CN110837540A - Method and system for processing spatial position data - Google Patents

Abstract

The invention discloses a method and a system for processing spatial position data. First, quality evaluation on spatio-temporal sampling needs to be performed on the sparsely sampled spatial position big data. Therefore, the present invention proposes a sparsely sampled spatial location big data through dynamic time period division, computer random simulation and calculation, and regression analysis, starting from the basic feature of sparsely sampled individual location big data, which is the inherent spatiotemporal distribution. A quantitative loss assessment model that depicts the typical characteristics of urban residents' activities, and clearly explains the quality loss distribution characteristics corresponding to different sampling rates. Finally, the sample data set that meets the actual needs is selected according to the distribution of the obtained deviation, so that the conclusions of related research and analysis can be more reliable and more scientific to guide the practical application.

Description

A method and system for processing spatial position data

technical field

The present invention relates to the technical field of spatial position data processing, and more particularly, to a method and system for processing spatial position data.

Background technique

The rapid development of technologies such as communication and information technology and location-aware technology has facilitated the collection of spatial location movement data of large-scale individuals and the sharing of information in a relatively low-cost, large-scale, and very fast manner, with time stamps and spatial Big data of location coordinates is already at your fingertips. The spatiotemporal big data is applied in the fields of urban residents’ mobility dynamics analysis, spatiotemporal pattern mining, traffic analysis, and urban planning. While providing new research perspectives, in order to more effectively and reasonably answer the questions that need to be studied, in the analysis process. There will be data quality issues.

Sparsely sampled location big data is an important data source in the current research and application of urban spatial analysis, such as mobile phone signaling spatiotemporal location data, check-in spatiotemporal location data, etc. One is that mobile communication equipment is widely used among urban residents, has a high penetration rate, and is carried around by users and has a long use time; the other is that the communication base stations in cities have the characteristics of large-scale and high-density establishment. However, the location big data such as the above also have the characteristics of randomness, spatio-temporal sparsity and uncertainty in the sampling location of individuals to a certain extent. Most of the data are simply screened, and the quality of the data is less considered, so the uncertainty brought by the current analysis and application work cannot be measured.

SUMMARY OF THE INVENTION

Aiming at the defects of the prior art, the purpose of the present invention is to solve the problems in the current research and application of location-based big data. In the data preprocessing link, the data is mostly simply screened, and the quality of the data is less considered, so as to analyze the current situation. And the uncertainty brought about by the application work is also not measurable technical issues.

In order to achieve the above object, in a first aspect, the present invention provides a method for processing spatial position data, comprising the following steps:

Step 1, establish an urban spatial database, import sparsely sampled spatial location data to the urban spatial database; divide the sparsely sampled spatial location data into spatial location data covering the entire period and spatial location data covering part of the period;

Step 2: Divide the spatial position data covering the whole period according to the period, and divide it into the time series spatial position data of M periods; M is a positive integer greater than or equal to 2;

Step 3, randomly select the spatial position data of C groups of m time periods from the time series space position data of the M time periods, and the spatial position data of each time period in the time series space position data of the M time periods is in the C group m. The spatial location data of the time period are selected at least k times; the initial value of m is 2, and m is a positive integer less than M; k is a positive integer less than or equal to C, and C is a positive integer;

Step 4: Calculate the index value of each user corresponding to the spatial location data of each group of m time periods, and average the C group index values of each user in m time periods as the index of each user in m time periods and calculating the index value of each user in the whole time period corresponding to the time-series space position data of M time periods; the index value includes: the user's space activity range, the user's activity path length within the space activity range, and Differences and imbalances of users in different spatial positions within the spatial activity range;

Step 5. Determine the index value deviation corresponding to m time periods according to the index value of each user in m time periods and the index value of each user in the whole time period, and based on the index value deviation corresponding to each user in m time periods Determine the quality loss coefficient of the index value of each user in m time periods;

Step 6. If m=M, go to step 7, if m is less than M, add 1 to m as a new value of m, go to step 3;

Step 7: Determine the number of users covering the spatial location data of m time periods in the spatial location data covering part of the time period, the number of users covering the spatial location data of m time periods determined according to the spatial location data covering part of the time period, The weighted quality loss coefficient of the sparsely sampled spatial position data is determined by the quality loss coefficient of the index value of each user in m time periods and the number of all users determined by the spatial position data covering the whole period, 2≤m≤M.

In an optional embodiment, the step 1 specifically includes the following steps:

The sparsely sampled spatial location data is imported into the urban spatial database, and each spatial location data is converted into a preset coordinate system, and each spatial location data includes sampling coordinates and sampling time.

