CN109918581B - Method for identifying multiple points of interest and multiple results of user based on space-time data - Google Patents
Method for identifying multiple points of interest and multiple results of user based on space-time data Download PDFInfo
- Publication number
- CN109918581B CN109918581B CN201910168619.6A CN201910168619A CN109918581B CN 109918581 B CN109918581 B CN 109918581B CN 201910168619 A CN201910168619 A CN 201910168619A CN 109918581 B CN109918581 B CN 109918581B
- Authority
- CN
- China
- Prior art keywords
- class
- calculation
- judgment
- time
- user
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Abstract
The invention discloses a method for identifying multiple points of interest and multiple results of a user based on space-time data, which comprises the following steps: 1) Error data filtering and reordering; 2) DBSCAN clustering; 3) Judging a clustering result; 4) A first major class of cyclic calculation and result marking; 5) A second major class of cyclic calculation and result marking; 6) And outputting the feature merging calculation and modeling results. The invention can excavate the family work addresses of most users only through the travel data; the main interest points of the user are mined out through DBSCAN clustering of the main interest points to serve as potential home work addresses. The invention filters all users which do not reach the clustering condition, and reduces the flow and the calculation cost.
Description
Technical Field
The invention relates to a method for identifying multiple points of interest and multiple results of a user, in particular to a method for identifying multiple points of interest and multiple results of a user based on space-time data, and belongs to the technical field of data mining.
Background
In modern life, self-driving travel has become one of the most important means of transportation for people. Along with the development of vehicle-mounted intelligent equipment and intelligent mobile phones, more and more equipment has satellite positioning capability, so that the track recording becomes possible.
The prior art is based on the fact that relevant judgment is carried out by collecting data of all time periods of a user, and the data of all time periods are harder to collect along with importance of the user on privacy. On one hand, the GPS data of the user can be adopted for a long time, and on the other hand, the data volume of the client is large and the privacy information is also large. In the prior art, no commercialized method for directly mining the home work address by using trip data exists.
Disclosure of Invention
The invention aims to provide a method for identifying multiple points of interest and multiple results of a user based on space-time data, which can avoid long-time acquisition of GPS data of the user, and has the advantages of less data quantity and less user privacy information.
In order to achieve the above purpose, the present invention provides the following technical solutions: a method for multi-point-of-interest multi-result recognition for a user based on spatio-temporal data, comprising the steps of:
1) Error data filtering and reordering;
2) DBSCAN clustering;
3) Judging a clustering result;
4) A first general class Loop calculation and result marking (MHWA-Loop 1);
5) A second broad class of Loop computation and result labeling (MHWA-Loop 2);
6) Feature merge computation and modeling result output (MHWA-Model 3).
As a further technical scheme of the invention: the time-space data is complete travel data of the user, each piece of history travel information of the user is contained, each piece of history travel information contains more accurate starting point longitude and latitude, starting point GPS time, ending point longitude and latitude and ending point GPS time, and the time is accurate to time of year, month, day, hour, minute and second.
As a further technical scheme of the invention: in step 1), filtering the user potentially erroneous trip data includes, but is not limited to, filtering users with a small total number of trips and reordering trips that occur in reverse order. The user filtered out because of the smaller total number of passes does not perform any subsequent steps to reduce traffic and computational costs.
As a further technical scheme of the invention: in the step 2), the DBSCAN clustering is to use the existing DBSCAN algorithm to cluster the end points of the users and find out the longitude and latitude coordinates of the potential families or working address classes and class centers which can be used for address mining.
As a further technical scheme of the invention: in the step 3), clustering judgment is carried out according to the clustered result, and according to the conditions of the total residence time, the inter-class distance and the like calculated by each class, the judgment is carried out on the users with the clustering number more than or equal to 2, and the subsequent step in the invention is carried out if the conditions are met;
as a further technical scheme of the invention: in the step 4), calculating related variables of the class with the first rank of the total residence time, starting cyclic comparison calculation from the second class, and outputting judgment conditions; if the corresponding condition is met, such is marked as 1 (default mark is 0), and all classes are calculated in turn.
As a further technical scheme of the invention: in the step 5), a class with a mark of 0 is found in the classes after the first cycle, and if not, the judgment is abandoned. Calculating related variables of the first class of the total residence time ranking of the class marked as 0, starting cyclic comparison calculation from the second class and outputting judgment conditions; if the corresponding condition is met, such is marked as 2 (default marked as 0), and all classes are computed in turn.
