster processing speed compared to manual calculations. Simultaneously, it can train the model to obtain parameters (i.e., kernel function parameter estimates) based on the real-time input dataset, and then use SVM to train classification. Compared to traditional statistical algorithms, machine learning can adapt to the latest data, significantly improving prediction accuracy and robustness for real-time passenger flow statistics. It can efficiently classify and predict real-time users, resulting in more accurate regional passenger flow figures.

[0103] Based on the above embodiments, the indicators of the user's signaling data include at least one of the following: the duration of the user's connection to a base station within the target area; a first ratio of the number of base stations the user connects to within the target area to the total number of base stations within the target area; a second ratio of the number of central base stations the user connects to within the target area to the total number of connected base stations; the user's movement speed within the target area; and the trajectory similarity of the user within the target area.

[0104] It should be noted that, in order to distinguish between ambiguous samples, five indicators are calculated based on the original signaling data of the samples, and relevant thresholds are given based on the training samples.

[0105] Directly using user signaling data features as input to a semi-supervised support vector machine (SVM) classification model based on a Gaussian mixture model kernel may result in some user features being unrecognized, leading to model insensitivity to different user data and ultimately poor prediction results or serious errors. Therefore, it is necessary to process user signaling data to extract hidden indicator features. These hidden indicators include duration, first proportion, second proportion, movement speed, and trajectory similarity. The indicator features and their calculation methods are as follows:

[0106] Duration: The duration of connection to the base station within the target area.

[0107] In the user trajectory, the base station tracking records of the user within the target area are determined. Based on the tracking records, the total length of time the user stays within the base station set is obtained. If the stay time is too short or too long, the user is less likely to be a tourist in this area.

[0108] The first ratio is: the number of regional base stations among the base stations connected to the user / the total number of base stations within the region.

[0109] By linking the user trajectory table with the total number of base stations within the target area, the number of base stations with which the user has a tracking record within the target area is determined. The larger the ratio of the number of base stations with tracking records in the target area to the total number of base stations in the target area, the more likely the user is to be a tourist in that area.

[0110] The second ratio is: the number of "central base stations" connected to by users / the number of base stations connected to by users within the target area.

[0111] Among all users, the "central base station" with the highest number of known tourist connections within the target area is identified. The selection method for known tourists is as follows: From 5000 trajectories collected within a day, the user with the highest overlap with base station data points within the target area is selected. Users who might be staff members are removed, and the remaining users are considered tourists in this target area. The connection count of all central base stations must account for 90% of the total connection count of all base stations within the target area. After selecting the central base station, the higher the proportion of a user's data points at the central base station relative to the total number of data points at base stations within the target area, the more likely they are to be a tourist in that area.

[0112] Movement speed: The user's third-order smoothed movement speed calculated based on base station data.

[0113] Based on the user's location and timing at the base station, the user's movement speed can be calculated by measuring the third-order motion smoothing speed under the base station signaling data. If the movement speed is too fast and the user's connection to the "central base station" is not high, then the user is unlikely to be a tourist in the area.

[0114] Trajectory similarity: The similarity of user trajectories within the target area.

[0115] The trajectory similarity is calculated based on the determination of the "central base station" and the movement matrix principle of Markov chains. It assumes that the probability of a tourist switching between base stations within the target area during the same time period depends only on the current base station and is independent of the previous connected base station. Based on the above method for selecting known tourists, the transfer frequency between each pair of central base stations is calculated, and this frequency is used as the score for each pair of base station switches within the target area. Since the user's switching between base stations is independent of the user's previous base station location, the user's final trajectory similarity score is the sum of the scores for each pair of base station switching frequencies within the target area, serving as an indicator of trajectory similarity. A higher trajectory similarity score indicates a greater likelihood that the user is a regional tourist. The method for determining the trajectory similarity of a user within the target area in this application is easy to calculate and is not affected by large errors in the latitude and longitude records of the user by the base stations, offering significant advantages in the application scenario of regional tourists.

[0116] The above five metrics apply to each user x i (i.e., sample points) constitute the training metric set {x} i (1) ,x i (2) ,x i (3) ,x i (4) ,x i (5) By inputting the set of indicators into the classification model for training, the weights and scores of each indicator can be obtained. Based on the weights and scores of each indicator, the importance of the indicator can be determined and whether the user is a tourist can be predicted, thereby statistically analyzing the tourist traffic in the target area.

[0117] This application embodiment improves the accuracy of the model by determining the indicators of the user's signaling data and training the model based on the indicators.

[0118] Based on the above embodiments, after collecting signaling data of users within the target area, the method further includes: determining the location of the user's number and the collection information of the signaling data to classify the user's signaling data; and determining the classification model for different input data based on the classified signaling data.