In an optional embodiment, the step 2 specifically includes the following steps:

The spatial position data covering the whole time period is divided into spatial position data of M time periods according to actual requirements or in an adaptive manner.

In an optional embodiment, the index value specifically includes:

The space activity range index is the radius of gyration R _g :

The index of the activity path length in the space activity range is the moving distance S:

The index of access difference and imbalance at different spatial locations within the scope of spatial activities is entropy E:

Among them, n is the spatial location data of each combination of m time periods or the total number of spatial location sampling points of each user in the spatial location data of M time periods, and (x _j , y _j ) is the jth sampling point of each user. Coordinate value, (x _c , y _c ) is the center of gravity of all sampling points for each user, n′ is the number of different sampling positions for each user, p _i is the probability of the i-th different sampling point for each user appearing ;

The calculation formula of the center of gravity (x _c , y _c ) of all sampling point positions of each user is:

In an optional embodiment, the step 5 specifically includes the following steps:

According to the deviation measurement model, the quality loss coefficient of the index value of each user under the number m of each time period is obtained, and the deviation measurement model is:

F _m (X _u )=A _m X _u -B

Among them, F _m (X _u ) represents the index value of each user u under the number _m of each time period, X _u represents the index value of each user in the whole time period, Am is the regression coefficient, and B is a constant;

The mass loss coefficient _QLm is determined by the following formula:

QL _m =1-|A _m |

Among them, |A _m | is the absolute value of the coefficient A _m ; the regression coefficient corresponding to each index value of each user in each time period m is brought into the above formula, and each user can be obtained in each time period. The mass loss coefficients QL _{m_Rg} , QL _{m_S} and QL _{m_E} corresponding to the gyration radius index value, the moving distance index value and the entropy index value.

In an optional embodiment, the step 7 specifically includes the following steps:

The quality loss coefficient w _QL corresponding to each index value of the sparsely sampled spatial location data is determined by the following formula:

Among them, users _m represents the number of users who cover the spatial location data of m time periods in the spatial location data covering part of the time period, users represents the number of all users; QL _m respectively represents the quality loss of the index value of each user in m time periods Coefficient, QL _m specifically includes: QL _{m_Rg} , QL _{m_S} or QL _{m_E} ;

The weighted quality loss coefficient W _QL of the sparsely sampled spatial location data is calculated by calculating the mass loss coefficients w _{QL_Rg} , w _{QL_S} and w _{QL_E} of the radius of gyration, moving distance and entropy, respectively:

In a second aspect, the present invention provides a processing system for spatial position data, comprising:

A data sampling unit, used for establishing an urban spatial database, and importing sparsely sampled spatial location data into the urban spatial database; dividing the sparsely sampled spatial location data into spatial location data covering the entire time period and spatial location covering part of the time period data;

A full-period data processing unit, configured to divide the spatial position data covering the full period of time according to periods, and divide it into time-series spatial position data of M periods; M is a positive integer greater than or equal to 2; The spatial position data of C groups of m time periods are randomly selected from the time-series spatial position data of the C groups, and the spatial position data of each time period in the time-series space position data of the M time periods are at least equal to the spatial position data of the C groups of m time periods. Select more than k times; the initial value of m is 2, m is a positive integer less than M; k is a positive integer less than or equal to C, and C is a positive integer; calculate the spatial location data of each group of m time periods corresponding to each user's index value, and average the C group index values of each user in m time periods as the index value of each user in m time periods; The index value under the time period; the index value includes: the user's space activity range, the user's activity path length within the space activity range, and the difference and imbalance of the user's different spatial positions within the space activity range ; Determine the index value deviation corresponding to m time periods according to the index value of each user in m time periods and the index value of each user in the whole time period, and determine the index value deviation corresponding to each user in m time periods. The quality loss coefficient of the index value of the number of users in m time periods; if m=M, end the process, if m is less than M, add 1 to m as the new m value, and continue from the time series space of the M time periods Randomly select the spatial position data of C groups of m time periods in the position data, to determine the quality loss coefficient of each user index value under the numerical value m of different time periods according to the spatial position data covering the whole period;

A part-time-period data processing unit, configured to determine the number of users covering the spatial position data of m time periods in the spatial position data covering part of the period; m is an integer from 2 to M respectively;

A data quality evaluation unit, configured to determine the number of users of the spatial location data covering m time periods according to the spatial location data covering part of the time period, and each user determined by the spatial location data covering the whole time period is under m time periods The quality loss coefficient of the index value and the number of all users determine the weighted quality loss coefficient of the sparsely sampled spatial location data, 2≤m≤M.