As a further technical scheme of the invention: in the step 6), when the feature combination calculation and modeling result output (MHWA-Model) is used, feature variables for modeling are calculated for the classes marked 1 and 2 respectively, and are input into a pre-trained logistic regression Model to output probabilities.
As a further technical scheme of the invention: in the step 6), the following judgment is performed according to the calculated probabilities of the potential workplace and the potential family: if the probability of the potential family land is greater than a certain threshold value, judging that all subclasses under the large class are family addresses; if the probability of the potential working address is greater than a certain threshold, judging all subclasses under the large class as working addresses; if the probabilities of the two potential addresses are smaller than a certain threshold value, the judgment is abandoned.
Compared with the prior art, the invention has the beneficial effects that: the invention can excavate the family work addresses of most users only through the travel data; the main interest points of the user are mined out through DBSCAN clustering of the main interest points to serve as potential home work addresses. The invention filters all users which do not reach the clustering condition, and reduces the flow and the calculation cost. According to the invention, the recognition precision is improved by accurately calculating the related variable and judging the characteristics. The invention establishes a logistic regression model by carrying out cyclic calculation and feature combination on a plurality of clusters, and judges whether the clusters are families or working addresses according to the probability output by the model.
Drawings
FIG. 1 is a flow chart of a method for mining a plurality of home or work addresses with a trip in embodiment 1 of the present invention;
FIG. 2 is a schematic flow chart of the clustering result judgment in embodiment 1 of the present invention;
FIG. 3 is a schematic flow chart of the first general class of Loop calculation and result marking (MHWA-Loop 1) in embodiment 1 of the present invention;
FIG. 4 is a flow chart of the second most general type of Loop calculation and result marking (MHWA-Loop 2) in embodiment 1 of the present invention;
FIG. 5 is a flow chart of the feature merge calculation and modeling result output (MHWA-Model) in embodiment 1 of the present invention.
Detailed Description
The technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention, and it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
A method for multi-point-of-interest multi-result recognition for a user based on spatio-temporal data, comprising the steps of:
1) Error data filtering and reordering;
2) DBSCAN clustering;
3) Judging a clustering result;
4) A first major class of cyclic calculation and result marking;
5) A second major class of cyclic calculation and result marking;
6) And outputting the feature merging calculation and modeling results.
The time-space data is complete travel data of the user, each piece of history travel information of the user is contained, each piece of history travel information contains more accurate starting point longitude and latitude, starting point GPS time, ending point longitude and latitude and ending point GPS time, and the time is accurate to time of year, month, day, hour, minute and second.
In step 1), filtering the user potentially erroneous trip data includes, but is not limited to, filtering users with a small total number of trips and reordering trips that occur in reverse order.
In the step 2), the DBSCAN clustering is to use the existing DBSCAN algorithm to cluster the end points of the users and find out the longitude and latitude coordinates of the potential families or working address classes and class centers which can be used for address mining.
In the step 3), clustering judgment is carried out according to the clustered result, relevant characteristics are calculated for users with the clustering number more than or equal to 2, and whether the subsequent step can be carried out is judged.
In the step 4), calculating related variables of the class with the first rank of the total residence time, starting cyclic comparison calculation from the second class, and outputting judgment conditions; if the corresponding condition is satisfied, such is marked as 1, the default is marked as 0, and all classes are calculated in turn.
In the step 5), a class with a mark of 0 is found in the classes after the first circulation, and if the class is not found, the judgment is abandoned; calculating related variables of the first class of the total residence time ranking of the class marked as 0, starting cyclic comparison calculation from the second class and outputting judgment conditions; if the corresponding condition is met, such is marked as 2, and all classes are computed in turn.
In the step 6), when the feature combination calculation and the modeling result output are used, feature variables for modeling are calculated for the classes with the marks of 1 and 2 respectively, and the feature variables are input into a pre-trained logistic regression model to output probability.
In the step 6), the calculated probabilities of the potential workplace and the potential family are judged as follows: if the probability of the potential family land is greater than a certain threshold value, judging that all subclasses under the large class are family addresses; if the probability of the potential working address is greater than a certain threshold, judging all subclasses under the large class as working addresses; if the probabilities of the two potential addresses are smaller than a certain threshold value, the judgment is abandoned.
Example 1
Referring to fig. 1, in an embodiment of the present invention, a method for identifying multiple points of interest and multiple results of a user based on spatio-temporal data includes the following steps:
step S10, error data filtering and reordering.