[0119] It should be noted that, in order to simplify the steps in the model operation process and improve operational efficiency, classification models with different input variables can be established based on different types of signaling data. Specifically, the location of the user's number and the information on the collection of signaling data are determined to classify the user's signaling data, and then the classification model for different input data is determined based on the classified signaling data.

[0120] For example, different metrics can be reasonably combined and calculated based on different types of user signaling data. The user signaling data classification and the combined metrics calculated are as follows:

[0121] (1) If the location of the user's mobile phone number is the city where the target area is located, the combined calculation indicators include: duration, first ratio, movement speed and trajectory similarity.

[0122] If a user's phone number indicates that the user is a local resident, then the model needs more complex indicators to distinguish whether the user is a resident living near the area, passing through the area, or a tourist. Therefore, four indicators are needed to make the judgment simultaneously.

[0123] (2) If the location of the user's mobile phone number is not the city where the target area is located, the combined calculation indicators include: the first ratio and the second ratio.

[0124] If a user's mobile phone number shows that the user is not a local resident, it indicates that the user is only visiting the area for a short period of time. The first and second ratios can be used to determine whether the user is a tourist in the area.

[0125] (3) If the user’s signaling data has serious time collection gaps, the combined calculation indicators include: first ratio, second ratio and trajectory similarity.

[0126] If there is significant time loss in the collected user data, and only the user's latitude and longitude information is usable in the original indicators, the above three indicators are calculated to improve the practicality of the indicators.

[0127] Based on signaling data from different types of users, establish classification models with different input variables.

[0128] This application embodiment classifies user signaling data by determining the user's phone number's location and signaling data collection information, and then determines a classification model for different input data based on the classified signaling data. This simplifies the model operation process, improves operational efficiency, and consequently enhances passenger flow statistics efficiency.

[0129] The passenger flow counting device provided in the embodiments of this application is described below. The passenger flow counting device described below can be referred to in correspondence with the passenger flow counting method described above.

[0130] refer to Figure 3 , Figure 3 This is a schematic diagram of the passenger flow statistics device provided in the embodiments of this application. The passenger flow statistics device of this application includes a data acquisition module 301, a Gaussian kernel function determination module 302, a classification model determination module 303, and a passenger flow statistics module 304.

[0131] The acquisition module 301 is used to acquire signaling data of users within the target area;

[0132] Gaussian kernel function determination module 302 is used to determine sample data based on the user's signaling data and determine the Gaussian kernel function based on the sample data;

[0133] The classification model determination module 303 is used to pass the Gaussian kernel function as a parameter into the support vector machine algorithm to determine the classification model;

[0134] The passenger flow statistics module 304 is used to perform passenger flow statistics based on the classification model.

[0135] The passenger flow statistics device provided in this application collects signaling data from users within a target area; determines sample data based on the user signaling data; determines a Gaussian kernel function based on the sample data; inputs the Gaussian kernel function as a parameter into a support vector machine algorithm to determine a classification model; and performs passenger flow statistics based on the classification model. This application embodiment determines the classification model using an SVM algorithm based on user signaling data and a Gaussian mixture semi-supervised model, thereby improving the model's accuracy and consequently enhancing the accuracy and efficiency of passenger flow statistics.

[0136] In one embodiment, the passenger flow statistics module 304 is specifically used for:

[0137] Input the indicator features in the user's signaling data into the classification model, and obtain the weight value and score of each indicator output by the classification model;

[0138] User types are determined based on the weight value and score of each of the aforementioned indicators, and passenger flow statistics are performed based on the user types.

[0139] In one embodiment, the Gaussian kernel function determination module 302 is specifically used for:

[0140] Determine a first ratio of the number of base stations the user is connected to to the total number of base stations in the target area, and the user's movement speed in the target area;

[0141] Based on the first ratio and / or the moving speed, determine the user type corresponding to the user's signaling data;

[0142] The signaling data is obtained by labeling a set proportion of the signaling data based on the user type.

[0143] The passenger flow statistics device further includes an indicator determination module, which is specifically used for:

[0144] The duration of the user's connection to the base station within the target area;

[0145] Within the target area, the number of user-connected base stations is in a first ratio to the total number of base stations within the target area;

[0146] Within the target area, a second ratio is established between the number of user connection center base stations and the number of connected base stations;

[0147] The user's movement speed within the target area;

[0148] The similarity of the user's trajectory within the target area.

[0149] In one embodiment, the indicator determination module is specifically used for:

[0150] Based on the movement matrix principle of Markov chains, the transfer frequency of the user between the two central base stations is determined;

[0151] The trajectory similarity of the user within the target area is determined based on the transfer frequency.