In an optional embodiment, the full-time data processing unit, according to the deviation measurement model, obtains the quality loss coefficient of the index value of each user under the number m of each time period, and the deviation measurement model is: F _m ( X _u )=A _m X _u -B; wherein, F _m (X _u ) represents the index value of each user u under the number m of each time period, X _u represents the index value of each user in the whole time period, A _m is the regression coefficient, and B is a constant; the quality loss coefficient QL _m is determined by the following formula: QL _m =1-|A _m |; where |A _m | is the absolute value of the coefficient _Am ; The regression coefficient corresponding to each index value under the number m of time periods is brought into the above formula, and the quality loss coefficient QL corresponding to the index value of the gyration radius, the index value of moving distance and the index value of entropy for each user under the number m of time periods can be obtained respectively. _{m_Rg} , QL _{m_S} and QL _{m_E} .

In an optional embodiment, the data quality evaluation unit determines the quality loss coefficient w _QL corresponding to each index value of the sparsely sampled spatial location data by the following formula:

Among them, users _m represents the number of users who cover the spatial location data of m time periods in the spatial location data covering part of the time period, users represents the number of all users; QL _m respectively represents the quality loss of the index value of each user in m time periods Coefficient, QL _m specifically includes: QL _{m_Rg} , QL _{m_S} or QL _{m_E} ;

The weighted quality loss coefficient W _QL of the sparsely sampled spatial location data is calculated by calculating the mass loss coefficients w _{QL_Rg} , w _{QL_S} and w _{QL_E} of the radius of gyration, moving distance and entropy, respectively:

In a third aspect, the present invention provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, realizes the processing of the spatial location data according to the first aspect above. Approach.

In general, compared with the prior art, the above technical solutions conceived by the present invention have the following beneficial effects:

The present invention provides a method and system for processing spatial position data, which can overcome the problems of randomness, sparseness and the resulting uncertainty in the sampling process of the space-time position data. Based on the evaluation of the quality loss coefficient of the position data corresponding to each time period and different combination methods proposed in the present invention, the sampling of the spatiotemporal position data with deterministic quality and density in the current practical application can be directly guided.

The present invention provides a method and system for processing spatial position data, and a quantitative evaluation model for the quality of spatial position data proposed. The calculation process proposed by the present invention fully considers the influence of different sampling features on the evaluation results, which not only effectively guarantees the unbiasedness of the calculation process, but also fills the gap in the current research and application of location-based big data related to data quality evaluation research.

The invention provides a processing method and system for spatial position data, and the data sampling method is scientific and evidence-based. Based on the quality loss assessment result proposed in the present invention, data with different reliability and quality can be selected intuitively and effectively for urban spatial analysis, so as to adapt to the data conditions.

The present invention provides a method and system for processing spatial position data, which have a wide range of applications. The evaluation model and method proposed in the present invention can be used for various types of spatiotemporal big data of individual location sparse sampling, such as mobile phone signaling spatiotemporal data, social media check-in spatiotemporal data, credit card consumption record spatiotemporal data, and the like.

The present invention provides a method and system for processing spatial position data, which can customize data on demand. The loss estimation evolution rule curve and the calculation method of the typical index proposed by the present invention can calculate the deviation rule of various types of sparse sampling location big data, and has intuitive and scientific guidance for selecting and customizing the data set for specific needs.

The present invention provides a method and system for processing spatial position data, which saves costs. The present invention does not require additional large-scale equipment and equipment, does not need to spend a lot of manpower and material resources for investigation, and only requires a small number of staff for maintenance, but makes full use of the sampling characteristics of the data itself to evaluate the quality of the data.

Description of drawings

1 is a flowchart of a method for processing spatial position data provided by the present invention;

2 is a schematic diagram of a resident activity trajectory reflected by the sparsely sampled location big data provided by the present invention;

Fig. 3 is a mass loss coefficient fitting curve diagram provided by the present invention;

FIG. 4 is an architecture diagram of a processing system for spatial location data provided by the present invention.

Detailed ways

In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, but not to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

Aiming at the defects of non-standardization and uncertainty in preprocessing and use of existing spatial location big data, the present invention introduces the basic format of spatial location big data and its existing sparse sampling problem, and uses data sampling points on this basis. Space-time distribution is an intuitive and effective inherent feature. The time period in the sampling distribution is divided by dynamic time window, and a deviation measurement model for sparse sampling data to describe the characteristics of urban residents' activities is proposed. This paper draws out the general conclusions of the spatial location big data to describe the deviation of urban residents' activity characteristics, and clearly explains the deviation distribution characteristics corresponding to different sampling rates. Finally, the sample data set that meets the actual needs is selected according to the distribution of the obtained deviations, which makes the conclusions of related research and analysis more reliable and guides practical applications more scientifically.

