An urban resident activity space access pattern analysis method based on access probability

Abstract

The application relates to the technical field of urban resident activity space analysis, in particular to a kind of urban resident activity space access mode analysis method based on access probability, comprising the following steps: step one: calculating the urban resident activity space on the scale of traffic subdistricts;Step two: calculating the access probability of urban residents to traffic subdistricts in their activity space;Step three: constructing the characteristic vector of the access mode of the activity space of urban residents;Step four: using the characteristic vector to cluster the access mode of the activity space of urban residents, and using the clustering result evaluation index to determine the final cluster number;Step five: using the access intensity to analyze the clustering result.The method can mine the potential access mode of urban residents, evaluate the common characteristics of the use of the activity space of urban residents in the same access mode by using the access intensity, comprehensively analyze the travel characteristics of urban residents in their activity space, and form a relatively perfect access mode analysis method.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of urban resident activity space analysis, and more particularly to a method for analyzing urban resident activity space access mode based on access probability. BACKGROUND

[0002] The access mode of urban residents to their activity space reflects the tendency of urban residents to choose activity locations in their activity space, and is closely related to the built environment in which the urban residents are located, the social and economic attributes of the urban residents, and the demographic attributes. Analyzing the access mode of urban residents to their activity space helps to understand their use of activity space and provides a reference for urban space resource allocation.

[0003] Existing methods for analyzing the access mode of urban residents in their activity space often use indicators such as travel distance, travel time, travel frequency, and travel purpose to analyze the characteristics of urban residents' travel in time and space and the choices and preferences of urban residents during travel. However, the access of urban residents to spatial locations within their activity space represents the degree of use of their activity space by urban residents and can be used as an indicator to measure the access mode of urban residents to their activity space. At present, methods for analyzing access mode using the access probability of urban residents to traffic zones within their activity space need to be improved. SUMMARY

[0004] The present application provides a method for analyzing the access mode of urban residents to their activity space based on access probability, with the aim of providing a comprehensive analysis of the access mode of urban residents.

[0005] The above-mentioned object is achieved by the following technical solutions:

[0006] A method for analyzing the access mode of urban residents to their activity space based on access probability, comprising the following steps:

[0007] Step one: calculate the activity space of urban residents at the traffic zone scale;

[0008] Step two: calculate the access probability of urban residents to traffic zones within their activity space;

[0009] Step three: construct a feature vector of the access mode of urban residents to their activity space;

[0010] Step four: use the feature vector to cluster the access mode of urban residents to their activity space, and use a clustering result evaluation index to determine the final number of clusters;

[0011] Step five: analyze the clustering results using access intensity.

[0012] In step one, a path planning server is built, and the urban residents are travelers using travel tools.

[0013] The path planning server comprises fused OTP, OSM and GTFS data.

[0014] The travel tool is one or more of shared bicycles, online taxis, online rental cars or public buses.

[0015] The calculation method of the access probability: estimating the travel destination of urban residents, aggregating the travel destination of urban residents at the traffic cell scale, and calculating the ratio of the access frequency of urban residents to the traffic cells in the activity space to the total travel frequency.

[0016] The travel destination of the travel tool is estimated by using the network kernel density method.

[0017] The calculation method of the access probability: taking the i-th residential traffic cell as an example, the urban residents in the i-th residential traffic cell have a total of C i trips in the statistical period. i N represents the total number of traffic cells in the activity space of the i-th residential traffic cell, including the residential traffic cell, then:

[0018]

[0019] In the formula, is the access frequency of the i-th residential traffic cell to the n-th traffic cell in the activity space in the statistical period;

[0020] For the urban residents in the i-th residential traffic cell, the access probability of the n-th traffic cell in the activity space in the statistical period is:

[0021]

[0022] The construction method of the feature vector: the access probability of the urban residents to the traffic cells in the activity space is combined to obtain a two-dimensional matrix of MxN, wherein M represents the number of residential traffic cells of the urban residents, and N represents the total number of traffic cells visited by the urban residents in the statistical period;

[0023] Among them, represents the access probability of the urban residents in the i-th residential traffic cell to the traffic cells, represents the probability that the j-th traffic cell is accessed by the urban residents in M residential traffic cells, M i As the characteristics of the access mode of the residents in the i-th residential traffic cell to the activity space, there are N characteristics, and each M i As a sample, there are M samples.

