A method for estimating the length of stay in commercial and leisure spaces based on sparse location data

Abstract

The present invention relates to a method for estimating the dwell time in commercial and leisure spaces based on sparse positioning data. Using sparse positioning data, the total dwell time of visitors in commercial and leisure spaces is obtained, and the visitors' trajectories and spatial sequences along the way are estimated. Weighted least squares are then used to estimate the average dwell time in each space. This method utilizes relatively easy-to-obtain sparse positioning data, and the dwell time results can be used as an evaluation indicator of the attractiveness of commercial and leisure spaces, supporting spatial planning, design, and management. This method is applicable to similar commercial and leisure spaces, such as commercial blocks, shopping malls, parks, amusement parks, and scenic spots. It can serve as a core module and algorithm in space utilization monitoring and evaluation, and has good feasibility.

Description

Technical Field

[0001] The method of the present invention measures the length of time tourists stay in commercial and leisure spaces based on sparse positioning data, and is applicable to the fields of commercial street development and management, retail industry, tourism industry, and urban planning. Background Art

[0002] Commercial and leisure spaces, including commercial blocks, shopping malls, parks, amusement parks, and scenic spots, are crucial venues for urban life and can also serve as a vital gateway to a city and a source of economic growth. Visitor behavior within these spaces is a key indicator for evaluating operational performance and developing planning and management measures. The duration of visitor stay is a crucial indicator for measuring spatial vitality and evaluating space utilization efficiency. Currently, the mainstream method for evaluating the performance of human activity in commercial and leisure spaces is to measure and count visitor traffic. However, measurement of visitor duration remains relatively lacking, resulting in an incomplete evaluation of space utilization.

[0003] The lack of measurement of length of stay in commercial and leisure spaces stems primarily from the limitations of existing measurement methods. The most traditional method is visitor questionnaires, which often rely on visitors' recollections and statements about their activities within the space. This data collection process is time-consuming and inefficient, and data accuracy is significantly affected by personal factors. A more objective and accurate method uses sensor devices, such as RFID, Wi-Fi, and Bluetooth, to locate visitors. However, these methods are less practical due to three main limitations: First, sensors require specialized deployment, resulting in high investment and maintenance costs; second, they require visitors to carry signal transmitters or actively activate Wi-Fi or Bluetooth, making implementation difficult and controllable; and third, they require frequent recording of location data to calculate individual length of stay, placing high demands on data processing and storage, resulting in high implementation costs. Furthermore, while GPS can be used for visitor location tracking without the need for sensor devices and is widely available on personal devices, it is not commonly used due to the limitations of device battery life and the requirement for active interaction for data feedback. Furthermore, indoor location tracking is not possible, making practical feasibility limited.

[0004] In recent years, the emergence of new technologies and the expansion of data sources have provided a foundation for using sparse location data to measure dwell time in commercial and leisure spaces. Sparse location data refers to data with relatively coarse spatial positioning accuracy and low positioning frequency, such as mobile phone signaling and image recognition positioning. Depending on the application requirements, this type of location data can meet the required position accuracy at a specific spatial scale. More importantly, its low recording frequency places less pressure on data processing and storage. Appropriate algorithms can largely restore visitors' spatiotemporal trajectories, making the measurement of spatial dwell time feasible and more efficient. Therefore, this technology not only has a wide range of applications but also offers significant advantages over existing methods in terms of widespread implementation. Summary of the Invention

[0005] The purpose of the present invention is to calculate the length of stay in commercial and leisure spaces based on sparse positioning data and the application of a specific algorithm. The sparse positioning data is used to obtain the total length of stay of tourists in commercial and leisure spaces, and the tourists' trajectories and spatial sequences are estimated. The weighted least squares method is used to estimate the average length of stay in each space. The method uses relatively easy-to-obtain sparse positioning data, and the length of stay results can be used as an evaluation indicator of the attractiveness of commercial and leisure spaces to support space planning, design, management and other work. The method is applicable to similar commercial and leisure spaces such as commercial blocks, comprehensive shopping malls, parks, amusement parks, scenic spots, etc., and can be used as a core module and algorithm in space utilization monitoring and evaluation work, with good feasibility.