The invention proposes a quantitative and effective sparse sampling location big data quality assessment and sampling method to solve the uncertainty problem brought by the large-scale sparse sampling spatial location big data in the analysis and application of urban residents' activities.

The technical solution of the present invention is a quality assessment and sampling method for sparsely sampled spatial location big data, as shown in FIG. 1 , including the following steps:

Step S101, establishing an urban spatial database, and importing sparsely sampled spatial location big data into the database;

In one example, a unique ID is assigned to each resident individual, and the individual ID field is set as a commonly used index field; the spatial data imported into the database are converted into a coordinate system based on the 2000 National Geodetic Coordinate System; query separately Individual ID location big data, and sequentially construct the time series of sampling points of each resident individual movement trajectory, {P ₁ ,P ₂ ,...,P _n }, where P _i (x _i ,y _i ,t _i ) is the i-th (x _i , y _i ) is the geodetic coordinate value, _xi is the abscissa, y _i is the ordinate, and t _i is the vertical coordinate of the third dimension, which is represented by time; the introduction of 2000 national geodetic coordinates The system is used as a two-dimensional datum. The geodetic coordinate values of each sampling point in the two-dimensional datum generate a Voronoi polygon, and a space-time cuboid is constructed with time as the third-dimensional vertical axis. In the space-time cube, the time series of sampling points is used to restore each resident. The spatiotemporal trajectory of the individual; as shown in Figure 2, it is a schematic diagram of the resident activity trajectory reflected by the sparsely sampled location big data. According to the time series of each individual trajectory sampling point and the recovered space-time trajectory, the natural day is linearly divided into time periods.

It should be noted that the size of the time window used in the time period division process proposed in the present invention is adaptive according to actual requirements or the density distribution of data recording sampling points, and the number of time periods obtained by division is M, and M is greater than or equal to 2. ; and match the individual samples covered by the whole time period from the data set, that is, the individuals with sampling positions in each time period obtained from the division are used as sub-data set D, and the total number of individuals included is N; other individuals are used as data set D '.

It can be understood that the sub-data set D is the spatial position data covering the whole time period, and the data set D' is the spatial position data covering a part of the time period. Among them, the spatial location data covering the whole time period means that each time period in this part of the data contains the location data of all the sampling individuals; the spatial location data covering part of the time period means that no time period in this part of the data contains all the sampling individuals of the location data, that is, not every time period in this part of the data contains the spatial location data of each sampled individual. Specifically, the sampled individual here is each resident or each user.

Optionally, typical indicators that characterize the activities of urban residents, such as the radius of gyration R _g , the moving distance S and the entropy E, can be introduced, and the full-time typical indicator values {R _gu , S _u , E _u } of each individual u can be calculated, where u=1,2,.......,N;

Typical indicators: the radius of gyration R _g is used to describe the spatial activity range of residents; the moving distance S is used to describe the length of the residents’ activity paths in the space; the entropy E is used to describe the residents’ visits to different spatial locations in the space. Differences and Imbalances;

Turning Radius:

Moving distance:

entropy:

In the above calculation formula, n is the total number of locations visited by each individual, (x _j , y _j ) is the geodetic coordinate value of the j-th sampling point, and (x _c , y _c ) is the location of all sampling points for each individual The center of gravity, n' is the number of different sampling positions, and p _i is the probability of the i-th different sampling point appearing;

The centroid (x _c , y _c ) of all sampling point positions of each individual is calculated as:

Step S102: From the sub-data set D obtained by matching, sequentially select the location records covering m (2≤m≤M, m=2,3,...,M) time periods, and calculate the position records under m time periods respectively. Typical index values of the sampling period of each individual u {R _gmu' ,S _mu' ,E _mu' }, m=2,3,...,M;

In one embodiment, the rules to be followed for selecting m time periods proposed by the present invention are as follows:

For a specified number of time periods, the number of random selections is greater than or equal to C times (C is determined according to actual needs or data characteristics, and the recommended default value is 1000); and the time period combination method in each random case occurs only once, and their are independent of each other; if the specified number of time periods cannot meet the requirements of the number of random selections, all selections are carried out according to the upper limit of the theoretical combination method, and the combination method only appears once, and they are independent of each other; thus, the random selection can be obtained. Random times C.

Specifically, for the specified time periods of the same number, after C random times, it should be ensured that each time period must be selected at least k times or more (k depends on actual needs or data characteristics, and the recommended default value is 10).

