A village population hollowing fluctuation mode identification method and system

Abstract

The present application relates to a kind of village population hollowing fluctuation mode identification method and system, belong to geographic information technology field.The method includes: mobile phone location trajectory data pre-processing;Based on SMoT model, identify user daily residence, combine Chinese New Year holiday window to respectively statistics village population and Chinese New Year population;With population outflow rate, construct hollowing index, after natural breakpoint method grading coding, generate time series;Accordingly, divide annual non-hollow village, seasonal hollow village and annual hollow village;Finally, adopt Euclidean distance and K-means algorithm to cluster the last two kinds of villages, identify fluctuation mode.The present application solves the problem that existing technology is difficult to capture short cycle population dynamic change, realizes the accurate quantification and mode classification to the seasonal fluctuation characteristics of village population hollowing, provides scientific basis for the differentiated resource allocation of rural revitalization.

Description

Technical Field

[0001] This invention belongs to the field of geographic information technology, specifically relating to a method and system for identifying fluctuation patterns in village population hollowing. Background Technology

[0002] In the rapid process of industrialization and urbanization, the one-way migration from rural to urban areas has become a common phenomenon, and a large number of people have left the rural areas, which has led to a serious problem of population hollowing out and has hindered the sustainable development of rural areas. Existing research shows that most of the rural migrants do not leave the village permanently, but return to their hometowns regularly during specific periods such as busy farming seasons and holidays[1]. This causes the degree of population hollowing out in villages to fluctuate throughout the year. Therefore, identifying the fluctuation pattern of population hollowing out in villages can effectively avoid the problem of "one-size-fits-all" policies, improve the resource allocation efficiency of population hollowing out management at different times, and provide a reference for promoting rural revitalization.

[0003] From a temporal perspective, current research on the hollowing out of village population mainly includes annual scale, single time node and monthly scale. For example, Wen et al. [2] analyzed the spatiotemporal evolution characteristics of the hollowing out of population in 87 administrative villages in Yanchi County, Ningxia in 2005, 2010, 2015 and 2020; Zhang Ru et al. [3] measured the level of hollowing out of population in Baiquan County, Northeast Black Soil Area in 2020 and evaluated the development potential of each village; Zhang Qingtian et al. [4] analyzed the monthly hollowing out of village in a mountainous county in western China in 2016 based on the electricity consumption data of farmers, and divided the villages into active and inactive villages according to the degree of fluctuation of monthly hollowing out.

[0004] The main shortcomings of existing technologies are as follows: (1) There is a lack of assessment methods for rural population hollowing at the level of individual daily activity characteristics. Most of them use panel data of regions (such as counties, towns, villages, etc.) for calculation; (2) They cannot capture the dynamic changes of rural population hollowing at short-term scales (such as weeks, months, quarters, etc.). Generally, they analyze rural population hollowing from the annual scale or a single time node; (3) There is a lack of identification of population hollowing fluctuation patterns. Only a few studies analyze the changes of rural population hollowing from the monthly or annual scale.

[0005] References:

[0006] [1] Wang D, Zhu Y, Zhao M, et al. Multi-dimensional hollowingcharacteristics of traditional villages and its influence mechanism based on the micro-scale: A case study of Dongcun Village in Suzhou, China[J]. LandUse Policy, 2021, 101: 105146.

[0007] [2] Wen Q, Li J, Ding J, et al. Evolutionary process and mechanism of population hollowing out in rural villages in the farming-pastoral ecotone ofNorthern China: A case study of Yanchi County, Ningxia[J]. Land Use Policy, 2023, 125: 106506.

[0008] [3] Zhang Ru, Du Guoming, Li Yuheng, et al. Study on the measurement and remediation path of rural hollowing in Northeast China based on human-land relationship—taking Baiquan County, Heilongjiang Province as an example [J]. Progress in Geography, 2024, 43(3): 488-503.

[0009] [4] Zhang Qingtian, Luo Xiaolong, Peng Haitao, et al. Analysis of seasonal changes and influencing mechanisms of rural hollowing-out in mountainous areas—taking County A in western China as an example [J]. Urban and Rural Planning, 2022(2): 30-38. Summary of the Invention

[0010] The purpose of this invention is to provide a method and system for identifying the fluctuation pattern of village population hollowing in order to solve the following three problems: (1) Based on the user's trajectory data, starting from the characteristics of individual daily activities, identify the resident population of the village and the population during the Spring Festival to assess the degree of village population hollowing; (2) Based on the high update frequency of trajectory data, construct a short-cycle time series of population hollowing to reflect the dynamic changes of rural population hollowing on a short-cycle scale; (3) Based on the village population hollowing time series, identify the fluctuation pattern of village population hollowing.