[0024] The two-dimensional matrix obtained in step three is clustered by using a hierarchical clustering algorithm, wherein a criterion parameter is selected as maxclust, and other parameters are selected as default values, and a silhouette coefficient is selected for evaluating the hierarchical clustering result.

[0025] Analysis method of clustering result: the access intensity is defined as the weighted average of the access probability of a city resident in a certain traffic zone in the activity space of the city resident in the access mode; the access intensity is calculated as follows:

[0026]

[0027]

[0028] In the formula, n k is the number of residential traffic zones in the kth access mode, is the total number of times of accessing the ith traffic zone by all residential traffic zones in the kth access mode; total k is the total number of trips of all residential traffic zones in the kth access mode, is an adjustment coefficient:

[0029] The common characteristics of the activity space used by all residents in the kth access mode can be represented by a vector composed of access intensities.

[0030] In the formula, m k is the number of traffic zones in the activity space of the residential traffic zone in the kth access mode; is the access intensity of the ith traffic zone in the activity space accessed by the residential traffic zone in the kth access mode.

[0031] The method has the following advantages:

[0032] First, the access probability of a city resident in a traffic zone in the activity space of the city resident is calculated, a feature vector of the access mode of the city resident is constructed by using the access probability, and clustering is performed, so that the potential access mode of the city resident can be mined. BRIEF DESCRIPTION OF DRAWINGS

[0033] Figure 1 The method is a flowchart of the method for analyzing the access mode of the activity space of a city resident based on an access probability.

[0034] Figure 2The result figure of the activity space of the public transport traveler;

[0035] Figure 3 The result figure of the feature vector of the visiting probability of the public transport traveler to the traffic zones in the activity space thereof;

[0036] Figure 4 The example figure of the two-dimensional array of the visiting probability of the public transport traveler to the traffic zones in the activity space thereof;

[0037] Figure 5 The figure of the relationship between the cluster number and the silhouette coefficient;

[0038] Figure 6 The example figure of the visiting mode clustering result;

[0039] Figure 7 The result figure of the visiting intensity calculation. DETAILED DESCRIPTION

[0040] An activity space visiting mode analysis method based on visiting probability, comprising the following steps:

[0041] Step one: obtaining the activity space of the urban resident through the activity space calculation method on the traffic zone scale;

[0042] For example, taking a public transport traveler in a city as an example, the data used are public transport card data, public transport station data, traffic zone data and road network data;

[0043] Specifically, the construction of the public transport traveling activity space needs to fuse the OTP (Open Trip Planner) and the OSM (OpenStreet Map) and the GTFS data to build a path planning server to complete;

[0044] Further, referring to Figure 2 When calculating the activity space of the public transport traveler, the departure time is selected as the earliest traveling time of the traveler in the residential traffic zone on the day, the time budget is selected as one day, the walking distance is set as 440 meters, and other parameters are selected as default values, so as to calculate the activity space of the residential traffic zone on the day. Similarly, the activity space of the subway traveler, the shared bicycle traveler and the online car-hailing traveler and the like does not deviate from the true range of the present application.