Specifically, in the random process, each time period will be selected and selected in a relatively uniform distribution manner to ensure a relatively balanced appearance of the time periods in the random process. For example, when the number of divided time periods is M=24, and the number of selected time periods is m=3, if the time periods are combined (#2, #5, #6), (#2, #6, #7), (# 2,#7,#16),(#2,#7,#19),(#2,#5,#7),(#2,#6,#11),(#2,#6,# 12), (#2, #7, #17), (#2, #7, #18) and (#2, #7, #22) have been selected, so the #2 period has been selected 10 times , but the periods #5, #6, #7, #11, #12, #16, #17, #18, #19 and #22 did not appear k times in these combinations, and the other 13 periods also does not appear, so in the remaining C-k random selections, priority will be given to the period that has not been selected for k times or more;

It can be understood that the rules proposed above guarantee the breadth and depth of random numbers, the unbiasedness and equilibrium of random processes, etc.

Finally, the user location records within the number m of time periods randomly selected each time are calculated, and a time-series location sequence is reconstituted.

Step S103: Calculate the mean value of the typical index values {R _g(m,c)u' , S _(m,c)u' , E _(m,c)u' } of the sampling period of each individual u under m time periods respectively , m=2,3,...,M; c=1,2,...,C. The number of time periods m should increase from 2 to M in turn; each time period is randomly C times;

Specifically, for the detailed flow of step S102 and step S103, reference may be made to step 2 to step 6 in the content of the invention. The present invention will not be described in detail here.

Step S104, the typical index mean values of these sampling periods {R _g(m,c)u' , S _(m,c)u' , E _(m,c)u' } and the full-period typical values under the full coverage of M complete periods The index values {R _gu , S _u , E _u } are compared and analyzed. Among them, the subscript u represents the parameters of the individual (resident or user) u.

The mean of the typical indicators of the whole period of N residents and the typical indicators of the sampling period are in the form of a one-to-one correspondence to form a coordinate pair, which are (R _gu ,R _g(m,c)u' ), (S _u ,S _(m ) _,c)u' ) and (E _u ,E _(m,c)u' ); calculate the deviation between the typical index value of the whole period and the typical index value of the sampling period.

The present invention proposes a quantitative deviation measurement model, and the calculation formula of the model is:

F _m (X _u )=A _m X _u -B

Among them, the typical index values R _gu , _{Su or E u in} the whole period calculated from the number of all location records in the whole period are taken as independent variables _Xu , and the typical index values in the sampling period calculated from the location records in a part of the time period selected by the random process The mean values of R _g(m,c)u' , S _(m,c)u' , E _(m,c)u' are respectively used as the dependent variable y in the regression model; A _m is the regression coefficient, and B is a constant .

In addition, when the typical index value of the whole period is 0, the typical index value of the corresponding sampling period is theoretically 0, so B in the regression model is forcibly set to 0;

Further use the following formula to solve the quality loss coefficient (Quality Loss, QL):

QL _m =1-|A _m |

where |A _m | is the absolute value of the coefficient _Am .

The quality loss coefficient QL _m is distributed in the range of 0 to 1; according to this formula, the QL _{m_Rg} , QL _{m_S} corresponding to typical indicators such as the radius of gyration, moving distance and entropy of the spatial position data of group C under m time periods can be obtained. and QL _{m_E} ;

Finally, a corresponding quality loss coefficient value can be calculated for each typical index in different time periods m.

Further, calculate the maximum value max, minimum value min, quartile value and standard deviation std of the quality loss coefficient value of each typical index (gyroscope, moving distance and entropy) under different time periods m respectively; according to the quality loss coefficient The maximum value and the minimum value of the value, the distribution restriction area of the quality loss coefficient is drawn and stored; the mean value of the quality loss coefficient of each typical index (radius of gyration, moving distance and entropy) under the corresponding m time periods for N individuals in turn Perform curve or straight line fitting f _Rg , f _S and f _E , (may be quantitative mathematical relationships such as linear function, exponential function, power function, etc.), as shown in FIG. 3 is the mass loss coefficient fitting curve diagram in the present invention .

From this, the mathematical evolution rules of the loss coefficients of the typical indicators in different time periods are obtained respectively; the conclusions of the fitting curves f _Rg , f _S and f _E can be directly used to screen out the data sets under different estimation deviations, and according to the actual needs Reasonably specify the lower quality limit that the data should meet, and use it in the fields of urban spatial analysis, urban residents' movement dynamics analysis, and spatiotemporal pattern mining.