[0011] To achieve the above objectives, the technical solution of the present invention is: a method for identifying fluctuation patterns in village population hollowing-out, comprising:

[0012] Obtain mobile phone location trajectory data of users within the study area, and preprocess the mobile phone location trajectory data, including removing duplicate values, deleting null values, and deleting typical error values;

[0013] Based on the preprocessed trajectory data, a stop point identification model is used to identify the set of stop points for each user each day. The stop point with the longest stay during the nighttime period is selected as the corresponding user's daily residence. According to the distribution of users' daily residences within a study period, the village to which the corresponding user belongs is identified as the resident population of the corresponding village within the corresponding study period. Using the same rules, the Spring Festival holiday of the corresponding year within the study period is selected as the identification window to identify the population of each village during the Spring Festival.

[0014] Based on the number of residents in each village during each research period and the number of residents during the Spring Festival, the population hollowing-out index of each village in each research period is calculated; the total population hollowing-out index is screened and classified and coded to construct the population hollowing-out time series of each village.

[0015] Based on the aforementioned time series of population hollowing out, each village is classified into hollowing out types, resulting in year-round non-hollowing-out villages, seasonal hollowing-out villages, and year-round hollowing-out villages.

[0016] The Euclidean distance was used to measure the similarity of time series of different villages within the same type for seasonal and year-round hollow villages, and the K-means algorithm was used to cluster the time series of villages within the same type to identify the fluctuation patterns of village population hollowing under each category.

[0017] Furthermore, the dwell point identification model is the SMoT model, which sets a spatial threshold D and a time threshold M. Location points of the same user that meet the spatial threshold D and have a continuous dwell time that meets the time threshold M within the same day are aggregated into a dwell point, thus obtaining the corresponding user's dwell point set for that day. Each dwell point includes dwell time and latitude and longitude information.

[0018] Furthermore, identifying the resident population of a village during the research period specifically includes: counting the village where the corresponding user's daily residence appears most frequently during the research period as the corresponding user's longest daily residence; if the corresponding longest daily residence belongs to the target village, and the number of days the corresponding user belongs to the target village is not less than half of the total number of days in the corresponding research period, then the corresponding user is determined to be a resident population of the target village during the corresponding research period.

[0019] Furthermore, the screening and hierarchical coding of the total population hollowing-out index specifically includes:

[0020] Population hollowing-out indices less than or equal to 0 are encoded as 0 and removed from the set to be classified;

[0021] The population hollowing-out index with a remainder greater than 0 is divided into k categories using the natural discontinuity method, and coded as 1, 2, …, k in order of increasing hollowing-out degree, where k is an integer not less than 2;

[0022] The population hollowing-out index of each village in each research period is mapped to a corresponding code to obtain the population hollowing-out time series of the corresponding village.

[0023] Furthermore, k takes the value 4, where code 0 represents non-hollowing, code 1 represents mild hollowing, code 2 represents moderate hollowing, and code 3 represents deep hollowing.

[0024] Furthermore, the hollowing type classification rule is as follows:

[0025] If all values ​​in the time series of a village’s population hollowing-out are 0, it is classified as a village that is not hollowed out throughout the year.

[0026] If the number of zero values ​​in the time series is less than the total number of study periods and at least one value is greater than zero, it is classified as a seasonal hollow village.

[0027] If all values ​​in the time series are greater than 0, the village is classified as a hollow village throughout the year.

[0028] Furthermore, the formula for calculating the similarity of time series data between two villages using Euclidean distance is as follows:

[0029]

[0030] in, and Let L represent the time series of population depopulation for any two villages within a certain type, where L is the total number of study periods. and These are the hollowing-out level codes for the two villages in the i-th research period;

[0031] During the clustering process, the silhouette coefficient and / or the sum of squared errors are used to determine the optimal number of clusters for the K-means algorithm.

[0032] Furthermore, the research period is weekly, monthly, or quarterly, and the research period covers at least one consecutive year.

[0033] Furthermore, the nighttime period is 0:00–6:00, the spatial threshold D of the SMoT model is 500 meters, and the time threshold M is 30 minutes.