[0045] Step two: estimating the traveling destination of the urban resident, aggregating the traveling destination of the urban resident on the traffic zone scale, calculating the ratio of the visiting frequency of the urban resident to the traffic zone in the activity space thereof to the total traveling frequency, that is, calculating the visiting probability of the urban resident to the traffic zone in the activity space thereof;

[0046] Specifically, taking the i-th residential traffic zone as an example, the urban resident of the i-th residential traffic zone has Ci sub-trip, N i denotes the total number of traffic zones in the activity space of the i-th residential traffic zone, including the residential traffic zone, then:

[0047]

[0048] where, is the visiting frequency of the i-th residential traffic zone to the n-th traffic zone in the activity space in the statistical time period;

[0049] For urban residents in the i-th residential traffic zone, the visiting probability of the n-th traffic zone in the activity space in the statistical time period is:

[0050]

[0051] For example, to calculate the visiting probability of bus travelers to traffic zones in their activity space, first, estimate the bus trip destinations using the network kernel density method; second, aggregate the trip frequencies of bus travelers at the traffic zone scale; finally, calculate the ratio of the visiting frequencies of bus travelers to traffic zones in their activity space to the total trip frequencies, and obtain the visiting probability of the residential traffic zone to each traffic zone in the activity space within a week; reference Figure 3 , shows part of the example of the visiting probability of the 257th traffic zone to traffic zones in its activity space.

[0052] Step three: construct the feature vector of the visiting mode of urban residents to their activity space, combine the visiting probability of urban residents to traffic zones in their activity space, and obtain a two-dimensional matrix of MxN, where M represents the number of residential traffic zones of urban residents, and N represents the total number of traffic zones visited by urban residents in the statistical time period;

[0053] where, denotes the visiting probability of urban residents in the i-th residential traffic zone to traffic zones, denotes the visiting probability of the j-th traffic zone by urban residents in M residential traffic zones, M i As a feature of the visiting mode of residents in the i-th residential traffic zone to their activity space, there are N features, and each M i As a sample, there are M samples.

[0054] For example, construct the feature vector of the visiting mode of bus travelers, combine the visiting probability calculated in step two, and obtain a two-dimensional array of 137x408; reference Figure 4 is part of the example of the two-dimensional array; in a week, the 137 residential traffic zones with trip records visited a total of 408 traffic zones.

[0055] Step four: using the eigenvector obtained in step three, the access patterns of urban residents' activity space are clustered by clustering algorithm, and the final cluster number, i.e. the number of access patterns, is determined using the clustering result evaluation index;

[0056] For example, the access patterns of bus travelers' activity space are clustered; the two-dimensional matrix calculated in step three is clustered using hierarchical clustering algorithm, the criterion parameter in the hierarchical clustering algorithm is selected as maxclust, and the other parameters are selected as default values, the silhouette coefficient is used to evaluate the hierarchical clustering result, and the larger the value of the silhouette coefficient, the better the clustering effect; reference Figure 5 The silhouette coefficient is calculated for 2-137 cluster numbers, and the silhouette coefficient is the largest when the cluster number is 3. That is, the access patterns of bus travelers in the city are divided into 3 categories, and part of the access pattern clustering results are shown as Figure 6 .

[0057] Step five: analyze the clustering results using access intensity, which is defined as the weighted average of the access probability of urban residents in a class of access patterns to a certain traffic zone in their activity space; the access intensity is calculated as follows:

[0058]

[0059]

[0060] In the formula, n k is the number of residential traffic zones in the kth access pattern, is the total number of visits to the ith traffic zone by all residential traffic zones in the kth access pattern; total k is the total number of trips of all residential traffic zones in the kth access pattern, is the adjustment coefficient:

[0061] The common characteristics of the residents in all residential traffic zones in the kth access pattern to their activity space use can be represented by the vector composed of access intensities;

[0062] In the formula, m k is the number of traffic zones in the activity space of the residential traffic zones in the kth access pattern; is the access intensity of the ith traffic zone in the activity space visited by the residential traffic zones of the kth access pattern;

[0063] Specifically, the access intensity of each access pattern is calculated for the access pattern clustering result obtained in step four, and the results are as follows Figure 7As shown in the figure; taking access mode 3 as an example, there are 33 residential traffic zones in access mode 3, which are mainly distributed in the central area of the city; the urban residents in these residential traffic zones generated a total of 41151 trips within a week, and visited a total of 301 destination traffic zones, with an average of 1247 trips per residential traffic zone within a week; the access focus area is concentrated in traffic zones 200-300, i.e. the central area of the city; combined with the distribution of traffic zones visited by bus travelers, it is found that the activity range of bus travelers under this mode is relatively large; combined with the land use type, these traffic zones mainly include central business districts, residential areas, open spaces and some medical service facilities;

[0064] Step six: the access mode analysis process ends.