Step S105: Finally, the weighted quality loss coefficient of the entire data set can be calculated through f _Rg , f _S and f _E , and the calculation method is as follows:

Count the number of time-period distributions of the sampling records of each individual in the data set D' respectively, and bring them into f _Rg , f _S and f _E in turn to calculate the quality loss coefficient of the typical index of the individual;

Set the quality loss coefficient of each individual in dataset D to 0;

According to the number of users covering different time periods in the data sets D and D' as weights, calculate the weighted quality loss coefficient w _QL of each typical index under the data set;

Among them, users _m represents the number of users whose location records cover m time periods, users represents the number of all individuals; QL _m represents the quality loss coefficient of a typical indicator corresponding to the number of time periods;

By calculating the weighted quality loss coefficients w _{QL_Rg} , w _{QL_S} and w _{QL_E} of typical indicators such as radius of gyration, moving distance and entropy, the weighted quality loss coefficient W _QL of the entire data can be calculated;

The formula for calculating the weighted quality loss coefficient W _QL of the entire sparsely sampled data is:

specifically. The weighted quality loss coefficient W _QL is used to evaluate the quality of the entire data set; the smaller the weighted quality loss coefficient, the higher the overall quality of the data; the larger the weighted quality loss coefficient, the worse the overall quality of the data; the distribution area of the quality loss coefficient Any value in represents a sampling combination that effectively guides the extraction of data records of a specified quality.

4 is a system for processing spatial position data provided by the present invention, as shown in FIG.

The data sampling unit 410 is used for establishing an urban spatial database, and importing sparsely sampled spatial location data into the urban spatial database; dividing the sparsely sampled spatial location data into spatial location data covering the whole time period and space covering part of the time period location data;

The full-time data processing unit 420 is configured to divide the spatial position data covering the full period according to time periods, and divide it into time-series spatial position data of M time periods; M is a positive integer greater than or equal to 2; The spatial position data of C groups of m time periods are randomly selected from the time-series spatial position data of the time period, and the spatial position data of each time period in the time-series space position data of the M time periods are at least in the spatial position data of the C groups of m time periods. Selected more than k times; m initial value is 2, m is a positive integer less than M; k is a positive integer less than or equal to C, C is a positive integer; calculate each user corresponding to the spatial location data of each group of m time periods , and average the C group index values of each user in m time periods as the indicator value of each user in m time periods; The index value in the whole time period; the index value includes: the user's space activity range, the user's activity path length within the space activity range, and the difference and imbalance of the user's different spatial positions within the space activity range The index value deviation corresponding to m time periods is determined according to the index value of each user in m time periods and the index value of each user in the whole time period, and is determined based on the index value deviation corresponding to each user in m time periods The quality loss coefficient of the index value of each user in m time periods; if m=M, end the process, if m is less than M, add 1 to m as a new m value, and continue from the sequence of the M time periods In the spatial location data, the spatial location data of C groups of m time periods is randomly selected to determine the quality loss coefficient of each user index value under the numerical value m of different time periods according to the spatial location data covering the whole time period;

A partial period data processing unit 430, configured to determine the number of users covering the spatial position data of m time periods in the spatial position data covering the partial period; m is an integer from 2 to M respectively;

The data quality evaluation unit 440 is configured to determine the number of users of the spatial location data covering m time periods according to the spatial location data covering part of the time period, and each user determined by the spatial location data covering the whole time period is in m time periods The quality loss coefficient of the lower index value and the number of all users determine the weighted quality loss coefficient of the sparsely sampled spatial location data, 2≤m≤M.

In an optional embodiment, the full-time data processing unit 420 obtains, according to the deviation measurement model, the quality loss coefficient of the index value of each user under the number m of each time period, and the deviation measurement model is: F _m (X _u )=A _m X _u -B; wherein, F _m (X _u ) represents the index value of each user u under the number m of each time period, X _u represents the index value of each user in the whole time period, A _m is a regression coefficient, and B is a constant; the quality loss coefficient QL _m is determined by the following formula: QL _m =1-|A _m |; where |A _m | is the absolute value of the coefficient A _m ; The regression coefficient corresponding to each index value under the number m of each time period is brought into the above formula, and the quality loss coefficient corresponding to the index value of the gyration radius, the index value of moving distance and the index value of entropy can be obtained for each user under the number m of each time period. QL _{m_Rg} , QL _{m_S} and QL _{m_E} .

In an optional embodiment, the data quality evaluation unit 440 determines the quality loss coefficient w _QL corresponding to each index value of the sparsely sampled spatial location data by the following formula:

Among them, users _m represents the number of users who cover the spatial location data of m time periods in the spatial location data covering part of the time period, users represents the number of all users; QL _m respectively represents the quality loss of the index value of each user in m time periods coefficient, QL _m specifically includes: QL _{m_Rg} , QL _{m_S} or QL _{m_E} ; by calculating the mass loss coefficients w _{QL_Rg} , w _{QL_S} and w _{QL_E} of the radius of gyration, moving distance and entropy respectively, the weighting of the sparsely sampled spatial position data is calculated Mass loss factor W _QL :

Specifically, for the functions of each unit, reference may be made to the foregoing method embodiments, and details are not described herein.