[0034] This invention also provides a system for recognizing patterns of village population hollowing out, comprising:

[0035] The data preprocessing module is used to acquire mobile phone location trajectory data of users within the study area and to preprocess the data by removing duplicate values, deleting null values, and deleting typical error values.

[0036] The population identification module is used to identify users' daily residence based on the SMoT model, and then identify the resident population of each village during each research period and the population of each village during the Spring Festival.

[0037] The index construction module is used to calculate the population hollowing-out index of each village for each research period based on the resident population and the population during the Spring Festival, and to perform hierarchical coding to construct the population hollowing-out time series of each village.

[0038] The classification module is used to classify villages into year-round non-hollow villages, seasonal hollow villages, and year-round hollow villages based on time series.

[0039] The pattern recognition module is used to perform time series clustering on seasonally hollowed-out villages and year-round hollowed-out villages using Euclidean distance combined with the K-means algorithm, and output the results of village population hollowing-out fluctuation pattern recognition.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] (1) It can capture the seasonal changes of rural population hollowing out within the year, and further supplement the study of the spatiotemporal characteristics of rural population hollowing out;

[0042] (2) Identify the fluctuation patterns of village population hollowing out to provide a scientific basis for differentiated governance of rural population hollowing out. Attached Figure Description

[0043] Figure 1 This is a flowchart of a method for identifying fluctuation patterns in village population hollowing according to the present invention.

[0044] Figure 2 It serves as the cluster center for the fluctuation patterns of hollowed-out villages throughout the year.

[0045] Figure 3 This represents the C1 fluctuation pattern of seasonally hollowed-out villages.

[0046] Figure 4 This represents the C2 fluctuation pattern of seasonally hollowed-out villages. Detailed Implementation

[0047] The technical solution of the present invention will now be described in detail with reference to the accompanying drawings.

[0048] This invention provides a method for identifying patterns of population hollowing out in villages, the main contents of which include:

[0049] (1) Preprocessing of mobile phone location data;

[0050] (2) Identification of village residents and population during the Spring Festival;

[0051] (3) Construction of time series of village population hollowing-out;

[0052] (4) Classification of village population hollowing-out types;

[0053] (5) Identification of fluctuation patterns in village population hollowing.

[0054] The specific process of the method of this invention is as follows: Figure 1 It includes the following steps:

[0055] The first step is to preprocess the mobile phone location data. (1) Remove duplicate values; (2) Delete null values; (3) Delete typical error values, which refers to records with abnormal ID, time, latitude and longitude.

[0056] The second step involves identifying the village's resident population and the population during the Spring Festival. First, based on the preprocessed user trajectory data and the SMoT model, spatial thresholds for key parameters are set. meters and time threshold Minutes were allocated to identify the daily set of locations for each user within the study area. Then, choose nighttime. The time period during which the user spends the longest is considered their daily residence. Next, determine the user's status during the research period. (Because the study focuses on the seasonal changes in hollowing out, the research period can be chosen as week, month, or season.) Longest daily residence If it is greater than or equal to half of the research period, then it is identified as... The corresponding village during the research period The resident population within the village. Similarly, for identifying the village's population during the Spring Festival, the entire research period is selected. (Because the study focuses on changes within the year, so) It must cover the Spring Festival holiday in the corresponding year (at least one consecutive year). As a research period for population identification during the Spring Festival, the population of villages during the Spring Festival was identified based on the aforementioned residential population identification rules.

[0057] The third step is to construct a time series analysis of village population depopulation. First, based on the village research period obtained in the second step... Resident population and research period The population during the Spring Festival is represented by the population outflow rate (Formula 4) during the research period. The degree of population hollowing out, constructing T for each research cycle iThe village population hollowing-out index. Afterwards, for ( / This indicates how many sub-periods the study period T is divided into, which is also the length of the population hollowing-out time series multiplied by... (Indicates the number of villages in the study area) The values ​​are processed as follows: (1) Encode the values ​​less than or equal to 0 as 0, and then remove them. (2) For the remaining values ​​in the set, divide them according to the natural discontinuity. Classes, and encoded in ascending order as follows: (Among them, the larger the value of the code, the more serious the hollowing out of the village's population); (3) According to the above code, each research period in the village is divided into groups. The population hollowing-out index is mapped to the corresponding code to complete the construction of the time series of village population hollowing-out.