Claims

1. A method for analyzing access patterns of urban residents' activity spaces based on access probability, characterized in that, The method comprises the following steps: Step 1: Calculate the activity space of urban residents on the traffic block scale; Step 2: Calculate the access probability of urban residents to the traffic blocks in the activity space; The access probability is calculated by estimating the travel destination of urban residents, aggregating the travel destination of urban residents on the traffic block scale, and calculating the ratio of the access frequency of urban residents to the traffic blocks in the activity space to the total travel frequency; Step 3: Construct a feature vector of the access mode of the activity space of urban residents; The two-dimensional matrix obtained in step 3 is clustered by using a hierarchical clustering algorithm, the criterion parameter in the hierarchical clustering algorithm is selected as maxclust, and the other parameters are selected as default values, and the silhouette coefficient is used to evaluate the hierarchical clustering result; Step 4: Cluster the access mode of the activity space of urban residents by using the feature vector, and determine the final cluster number by using a cluster result evaluation index; Step 5: Analyze the clustering result by using the access intensity; The access intensity is defined as the weighted average of the access probability of urban residents in an access mode to a certain traffic block in the activity space, and the access intensity is calculated as follows: ； ； In the formula, It is the first The number of residential traffic zones in the class access pattern. It is the first Access patterns for all residential communities in the same access mode The total number of times traffic zone number 1 was counted; It is the first The sum of all residential community trips in the access pattern. This is the adjustment coefficient.

2. The access mode analysis method of the activity space of urban residents based on the access probability according to claim 1, wherein in step 1, a path planning server is built, and the urban residents are travelers using travel tools.

3. The access mode analysis method of the activity space of urban residents based on the access probability according to claim 2, wherein the path planning server comprises fused OTP, OSM and GTFS data.

4. The access mode analysis method of the activity space of urban residents based on the access probability according to claim 2 or 3, wherein the travel tool is one or more of shared bicycles, online taxis, online rental cars or public buses.

5. The access mode analysis method of the activity space of urban residents based on the access probability according to claim 1, wherein a network kernel density method is used to estimate the travel destination of the travel tool.

6. The method of claim 5, wherein the access probability is calculated by: taking the i-th residential traffic zone as an example, the total number of trips of the urban residents in the i-th residential traffic zone in the statistical time period is the total number of trips of the urban residents in the i-th residential traffic zone in the statistical time period is the total number of trips of the urban residents in the i-th residential traffic zone in the statistical time period is the total number of trips of the urban residents in the i-th residential traffic zone in the statistical time period is the total number of trips of the urban residents in the i-th residential traffic zone in the statistical time period is ； In the formula, is No. of residential traffic zones in the statistical time period for the activity space No. of traffic zones visited; For a city resident living in a traffic cell with number i, the probability of visiting a space in the activity space during the statistical period is: ​ 。 7. The method of claim 1, wherein the feature vector is constructed by combining the access probabilities of the traffic zones within the activity space of the urban resident to obtain a two-dimensional matrix, wherein represents the number of traffic zones in the activity space of the urban resident, and represents the total number of the traffic zones visited by the urban resident within the statistical time. ​​​ wherein, denotes denotes denotes denotes denotes as the characterizes the visiting pattern of the residents in the characterizes the visiting pattern of the residents in the as a sample, there are as a sample, there are 8. The access mode analysis method of the activity space of urban residents based on the access probability according to claim 1, wherein the analysis method of the clustering result is: No. The common characteristics of residents' use of their activity space within all residential traffic zones under the same access pattern can be represented by a vector composed of access intensities. Characterization; wherein, It is the first The number of traffic cells within the residential traffic cell activity space in the class access pattern; It is the first in the activity space Traffic Community No. 1 was the Access intensity of residential area access based on access pattern.

Images