[0006] The technical solution of the present invention is:

[0007] A method for estimating the length of stay in a commercial leisure space based on sparse positioning data includes the following steps:

[0008] Step 1. Condition setting

[0009] [Assumption 1] A commercial leisure space consists of J subspaces, each of which is represented by j = {1, 2, ..., J}; based on the actual length of stay of each tourist in each subspace, there is an average length of stay in each subspace. That is, the object to be measured by this algorithm;

[0010] [Set 2] Sparse location trajectory samples of several tourists in the commercial leisure space have been obtained; each trajectory records several locations of the tourist and the time when the record occurred; several adjacent location pairs can be extracted from each trajectory; the total number of all location pairs is N, and each location pair is represented by n = {1, 2, ..., N}; for each location pair, its starting position is represented as s according to the time sequence of the tourist's behavior. n , the end position is represented by e n , the corresponding recording time is T sn and T en , then the total duration of the tourist's activities between the location points is T en -T sn ;

[0011] [Set 3] For any point pair, based on the specific spatial form of the commercial and leisure space, the number of times tourists pass through each subspace B has been inferred n ={b 1n ,b 2n ,…,b jn ,…,b Jn}, and the total duration of tourists' stay activities t n , which is equivalent to the total duration of the activity minus the duration of movement t wn, that is, t n =T en -T sn -t wn ;

[0012] Step 2. Average stay duration estimation algorithm

[0013]

Step 2.1

[0014]

[0015] where ε j is the dwell time error in space j, assumed to be independently and identically normally distributed for all n Its mean is 0 and its standard deviation is ω is a positive constant, that is, the standard deviation is proportional to the average length of stay, and all ε j The covariance between is 0; that is:

[0016]

[0017] Since the sum of independent normal distributions is still a normal distribution, formula (2) can be written as:

[0018]

[0019] where ε n The mean is 0 and the standard deviation is ωδ n Normal distribution of:

[0020]

[0021]

Step 2.2

[0022]

[0023] in ε is a normal distribution with mean 0 and standard deviation ω;

[0024] [Step 2.3] Solve equation (5) using the linear regression model to obtain the parameters

[0025]

Step 2.4

[0026] The method of the present invention is applicable to similar commercial and leisure spaces such as commercial blocks, comprehensive shopping malls, parks, amusement parks, scenic spots, etc. It can be used as a core module and algorithm in space utilization monitoring and evaluation work, and has good feasibility.

[0027] The advantages of the present invention are:

[0028] (1) Using easily accessible sparse positioning data to measure the length of stay in commercial and leisure spaces, the accuracy can meet the continuous monitoring needs of general management and decision-making;

[0029] (2) After determining the required number of trajectories, the application model can directly obtain the dwell time, which is convenient and feasible. It is applicable to a wide range of scenarios, such as squares, parks, and other places where sparse positioning data can be obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 Algorithm flow chart

[0031] Figure 2 Table 1 Example of probability setting for each visit duration in each block

[0032] Figure 3 Table 2 Examples of consumer trajectories

[0033] Figure 4 Table 3 Data conversion form DETAILED DESCRIPTION

[0034] The technical solution of the present invention is further described below with reference to the accompanying drawings.

[0035] The technical solution adopted by the present invention is ( Figure 1 ):

[0036] Step 1. Condition setting

[0037] [Assumption 1] A commercial leisure space consists of J subspaces (e.g., a block, a plot of land, a store in a commercial street), each subspace is represented by j = {1, 2, ..., J}. Theoretically, based on the actual length of time each tourist stays in each subspace, there is an average length of time for each subspace: That is, the object to be measured by this algorithm.

[0038] [Set 2] Sparse location trajectory samples of several tourists in the commercial leisure space have been obtained. Each trajectory records several locations of the tourist and the time when the record occurred. From each trajectory, several adjacent location pairs can be extracted; the total number of all location pairs is N, and each location pair is represented by n = {1, 2, ..., N}. For each location pair, according to the time sequence of the tourist's behavior, its starting position is represented as s n , the end position is represented by e n, the corresponding recording time is T sn and T en , then the total duration of the tourist's activities between the location points is T en -T sn .

[0039] [Set 3] For any point pair, based on the specific spatial form of the commercial and leisure space, the number of times tourists pass through each subspace B has been inferred n ={b 1n ,b 2n ,…,b jn ,…,b Jn}, and the total duration of tourists' stay activities t n , which is equivalent to the total duration of the activity minus the duration of movement t wn , that is, t n =T en -T sn -t wn .