[0058] The fourth step is to classify the types of village depopulation. Based on the time series of depopulation constructed in the third step, the types of village depopulation are classified as follows: (1) Non-depopulated villages throughout the year, that is, all values ​​in the time series of depopulation are 0; (2) Seasonally depopulated villages, that is, the number of values ​​with 0 in the time series of depopulation is less than 100%. (3) Hollow villages throughout the year, that is, all values ​​in the time series of population hollowing are greater than 0.

[0059] Step 5: Identification of fluctuation patterns in village population hollowing. The fluctuation patterns of seasonal hollowing villages and year-round hollowing villages obtained in Step 4 are identified separately. The specific methods are as follows: (1) Euclidean distance is used to measure the similarity of time series in a certain type; (2) K-means algorithm is used to cluster the time series in the type to identify different fluctuation patterns.

[0060] The specific steps for identifying the village's resident population and the population during the Spring Festival, constructing a time series of village population hollowing-out, and identifying the fluctuation patterns of village population hollowing-out are as follows:

[0061] (1) Identification of village residents and population during the Spring Festival

[0062] First, the SMoT model is used, and its key parameter spatial thresholds are set. meters and time threshold Minutes are used to obtain a collection of the user's daily dwell times. .

[0063] (1)

[0064] in, For users exist day( Let T be the set of places of stay on a specific day within the study period T. This refers to the stop points in the set, which include the duration of the stay and the latitude and longitude.

[0065] Next, select nighttime. The time period during which the user spends the longest is considered their daily residence. .

[0066] (2)

[0067] in, This represents the function that takes the maximum value. Indicates the point of stay During the nighttime period Duration of stay within.

[0068] Subsequently, based on the user's research period Daily residence status within the area, identifying the research cycle. The population of the inner village.

[0069] (3)

[0070] in, Indicates village During the research period The result of determining the number of residents in the area (1 for yes, 0 for no); Indicates the user's longest-staying residence; This indicates that the user's longest-staying residence is in a village. The number of days.

[0071] Population identification in villages during the Spring Festival and villages in the research period The method for identifying the resident population is the same, only the research period is longer. Change to selecting the entire research period Chinese New Year holiday in the corresponding year .

[0072] (2) Construction of time series of village population hollowing

[0073] Each research cycle obtained in the second step Village population and study period Population statistics were conducted by region during the Spring Festival to obtain data for each village in each research period. Resident population and population during the Spring Festival Then, the population outflow rate was used to characterize the degree of village depopulation, and the village was calculated in each study period. Population hollowing-out index .

[0074] (4)

[0075] in, For the first The village in the Population hollowing-out index for each research period, For the first The village in the The number of residents in each research period For the first The villages during the research period Population numbers during the Spring Festival.

[0076] Subsequently, all villages were constructed in each research cycle. The complete set of population hollowing-out indexes.

[0077] ((5)

[0078] in, The total number of villages in the study area. This represents the total number of research periods.

[0079] For sets Elements less than 0 are filtered and encoded to obtain a new set. .

[0080] ((6) ((7)

[0081] For sets Divided into using the natural discontinuity method Classes, coded in ascending order as follows: The population hollowing-out level code was obtained.

[0082] ((8)

[0083] in, For the first The village in the Research cycles The population hollowing-out level is indicated by a higher value, signifying a more severe degree of hollowing-out. It is a function with natural discontinuities.

[0084] Finally, the time series of population depopulation in each village was completed.

[0085] ((9)

[0086] in, For the first The time series of population depopulation in several villages, with a series length of [number missing]. .

[0087] (3) Identification of fluctuation patterns in village population hollowing

[0088] Following the rules in step four, villages are categorized into year-round non-hollow villages, seasonally hollow villages, and year-round hollow villages. Then, fluctuation pattern identification is performed on the seasonally hollow villages and year-round hollow villages to explore the fluctuation patterns of village population hollowing.

[0089] First, Euclidean distance is used to measure the similarity between time series in the type.

[0090] ((10)

[0091] in, and These are time series data on population depopulation for any two villages within a certain category.

[0092] Subsequently, the K-means algorithm was used to cluster the time series data within the category, identifying different patterns of population hollowing out. The determination of clusters can be made using parameters such as the silhouette coefficient (SC) and the sum of squares of errors (SSE). Figure 2 It serves as the cluster center for the fluctuation patterns of hollowed-out villages throughout the year.