In this embodiment, coarse filtering and sorting are performed on the input user travel data, including removing data with excessive mileage deviation due to the reason of GPS positioning accuracy, removing data with abnormal uploading time, removing data with longitude and latitude not in the china range, and so on. To prevent the reverse order of the user trip data, i.e., the start time of the n+1th trip is earlier than the n-th trip, the user data is ordered according to the trip start time. If the total travel number of the user is less than 5, the subsequent steps are not performed, and the judgment is abandoned.
Step S20, DBSCAN clustering.
In this embodiment, the existing DBSCAN method is directly used to cluster the end points of all the strokes of the user in the past 3 months, the set cluster radius threshold is r, and the minimum point threshold in the cluster range is c.
The specifically defined above threshold values include:
1) Clustering radius threshold 500 meters;
2) The minimum number of points in the clustering range is 5 points.
And S30, clustering judgment.
As shown in fig. 2, the clustering judgment includes the following steps:
s3001: if the clustering is more than or equal to 2, continuing the subsequent steps, otherwise, giving up judgment, and ending the algorithm;
s3002: calculating the total residence time of each type, and sorting according to the order from big to small;
s3003: taking two digits with the maximum ranking of the total residence time, and entering the subsequent step if the sum of the two digits is more than 8 hours;
s3004: if the center cluster of the two classes is more than 500km, the step S40 is entered;
s3005: if three classes exist, the total residence time difference of the second class and the third class is less than 10%, the step S40 is entered, otherwise, the subsequent step is entered;
s3006: if the third total residence time is greater than 10% of the total residence time of all classes, step S40 is entered, otherwise the decision is abandoned and the algorithm ends.
In step S40, the first broad class loops calculate and result tag (MHWA-Loop 1).
As shown in FIG. 3, the first broad class Loop calculation and result marking (MHWA-Loop 1) comprises the steps of:
s4001: and calculating variables of the first class. The variable computation of the class comprises the following four steps:
s40011: converting all end times of the class into time of day in seconds;
s40012: calculating 10% and 90% quantiles of the time;
s40013: calculating the time belonging to the day of the week and counting the times;
s40014: the 10% and 90% quantile values and the maximum number of times of two days are output.
S4002: starting calculation from the class with the second rank of total residence time;
s4003: the condition judgment and calculation comprises the following four steps:
s40031: the total residence time of the class is more than 4 hours, otherwise, the judgment is abandoned, and the default mark is 0;
s40032: counting the number of times the class ending time falls between 10% and 90% of the quantiles of the class ending time as condition one
S40033: calculate the first two days of the week with the highest number of times of this category
S40034: judging whether the calculated results of the first class related variables are consistent with the calculated results of the first class related variables or not, and outputting corresponding condition judgment for the subsequent steps.
S4004: judging whether the first condition is larger than 0.5 or not, and judging whether the second condition is met or not;
s4005: the class satisfying the condition is marked as 1;
s4006: and (5) circularly calculating and judging each class, and after the calculation is finished, entering step S50.
In step S50, a second broad class loops computation and result marking (MHWA-Loop 2).
As shown in FIG. 4, the second broad class of Loop computation and result marking (MHWA-Loop 2) comprises the steps of:
s5001: finding a class marked as 0, and counting times;
s5002: if the number of times is more than 0, entering the subsequent step;
s5003: marking the first class as 2, and marking the default as 0;
performing variable calculation in the step S4001 on the first class;
s5004: calculating from the second class;
entering the condition calculation and judgment in the step S4002;
s5005: judging whether the first condition is larger than 0.5 or not, and judging whether the second condition is met or not;
s5006: the class meeting the condition is marked as 2;
s5007: and (5) circularly calculating and judging each class, and after the calculation is finished, entering step S60.
In step S60, the feature is combined with the calculation and modeling result output (MHWA-Model).
As shown in FIG. 5, the feature merge calculation and modeling result output (MHWA-Model) comprises the steps of:
s6001: combining and calculating variables marked as 1 and 2, and averaging;
s6002: the calculated features include: step S6003 is executed, including a total residence time ratio, a trip number ratio, a non-working day residence time ratio, and the like;
s6003: the characteristic variable is brought into a logistic regression model, the probability p is calculated, and the step S6004 is executed;
s6004: and judging whether the output probability p is more than or equal to 0.6. If yes, the first class of total stay time ranking is the home address of the user, and the second class of total stay time ranking is the work address of the user; otherwise, step S6005 is performed;
s6005: and judging whether the output probability p is smaller than or equal to 0.5. If yes, the second major class of the total stay time rank is the place where the home address of the user is located, and the first major class of the total stay time rank is the place where the work address of the user is located; otherwise, the judgment is abandoned.