[0152] In one embodiment, the acquisition module 301 is specifically used for:

[0153] Determine the location of the user's number and the information collected on the signaling data in order to classify the user's signaling data;

[0154] The classification model for different input data is determined based on the classified signaling data.

[0155] In one embodiment, the classification model determination module 303 is specifically used for:

[0156] The sample data was trained multiple times to obtain multiple training errors;

[0157] The accuracy of the classification model is determined based on the training error.

[0158] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4 As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other via the communication bus 440. The processor 410 can call a computer program stored in the memory 430 to execute steps of a passenger flow statistics method, such as:

[0159] Collect signaling data from users within the target area;

[0160] Based on the user's signaling data, sample data is determined, and based on the sample data, a Gaussian kernel function is determined.

[0161] The Gaussian kernel function is passed as a parameter to the support vector machine algorithm to determine the classification model;

[0162] Passenger flow statistics are performed based on the classification model.

[0163] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0164] On the other hand, this application also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can perform the steps of the passenger flow statistics method provided in the above embodiments, such as including:

[0165] Collect signaling data from users within the target area;

[0166] Based on the user's signaling data, sample data is determined, and based on the sample data, a Gaussian kernel function is determined.

[0167] The Gaussian kernel function is passed as a parameter to the support vector machine algorithm to determine the classification model;

[0168] Passenger flow statistics are performed based on the classification model.

[0169] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0170] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0171] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for passenger flow statistics, characterized in that, include: Collect signaling data from users within the target area; Based on the user's signaling data, sample data is determined, and based on the sample data, a Gaussian kernel function is determined. The Gaussian kernel function is passed as a parameter to the support vector machine algorithm to determine the classification model; Passenger flow statistics are performed based on the classification model. The step of determining sample data based on the user's signaling data includes: Determine a first ratio of the number of base stations the user is connected to to the total number of base stations in the target area, and the user's movement speed in the target area; Based on the first ratio and / or the movement speed, the user type corresponding to the user's signaling data is determined; the user type includes regional tourists and non-regional tourists; The signaling data is obtained by labeling a set proportion of the signaling data based on the user type.

2. The passenger flow statistics method according to claim 1, characterized in that, The process of performing passenger flow statistics based on the classification model includes: Input the indicator features in the user's signaling data into the classification model, and obtain the weight value and score of each indicator output by the classification model; User types are determined based on the weight value and score of each of the aforementioned indicators, and passenger flow statistics are performed based on the user types.

3. The passenger flow statistics method according to claim 2, characterized in that, The metrics of the user's signaling data include at least one of the following: The duration of the user's connection to the base station within the target area; Within the target area, the number of user-connected base stations is in a first ratio to the total number of base stations within the target area; Within the target area, a second ratio is established between the number of user connection center base stations and the number of connected base stations; The user's movement speed within the target area; The similarity of the user's trajectory within the target area.

4. The passenger flow statistics method according to claim 3, characterized in that, Determining the trajectory similarity of users within the target area includes: Based on the movement matrix principle of Markov chains, the transfer frequency of the user between the two central base stations is determined; The trajectory similarity of the user within the target area is determined based on the transfer frequency.

5. The passenger flow statistics method according to claim 1, characterized in that, After collecting signaling data from users within the target area, the method further includes: Determine the location of the user's number and the information collected on the signaling data in order to classify the user's signaling data; The classification model for different input data is determined based on the classified signaling data.

6. The passenger flow statistics method according to claim 1, characterized in that, The step of passing the Gaussian kernel function as a parameter into the support vector machine algorithm to determine the classification model includes: The sample data was trained multiple times to obtain multiple training errors; The accuracy of the classification model is determined based on the training error.

7. A passenger flow counting device, characterized in that, include: The acquisition module is used to collect signaling data from users within the target area. A Gaussian kernel function determination module is used to determine sample data based on the user's signaling data and to determine a Gaussian kernel function based on the sample data. The classification model determination module is used to pass the Gaussian kernel function as a parameter into the support vector machine algorithm to determine the classification model. The passenger flow statistics module is used to perform passenger flow statistics based on the classification model. The Gaussian kernel function determination module is further configured to determine a first ratio of the number of base stations connected to the user to the total number of base stations in the target area, and the user's movement speed in the target area; determine the user type corresponding to the user's signaling data based on the first ratio and / or the movement speed; the user type includes regional visitors and non-regional visitors; and label a set proportion of signaling data based on the user type to obtain the sample data.

8. An electronic device comprising a processor and a memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the passenger flow statistics method according to any one of claims 1 to 6.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the passenger flow statistics method according to any one of claims 1 to 6.

Images