[0093] In a specific experiment, this invention takes a certain county in a certain city as the research area. Based on the large-scale anonymized mobile phone location data generated by the SDK mobile phone multi-application APP positioning from July 2023 to June 2024, it calculates the monthly population hollowing index of each village in the county and identifies the population hollowing fluctuation pattern of the villages.

[0094] First, based on the urban-rural classification codes published by the National Bureau of Statistics, 492 village (community) units in a certain county were divided into urban and rural areas, and 359 rural village units were selected as the analysis units for this experiment. Then, the mobile phone location data was preprocessed to obtain the trajectory data of users within the study area; using the SMoT model, with a 500-meter spatial threshold and a 30-minute time threshold, the daily dwelling points of each user were identified. Based on the dwelling point set, the period from 0:00 to 6:00 was defined as the nighttime period to identify the daily residence of users, and the monthly resident population of each village was statistically obtained according to the rules of this invention. Similarly, the Spring Festival holiday of 2024 (February 10th to February 17th) was selected, and the population of each village during the Spring Festival was statistically obtained according to the same resident population identification rules.

[0095] Secondly, based on the population hollowing-out index formula defined according to the present invention, the monthly population hollowing-out index of each village is calculated; then, the degree of hollowing-out is encoded, and this example sets four encoding types: non-hollowing-out (coded as 0), mild hollowing-out (coded as 1), moderate hollowing-out (coded as 2), and severe hollowing-out (coded as 3). Based on the encoding results, the time series of population hollowing-out for each village is constructed, and according to the village type classification rules of the present invention, villages with hollowing-out throughout the year and villages with hollowing-out seasonally are identified.

[0096] Finally, Euclidean distance is used to measure the similarity of time series of villages of the same type, and the K-means algorithm is used to cluster and identify the population hollowing-out fluctuation pattern of villages. In this example, the elbow rule is used to determine the optimal number of clusters.

[0097] The results showed that: (1) Of the 359 villages in a certain county, 350 were hollowed out throughout the year, 9 were hollowed out seasonally, and there were no villages that were not hollowed out throughout the year. (2) Five different fluctuation patterns were identified in the hollowed-out villages throughout the year, of which C1 had 38 villages, C2 had 137 villages, C3 had 85 villages, C4 had 34 villages, and C5 had 56 villages. Two different fluctuation patterns were identified in the seasonally hollowed-out villages, of which C1 had 7 villages and C2 had 2 villages. Figure 3 , Figure 4 (3) In villages with year-round hollowing out, the C1 pattern shows mild hollowing out during busy farming seasons and holidays, and moderate hollowing out during other periods; the C2 pattern shows moderate hollowing out only during the Spring Festival, and deep hollowing out during other periods; the C3 pattern shows a similar fluctuation pattern to the C2 pattern, with village changes only during the Spring Festival, but the degree of hollowing out is milder than that of the C2 pattern; the C4 pattern shows mild hollowing out throughout the year, with no fluctuations; the C5 pattern shows deep hollowing out in spring, mild hollowing out during the Spring Festival, and moderate hollowing out during other periods. (4) In villages with seasonal hollowing out, the C1 pattern shows changes similar to those in towns, with mild hollowing out during the Spring Festival and non-hollowing out for most of the other time; the C2 pattern shows non-hollowing out only in specific months, and hollowing out in other months.

[0098] The above are preferred embodiments of the present invention. Any changes made to the technical solution of the present invention that do not exceed the scope of the technical solution of the present invention shall fall within the protection scope of the present invention.

Claims

1. A method for identifying fluctuation patterns in village population hollowing-out, characterized in that, include: Obtain mobile phone location trajectory data of users within the study area, and preprocess the mobile phone location trajectory data, including removing duplicate values, deleting null values, and deleting typical error values; Based on the preprocessed trajectory data, a stop point identification model is used to identify the set of stop points for each user each day. The stop point with the longest stay during the nighttime period is selected as the corresponding user's daily residence. According to the distribution of users' daily residences within a study period, the village to which the corresponding user belongs is identified as the resident population of the corresponding village within the corresponding study period. Using the same rules, the Spring Festival holiday of the corresponding year within the study period is selected as the identification window to identify the population of each village during the Spring Festival. Based on the number of residents in each village during each research period and the number of residents during the Spring Festival, the population hollowing-out index of each village in each research period is calculated; the total population hollowing-out index is screened and classified and coded to construct the population hollowing-out time series of each village. Based on the aforementioned time series of population hollowing out, each village is classified into hollowing out types, resulting in year-round non-hollowing-out villages, seasonal hollowing-out villages, and year-round hollowing-out villages. The Euclidean distance was used to measure the similarity of time series of different villages within the same type for seasonal and year-round hollow villages, and the K-means algorithm was used to cluster the time series of villages within the same type to identify the fluctuation patterns of village population hollowing under each category.