In the method for mining a plurality of families or working addresses by using the journey in the embodiment, firstly, filtering and clustering according to the journey points to find potential families and working addresses, then, carrying out clustering judgment and accurate calculation and judgment of related variables and characteristics, and finally, judging that the clustering is the family or the working address through a logistic regression model. Compared with the prior art, the method of mining a plurality of families or work addresses with a journey in the present embodiment solves the problem that the data of the whole period must be used to obtain the user family work address. In the age when users pay more and more attention to data privacy, the general home address or work address of the users can be found by using the trip data to provide support for later user labels and user portraits.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.
Furthermore, it should be understood that although the present disclosure describes embodiments, not every embodiment is provided with a separate embodiment, and that this description is provided for clarity only, and that the disclosure is not limited to the embodiments described in detail below, and that the embodiments described in the examples may be combined as appropriate to form other embodiments that will be apparent to those skilled in the art.
Claims (1)
1. A method for multi-point-of-interest multi-result recognition of a user based on spatio-temporal data, comprising the steps of:
1) Error data filtering and reordering; in the step 1), filtering the user potentially wrong trip data includes, but is not limited to, filtering the user with a small total trip number and reordering the trips with reverse order;
2) DBSCAN clustering; in the step 2), the DBSCAN clustering is to cluster the end points of the users by using the existing DBSCAN algorithm, and find out the longitude and latitude coordinates of potential families or working address classes and class centers which can be used for address mining;
3) Judging a clustering result; in the step 3), clustering judgment is carried out according to the clustered result, relevant characteristics are calculated for users with the clustering number more than or equal to 2, and whether the subsequent step can be carried out is judged;
the clustering judgment comprises the following steps:
s3001: if the clustering is more than or equal to 2, continuing the subsequent steps, otherwise, giving up judgment, and ending the algorithm;
s3002: calculating the total residence time of each type, and sorting according to the order from big to small;
s3003: taking two digits with the maximum ranking of the total residence time, and entering the subsequent step if the sum of the two digits is more than 8 hours;
s3004: if the center cluster of the two classes is more than 500km, the step S40 is entered;
s3005: if three classes exist, the total residence time difference of the second class and the third class is less than 10%, the step S40 is entered, otherwise, the subsequent step is entered;
s3006: if the total residence time of the third name is more than 10% of the total residence time of all classes, the step S40 is entered, otherwise, the judgment is abandoned, and the algorithm is ended;
4) A first major class of cyclic calculation and result marking; in the step 4), calculating related variables of the class with the first rank of the total residence time, starting cyclic comparison calculation from the second class, and outputting judgment conditions; if the corresponding condition is met, marking the class as 1, marking the default as 0, and sequentially calculating all the classes;
the first general class of loop calculation and result marking comprises the following steps:
s4001: the variable calculation is carried out on the first class, and the variable calculation comprises the following four steps:
s40011: converting all end times of the class into time of day in seconds;
s40012: calculating 10% and 90% quantiles of the time;
s40013: calculating the time belonging to the day of the week and counting the times;
s40014: outputting 10% and 90% quantile values and the maximum number of times of two weeks;
s4002: starting calculation from the class with the second rank of total residence time;
s4003: the condition judgment and calculation comprises the following four steps:
s40031: the total residence time of the class is more than 4 hours, otherwise, the judgment is abandoned, and the default mark is 0;
s40032: counting the number of times the class ending time falls between 10% and 90% of the quantiles of the class ending time as condition one
S40033: calculate the first two days of the week with the highest number of times of this category
S40034: judging whether the calculated results of the first class related variables are consistent with the calculated results of the first class related variables or not, and outputting corresponding condition judgment for the subsequent steps;
s4004: judging whether the first condition is larger than 0.5 or not, and judging whether the second condition is met or not;