Those skilled in the art can easily understand that the above are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present invention, etc., All should be included within the protection scope of the present invention.

Claims (10)

1. A method for processing spatial location data, comprising the steps of:

step 1, establishing an urban spatial database, and importing sparse sampled spatial position data to the urban spatial database; dividing the sparsely sampled spatial position data into spatial position data covering a full period and spatial position data covering a partial period;

step 2, dividing the spatial position data covering the whole time interval into time sequence spatial position data of M time intervals according to the time intervals; m is a positive integer greater than or equal to 2;

step 3, randomly selecting C groups of spatial position data of M time periods from the time sequence spatial position data of the M time periods, wherein the spatial position data of each time period in the time sequence spatial position data of the M time periods is selected at least k times from the spatial position data of the C groups of M time periods; m is 2 as the initial value, and M is a positive integer smaller than M; k is a positive integer less than or equal to C, C is a positive integer;

step 4, calculating an index value of each user corresponding to the spatial position data of each group of m time periods, and averaging C groups of index values of each user in the m time periods to serve as the index values of each user in the m time periods; calculating the index value of each user corresponding to the time sequence space position data of the M time periods in the full time period; the index value includes: the spatial activity range of the user, the activity path length of the user in the spatial activity range, and the difference and the imbalance of the user at different spatial positions in the spatial activity range;

step 5, determining index value deviations corresponding to m time periods according to the index values of each user in m time periods and the index values of each user in the full time period, and determining the quality loss coefficient of the index value of each user in m time periods based on the index value deviations corresponding to m time periods of each user;

step 6, if M is equal to M, executing step 7, and if M is smaller than M, adding 1 to M to obtain a new M value, and executing step 3;

and 7, determining the number of users covering M time period spatial position data in the spatial position data covering part of the time period, determining the number of users covering M time periods spatial position data according to the number of users covering M time periods spatial position data determined by the spatial position data covering part of the time period, the quality loss coefficient of the index value of each user in M time periods determined by the spatial position data covering the whole time period, and the weighted quality loss coefficient of all users determined by the spatial position data covering the whole time period, wherein M is more than or equal to 2 and less than or equal to M.

2. The method for processing spatial location data according to claim 1, wherein the step 1 specifically includes the steps of:

and importing the sparsely sampled spatial position data into the urban spatial database, and converting each spatial position data into a preset coordinate system, wherein each spatial position data comprises a sampling coordinate and sampling time.

3. The method for processing spatial location data according to claim 1, wherein the step 2 specifically includes the steps of:

and dividing the spatial position data covering the whole time interval into spatial position data of M time intervals according to actual requirements or in a self-adaptive mode.

4. The method for processing spatial location data according to claim 1, wherein the index value specifically includes:

the spatial motion range index is the radius of gyration R_g：

The index of the length of the movement path in the spatial movement range is a movement distance S:

the index of access variability and disparity at different spatial positions within the spatial range of motion is entropy E:

where n is the total number of spatial position samples per user in the spatial position data of each combination of M periods or the spatial position data of M periods, (x)_j,y_j) Is the coordinate value of the jth sampling point of each user, (x)_c,y_c) Is the center of gravity of all the sampling point positions of each user, n' is the number of sampling points different for each user, p_iIs the probability of occurrence of the ith distinct sample point of each user;

center of gravity (x) of all sampling point positions of each user_c,y_c) The calculation formula of (2) is as follows:

5. the method for processing spatial location data according to claim 4, wherein the step 5 specifically includes the steps of:

according to a deviation measurement model, obtaining a quality loss coefficient of an index value of each user under each time period number m, wherein the deviation measurement model is as follows:

F_m(X_u)＝A_mX_u-B

wherein, F_m(X_u) An index value, X, representing each user u at each period number m_uAn index value, A, representing each user at the full time period_mIs a regression coefficient, B is a constant;

the mass loss coefficient QL is determined by the following formula:

QL_m＝1-|A_m|

wherein, | A_m| is coefficient A_mAbsolute value of (d); substituting the regression coefficient corresponding to each index value of each user in each period number m into the formula to obtain the quality loss coefficient QL corresponding to the gyration radius index value, the movement distance index value and the entropy index value of each user in each period number m_{m_Rg}，QL_{m_S}And QL_{m_E}。

6. The method for processing spatial location data according to claim 5, wherein the step 7 specifically includes the steps of:

determining the quality corresponding to each index value of the sparsely sampled spatial position data by the following formulaCoefficient of mass loss omega_QL：