2. The method for identifying village population hollowing-out fluctuation patterns according to claim 1, characterized in that, The dwell point identification model is the SMoT model, which sets a spatial threshold D and a time threshold M. Location points of the same user that meet the spatial threshold D and have a continuous dwell time that meets the time threshold M within the same day are aggregated into a dwell point, thus obtaining the corresponding user's dwell point set for the day. Each dwell point includes dwell time and latitude and longitude information.

3. The method for identifying village population hollowing-out fluctuation patterns according to claim 2, characterized in that, Identifying the resident population of a village during the research period specifically includes: counting the village where the corresponding user's daily residence appears most frequently during the research period as the corresponding user's longest daily residence; if the corresponding longest daily residence belongs to the target village, and the number of days the corresponding user belongs to the target village is not less than half of the total number of days in the corresponding research period, then the corresponding user is determined to be a resident population of the target village during the corresponding research period.

4. The method for identifying village population hollowing-out fluctuation patterns according to claim 1, characterized in that, The screening and hierarchical coding of the total population hollowing-out index specifically includes: Population hollowing-out indices less than or equal to 0 are encoded as 0 and removed from the set to be classified; The population hollowing-out index with a remainder greater than 0 is divided into k categories using the natural discontinuity method, and coded as 1, 2, …, k in order of increasing hollowing-out degree, where k is an integer not less than 2; The population hollowing-out index of each village in each research period is mapped to a corresponding code to obtain the population hollowing-out time series of the corresponding village.

5. The method for identifying village population hollowing-out fluctuation patterns according to claim 4, characterized in that, k takes a value of 4, where code 0 represents non-hollowing, code 1 represents mild hollowing, code 2 represents moderate hollowing, and code 3 represents deep hollowing.

6. The method for identifying village population hollowing-out fluctuation patterns according to claim 1, characterized in that, The hollowing type classification rule is as follows: If all values ​​in the time series of a village’s population hollowing-out are 0, it is classified as a village that is not hollowed out throughout the year. If the number of zero values ​​in the time series is less than the total number of study periods and at least one value is greater than zero, it is classified as a seasonal hollow village. If all values ​​in the time series are greater than 0, the village is classified as a hollow village throughout the year.

7. The method for identifying village population hollowing-out fluctuation patterns according to claim 1, characterized in that, The formula for calculating the similarity of time series data between two villages using Euclidean distance is as follows: in, and Let L represent the time series of population depopulation for any two villages within a certain type, where L is the total number of study periods. and These are the hollowing-out level codes for the two villages in the i-th research period; During the clustering process, the silhouette coefficient and / or the sum of squared errors are used to determine the optimal number of clusters for the K-means algorithm.

8. The method for identifying village population hollowing-out fluctuation patterns according to claim 1, characterized in that, The research period is weekly, monthly, or quarterly, and the research period covers at least one consecutive year.

9. The method for identifying village population hollowing-out fluctuation patterns according to claim 2, characterized in that, The nighttime period is 0:00–6:00, the spatial threshold D of the SMoT model is 500 meters, and the time threshold M is 30 minutes.

10. A system for recognizing patterns of village population hollowing-out fluctuations, characterized in that, include: The data preprocessing module is used to acquire mobile phone location trajectory data of users within the study area and to preprocess the data by removing duplicate values, deleting null values, and deleting typical error values. The population identification module is used to identify users' daily residence based on the SMoT model, and then identify the resident population of each village during each research period and the population of each village during the Spring Festival. The index construction module is used to calculate the population hollowing-out index of each village for each research period based on the resident population and the population during the Spring Festival, and to perform hierarchical coding to construct the population hollowing-out time series of each village. The classification module is used to classify villages into year-round non-hollow villages, seasonal hollow villages, and year-round hollow villages based on time series. The pattern recognition module is used to perform time series clustering on seasonally hollowed-out villages and year-round hollowed-out villages using Euclidean distance combined with the K-means algorithm, and output the results of village population hollowing-out fluctuation pattern recognition.

Images