[0040] Step 2. Average stay duration estimation algorithm

[0041]

Step 2.1

[0042]

[0043] where ε j is the dwell time error in space j, assumed to be independently and identically normally distributed for all n Its mean is 0 and its standard deviation is ω is a positive constant, that is, the standard deviation is proportional to the average length of stay, and all ε j The covariance between them is 0. That is:

[0044]

[0045] Since the sum of independent normal distributions is still a normal distribution, formula (2) can be written as:

[0046]

[0047] where ε n The mean is 0 and the standard deviation is ωδ n Normal distribution of:

[0048]

[0049]

Step 2.2

[0050]

[0051] in ε is a normal distribution with mean 0 and standard deviation ω.

[0052] [Step 2.3] Solve equation (5) using the linear regression model to obtain the parameters

[0053]

Step 2.4

[0054] Example

[0055] Taking the measurement of the length of time tourists stay in a plot in a commercial pedestrian street as an example, the implementation method of the above technical solution is explained using a simulation verification method.

[0056] 1. Stay behavior simulation

[0057] Assume that there are 7 plots in the commercial street (j = [1, 2, ..., 7]) and they are arranged linearly in order. There are also 7 lengths of stay of tourists in each plot (k = [1, 2, ..., 7]), corresponding to the length of stay of [0, 10min, ..., 60min]. Assume that the probability of each length of stay in each plot is p jk (Table 1), that is, the distribution of length of stay, based on which the expected length of stay for each plot can be obtained.

[0058] Generate a certain number of location point pairs. For a point pair n, randomly select the starting block b according to uniform distribution s , similarly generate the terminal plot b e , according to the setting of linear arrangement between plots, the plot path sequence s is obtained n If the starting plot is b2 and the ending plot is b4, the plot path sequence is {b2, b3, b4}. According to the distribution of the stay time of these three plots, a random number is drawn to determine the stay time in each plot (such as: 5min, 15min, 25min), and the total stay time in the plot of this record is t n (Table 2).

[0059] 2. Algorithm implementation:

[0060] [Step 1] Generate 4000 position point pairs (N). For any position pair n, the number of times it passes through each subspace is B n ={b 1n ,b 2n ,…,b jn,…,b 7n}, if it passes through plot j, then b jn =1, if not passed, then b jn =0, convert the data into the form of Table 3.

[0061] [Step 2] Construct the dwell time T according to the above average dwell time estimation algorithm n Model with the number of stays in each subspace:

[0062]

[0063] where ε n The mean is 0 and the standard deviation is ωδ n Normal distribution of:

[0064]

[0065]

Step 3

[0066]

[0067]

Step 4

[0068]

Step 5

Claims

1. A method for estimating the length of stay in commercial and leisure spaces based on sparse positioning data, characterized in that: Applicable to commercial blocks and shopping malls, it serves as the core module and algorithm for space utilization monitoring and evaluation. The algorithm includes the following steps: Step 1. Condition setting 【Set 1】A commercial leisure space consists of subspaces, each of which consists of According to the actual length of time each tourist stays in each subspace, there is an average length of time for each subspace , that is, the object to be measured by this algorithm; [Set 2] Sparse positioning trajectory samples of several tourists in the commercial leisure space have been obtained; each trajectory records several locations of the tourist and the time when the record occurred; several adjacent location pairs can be extracted from each trajectory; the total number of all location pairs is , each position point is used For each location point pair, according to the time sequence of tourist behavior, its starting position is represented as , the end position is expressed as , the corresponding recording time is and , then the total duration of the tourist's activities between the location points is ; [Set 3] For any point pair, the number of times tourists pass through each subspace has been inferred based on the specific spatial form of the commercial and leisure space. , and the total duration of tourists' stay activities , which is equal to the total duration of the activity minus the duration of movement ,Right now ; Step 2. Average stay time estimation algorithm 【Step 2.1】Build the dwell time Model with the number of stays in each subspace: in It's space The dwell time error is assumed to be Independent and identically normally distributed , whose mean is 0 and standard deviation is , is a positive constant, that is, the standard deviation is proportional to the average length of stay, and all The covariance between is 0; that is: Since the sum of independent normal distributions is still a normal distribution, formula (2) can be written as: in The mean is 0 and the standard deviation is The normal distribution of , that is: 【Step 2.2】Set the initial stay duration For any value, Substituting into formula (4), we get ; Divide formula (3) by ,have to: in , , The mean is 0 and the standard deviation is Normal distribution; [Step 2.3] Use the weighted least squares method of the linear regression model to solve equation (5) and obtain the parameters ; 【Step 2.4】Use Repeat steps 2.2 and 2.3 until the parameters converge, which is the final estimate of the spatial residence time.

Images