s4005: the class satisfying the condition is marked as 1;
s4006: each class is judged through cyclic calculation, and step S50 is carried out after calculation is completed;
5) A second major class of cyclic calculation and result marking; in the step 5), a class with a mark of 0 is found in the classes after the first circulation, and if the class is not found, the judgment is abandoned; calculating related variables of the first class of the total residence time ranking of the class marked as 0, starting cyclic comparison calculation from the second class and outputting judgment conditions; if the corresponding condition is met, marking the class as 2, and sequentially calculating all classes;
the second broad class of loop computation and result labeling comprises the steps of:
s5001: finding a class marked as 0, and counting times;
s5002: if the number of times is more than 0, entering the subsequent step;
s5003: marking the first class as 2, and marking the default as 0;
performing variable calculation in the step S4001 on the first class;
s5004: calculating from the second class;
entering the condition calculation and judgment in the step S4002;
s5005: judging whether the first condition is larger than 0.5 or not, and judging whether the second condition is met or not;
s5006: the class meeting the condition is marked as 2;
s5007: each class is judged through cyclic calculation, and step S60 is carried out after calculation is completed;
6) And 6) outputting a feature combination calculation and modeling result, wherein in the step 6), when the feature combination calculation and modeling result is used, feature variables for modeling are calculated for classes marked as 1 and 2 respectively, the feature variables are input into a pre-trained logistic regression model, and probability is output, and in the step 6), the calculated probabilities of potential workplace and potential family are judged as follows: if the probability of the potential family land is greater than a certain threshold value, judging that all subclasses under the large class are family addresses; if the probability of the potential working address is greater than a certain threshold, judging all subclasses under the large class as working addresses; if the probability of the two potential addresses is smaller than a certain threshold value, discarding the judgment;
the feature merging calculation and modeling result output comprises the following steps:
s6001: combining and calculating variables marked as 1 and 2, and averaging;
s6002: the calculated features include: step S6003 is executed, including a total residence time ratio, a trip number ratio, a non-working day residence time ratio, and the like;
s6003: the characteristic variable is brought into a logistic regression model, the probability p is calculated, and the step S6004 is executed;
s6004: judging whether the output probability p is greater than or equal to 0.6, if so, the first major class of total stay time ranking is the place where the home address of the user is located, and the second major class of total stay time ranking is the place where the work address of the user is located; otherwise, step S6005 is performed;
s6005: judging whether the output probability p is smaller than or equal to 0.5, if so, the second major class of the total stay time ranking is the place where the home address of the user is located, and the first major class of the total stay time ranking is the place where the working address of the user is located; otherwise, discarding the judgment;
the time-space data is complete travel data of the user, each piece of history travel information of the user is contained, each piece of history travel information contains more accurate starting point longitude and latitude, starting point GPS time, ending point longitude and latitude and ending point GPS time, and the time is accurate to time of year, month, day, hour, minute and second.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910168619.6A CN109918581B (en) | 2019-03-06 | 2019-03-06 | Method for identifying multiple points of interest and multiple results of user based on space-time data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910168619.6A CN109918581B (en) | 2019-03-06 | 2019-03-06 | Method for identifying multiple points of interest and multiple results of user based on space-time data |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109918581A CN109918581A (en) | 2019-06-21 |
| CN109918581B true CN109918581B (en) | 2023-09-22 |
Family
ID=66963634
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910168619.6A Active CN109918581B (en) | 2019-03-06 | 2019-03-06 | Method for identifying multiple points of interest and multiple results of user based on space-time data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109918581B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111178932A (en) * | 2019-11-26 | 2020-05-19 | 深圳壹账通智能科技有限公司 | User geographic portrait generation method and device, computer equipment and storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6449612B1 (en) * | 1998-03-17 | 2002-09-10 | Microsoft Corporation | Varying cluster number in a scalable clustering system for use with large databases |