Wherein, users_mRepresenting the number of users covering m time periods of the spatial position data covering the part of the time periods, wherein users represents the number of all users; QL_mQuality loss coefficient, QL, representing the index value for each user over m time periods_mThe method specifically comprises the following steps: QL_{m_Rg}，QL_{m_S}Or QL_{m_E}；

By calculating the mass loss coefficient w of the radius of gyration, the distance traveled and the entropy, respectively_{QL_Rg}，w_{QL_S}And w_{QL_E}Calculating a weighted mass loss factor W for the sparsely sampled spatial position data_QL：

7. A system for processing spatial location data, comprising:

the data sampling unit is used for establishing an urban spatial database and importing sparse sampled spatial position data into the urban spatial database; dividing the sparsely sampled spatial position data into spatial position data covering a full period and spatial position data covering a partial period;

the full-time-period data processing unit is used for dividing the spatial position data covering the full time period into time sequence spatial position data of M time periods according to the time periods; m is a positive integer greater than or equal to 2; randomly selecting C groups of spatial position data of M time periods from the time sequence spatial position data of the M time periods, wherein the spatial position data of each time period in the time sequence spatial position data of the M time periods is selected at least k times from the spatial position data of the C groups of M time periods; m is 2 as the initial value, and M is a positive integer smaller than M; k is a positive integer less than or equal to C, C is a positive integer; calculating an index value of each user corresponding to the spatial position data of each group of m time periods, and averaging C groups of index values of each user in the m time periods to serve as the index values of each user in the m time periods; calculating the index value of each user corresponding to the time sequence space position data of the M time periods in the full time period; the index value includes: the spatial activity range of the user, the activity path length of the user in the spatial activity range, and the difference and the imbalance of the user at different spatial positions in the spatial activity range; determining index value deviations corresponding to m time periods according to the index values of each user in m time periods and the index values of each user in the full time period, and determining the quality loss coefficient of the index value of each user in m time periods based on the index value deviations corresponding to m time periods of each user; if M is equal to M, ending the processing, if M is smaller than M, adding 1 to M to serve as a new M value, and continuing to randomly select C groups of spatial position data of M time periods from the time sequence spatial position data of the M time periods so as to determine the quality loss coefficient of each user index value under different time period values M according to the spatial position data covering the whole time period;

a partial time period data processing unit for determining the number of users covering the m time period spatial position data in the spatial position data covering the partial time period; m is an integer from 2 to M;

and the data quality evaluation unit is used for determining the weighted quality loss coefficient of the sparsely sampled spatial position data according to the number of users covering the spatial position data of M time periods determined by the spatial position data covering the whole time periods, the quality loss coefficient of the index value of each user in M time periods determined by the spatial position data covering the whole time periods and the number of all users, wherein M is more than or equal to 2 and less than or equal to M.

8. The system for processing spatial locality data according to claim 7, wherein said full-time-period data processing unit finds a quality loss coefficient of an index value for each user for each time period number m based on a deviation metric model, said deviation metric model being: f_m(X_u)＝A_mX_u-B; wherein, F_m(X_u) An index value, X, representing each user u at each period number m_uAn index value, A, representing each user at the full time period_mIs a regression coefficient, B is a constant; the mass loss coefficient QL_mDetermined by the following formula: QL_m＝1-|A_mL, |; wherein, | A_m| is coefficient A_mAbsolute value of (d); substituting the regression coefficient corresponding to each index value of each user in each period number m into the formula to obtain the quality loss coefficient QL corresponding to the gyration radius index value, the movement distance index value and the entropy index value of each user in each period number m_{m_Rg}，QL_{m_S}And QL_{m_E}。

9. The system for processing spatial locality data according to claim 7, wherein the data quality assessment unit determines a quality loss coefficient ω corresponding to each index value of the sparsely sampled spatial locality data by using the following formula_QL：

Wherein, users_mRepresenting the number of users covering m time periods of the spatial position data covering the part of the time periods, wherein users represents the number of all users; QL_mQuality loss coefficient, QL, representing the index value for each user over m time periods_mThe method specifically comprises the following steps: QL_{m_Rg}，QL_{m_S}Or QL_{m_E}；

By calculating the mass loss coefficient w of the radius of gyration, the distance traveled and the entropy, respectively_{QL_Rg}，w_{QL_S}And w_{QL_E}Calculating a weighted mass loss factor W for the sparsely sampled spatial position data_QL：

10. A computer-readable storage medium, characterized in that a computer program is stored thereon, which computer program, when being executed by a processor, carries out a method of processing spatial position data according to any one of claims 1 to 6.

Images