| CN105631475A (en) * | 2015-12-25 | 2016-06-01 | 石成富 | Computer data mining and clustering method based on time sequence |
| CN106528597A (en) * | 2016-09-23 | 2017-03-22 | 百度在线网络技术（北京）有限公司 | POI (Point Of Interest) labeling method and device |
| CN106649540A (en) * | 2016-10-26 | 2017-05-10 | Tcl集团股份有限公司 | Video recommendation method and system |
| CN107909344A (en) * | 2017-11-21 | 2018-04-13 | 杭州电子科技大学 | Workflow logs iterative task recognition methods based on relational matrix |
| CN109034187A (en) * | 2018-06-12 | 2018-12-18 | 上海中通吉网络技术有限公司 | A kind of subscriber household work address excavation process |
| CN109086323A (en) * | 2018-06-28 | 2018-12-25 | 上海中通吉网络技术有限公司 | The determination method and system of subscriber household and work address |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110313954A1 (en) * | 2010-06-18 | 2011-12-22 | Microsoft Corporation | Community model based point of interest local search |
-
2019
- 2019-03-06 CN CN201910168619.6A patent/CN109918581B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6449612B1 (en) * | 1998-03-17 | 2002-09-10 | Microsoft Corporation | Varying cluster number in a scalable clustering system for use with large databases |
| CN105631475A (en) * | 2015-12-25 | 2016-06-01 | 石成富 | Computer data mining and clustering method based on time sequence |
| CN106528597A (en) * | 2016-09-23 | 2017-03-22 | 百度在线网络技术（北京）有限公司 | POI (Point Of Interest) labeling method and device |
| CN106649540A (en) * | 2016-10-26 | 2017-05-10 | Tcl集团股份有限公司 | Video recommendation method and system |
| CN107909344A (en) * | 2017-11-21 | 2018-04-13 | 杭州电子科技大学 | Workflow logs iterative task recognition methods based on relational matrix |
| CN109034187A (en) * | 2018-06-12 | 2018-12-18 | 上海中通吉网络技术有限公司 | A kind of subscriber household work address excavation process |
| CN109086323A (en) * | 2018-06-28 | 2018-12-25 | 上海中通吉网络技术有限公司 | The determination method and system of subscriber household and work address |
Non-Patent Citations (3)
| Title |
|---|
| Dong Liang ; Zhaojing Zhang ; Mugen Peng ; .Access Point Reselection and Adaptive Cluster Splitting-Based Indoor Localization in Wireless Local Area Networks.IEEE Internet of Things Journal.2015,第573-585页. * |
| 基于时空密度算法的用户轨迹数据兴趣区域发现;周新丽;桑梓森;张越;;中国科技论文(08);第75-80页 * |
| 张铁映 ; 李宏伟 ; 许栋浩 ; 孟超越 ; 朱燕 ; .采用密度聚类算法的兴趣点数据可视化方法.测绘科学.2016,第157-162页. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109918581A (en) | 2019-06-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105354196B (en) | Information-pushing method and information push-delivery apparatus | |
| CN108346292B (en) | Urban expressway real-time traffic index calculation method based on checkpoint data | |
| CN105809292B (en) | Bus IC card passenger getting off car website projectional technique | |
| US9313616B2 (en) | System and method for automated identification of location types for geofences | |
| EP2842108B1 (en) | System and method for tracking driver hours and timekeeping | |
| CN104854884B (en) | Method and system for labeling visited locations based on contact information | |
| US20110307359A1 (en) | Systems and methods for managing address and tax inventory data | |
| CN111770447B (en) | Method and device for generating electronic fence and server | |
| CN108279017B (en) | Method for calculating and adding via points in real time in navigation process | |
| CN103646312A (en) | Public bicycle information reading method based on two-dimension code | |
| CN110046218B (en) | Mining method, device and system for user travel mode and processor | |
| CN109918581B (en) | Method for identifying multiple points of interest and multiple results of user based on space-time data | |
| CN103942312A (en) | Public transportation transfer line planning method and device | |
| WO2015177858A1 (en) | Trip attribute estimating system, trip attribute estimating method, trip attribute estimating program, and travel behavior survey system | |
| CN105373582A (en) | Government affair service guiding method and system | |
| CN114501336B (en) | Road traffic volume measuring and calculating method and device, electronic equipment and storage medium | |
| CN109919225B (en) | Method for identifying user interest points based on space-time data | |
| DE112021001926T5 (en) | SYSTEM AND METHOD FOR FILTERLESS THrottling OF VEHICLE EVENT DATA PROCESSING TO IDENTIFY PARKING AREAS | |
| CN109918582B (en) | Method for identifying single interest point of user based on space-time data | |
| CN101155380B (en) | Integrating system and method for wireless network test data | |
| CN109615727B (en) | Riding endpoint extraction method and system based on shared bicycle static GPS data | |
| CN109766440B (en) | Method and system for determining default classification information for object text description | |
| Pavlyuk et al. | Spatiotemporal dynamics of public transport demand: a case study of Riga | |
| CN113139740A (en) | Comprehensive information management service system suitable for small-scale operator | |
| CN101950324A (en) | River-health intelligent diagnosis method and equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |