Farmland non-farming data processing and analysis system and method

Abstract

The invention provides a cultivated land non-cultivation data processing and analysis system and method. The system comprises a non-cultivation rate influence factor analysis module and a non-cultivation rate prediction module. The non-farming rate influence factor analysis module integrates a geographic detector model, a panel model, a space model, a space-time geographic weighted regression model and other modules to obtain a significant influence factor result of non-farming in a target area; and the non-farming rate prediction module comprises a panel model prediction module and a time t prediction module to predict the non-farming rate corresponding to each administrative region in the next year. Based on the technical scheme of the invention, multi-source heterogeneous data such as land utilization, climate environment and the like are integrated; index classification is more detailed and comprehensive, so that an analysis result is more effective; time and space factors are fused, and variables of different regions are analyzed more accurately; the factor identification accuracy is remarkably improved, the space prediction precision is remarkably improved, and the time sequence prediction error is remarkably reduced.

Description

Technical Field

[0001] The present invention relates to the field of big data analysis and mining technology, and in particular to a system and method for processing and analyzing cultivated land and uncultivated land data. Background Art

[0002] While the current national, provincial, and municipal analysis of factors affecting uncultivated land has broad coverage, it still suffers from significant shortcomings. Existing technologies are often based on single-factor analysis, focusing on the impact of social and economic factors on uncultivated land, primarily using linear regression models. Individualized farmer factors are primarily analyzed using binary choice models, while natural factors are primarily analyzed through qualitative analysis or descriptive statistics. Overall, despite the diverse methodologies, these approaches are narrowly focused, preventing a holistic understanding of the factors influencing uncultivated land and neglecting the inherent connections between these factors. Consequently, current analysis of uncultivated land suffers from significant technical shortcomings in its application, relying heavily on single-factor analysis models and lacking a robust framework for multidimensional spatiotemporal data fusion and cross-model collaborative analysis. Existing technologies often employ traditional regression models or single geographic detectors and lack a systematic understanding of the coupling mechanisms between socioeconomic systems, farmer behavior, and natural endowments. Existing technologies fail to fully reveal the spatiotemporal interactions between urban and rural land use and topographic constraints, resulting in insufficient explanatory power for the complex processes of uncultivated land.

[0003] Furthermore, existing technologies generally fail to simultaneously consider proximity (e.g., the contagion effect of land use between villages), geographic distance (e.g., the impact of transportation accessibility on farmland transfer), and economic connections (e.g., the driving effect of regional industrial agglomeration on farmland use). This hinders the analysis of the structural contradictions between land used for specialty industries and land used for food production, and makes it difficult to reveal the evolutionary patterns and key drivers of uncultivated farmland through spatiotemporal data mining. Methodologically, the widespread application of traditional econometric models (e.g., ordinary least squares) struggles to capture the spatial heterogeneity of data. Although existing technologies have attempted to incorporate spatial autocorrelation analysis, the application of advanced spatial econometric methods such as geographically weighted regression and spatiotemporal autoregressive models remains exploratory. Furthermore, the lack of multi-source heterogeneous data fusion technology hinders the coordinated analysis of socioeconomic statistics, remote sensing monitoring data, and farmer survey data, limiting the refined identification of the drivers of uncultivated farmland. For example, while remote sensing image overlay analysis can identify suspicious areas, it lacks dynamic prediction capabilities.

[0004] On the other hand, there are currently few existing technologies in China for predicting the future trend of uncultivated farmland. Existing solutions primarily predict the total future cultivated land area in a given region, but not the uncultivated rate. Existing prediction technologies primarily use multiple regression models to predict cultivated land area. These models are primarily developed using multiple regression and neural networks, and are then compared and evaluated. Markov models are also used to predict the structural proportions and areas of cultivated land, forest land, grassland, water bodies, construction land, and unused land.

[0005] In statistics, p-value and q-value are important indicators used to measure the significance of hypothesis test results. The p-value indicates the probability of the observed result or a more extreme result when the null hypothesis is true. The q-value is the result after correcting the p-value. Existing technologies focus more on the prediction of changes in the total amount of cultivated land, but lack in-depth analysis of the dynamic evolution of the core indicator of the uncultivated rate. Since the changes in the total amount of cultivated land cannot reflect the deep adjustments in the land use structure, such as the transformation of fragmented cultivated land to large-scale construction land, the protection measures proposed based on the total amount analysis are often difficult to accurately respond to the regionally differentiated risks of cultivated land loss. For example, the uncultivated patterns of cultivated land in different terrains are significantly different, and a zoning and classification management strategy is needed, but existing methods are difficult to achieve such refined analysis. Summary of the Invention

[0006] In response to the problems in the above-mentioned prior art, this application proposes a data processing and analysis system for uncultivated cultivated land, which includes an influencing factor analysis model and a prediction model. The influencing factor analysis model includes multivariate analysis modules such as a geographic detector model module, a panel model module, a spatial model module, and a spatiotemporal geographic weighted regression model module. By implementing modeling and solving and statistical analysis on each module respectively, a common feature screening operation is carried out based on the significance indicators output by each module. Specifically, the p-value and q-value of the indicator, the quantitative parameter of the degree of influence represented by the q-value, and the positive and negative direction of the influence are used for multidimensional comprehensive analysis, thereby achieving scientific identification and accurate extraction of key influencing factors. The prediction model is built on the organic combination of panel data modeling methods and quadratic regression analysis technology. By integrating the advantages of panel models in dynamic analysis of time series and cross-sectional data, and superimposing the fitting ability of quadratic regression for nonlinear relationships, a prediction and analysis framework with both time series dynamic characteristics and spatial heterogeneity is formed. This invention constructs a comprehensive and detailed indicator system. Through the collaboration of multiple models, it conducts a comprehensive and in-depth analysis of the factors affecting uncultivated cultivated land from multiple dimensions such as geographic space and time series, and realizes comprehensive and refined quantitative analysis of the factors affecting uncultivated cultivated land in designated areas and future trend prediction.

[0007] The present invention provides a system for processing and analyzing uncultivated land data, comprising an uncultivated land rate influencing factor analysis module and an uncultivated land rate prediction module; The module for analyzing factors affecting the uncultivated rate integrates the geographic detector model module, the panel model module, the spatial model module, the spatiotemporal geographic weighted regression model module and the comprehensive analysis module; The uncultivated rate prediction module includes a panel model prediction module and a time t prediction module; The geographic detector model module collects uncultivated indicator data and pre-processes missing values; classifies different data types; calculates the q value and p value of each variable based on the geographic detector, determines the influencing factors through two key factor screening mechanisms, and obtains the influencing factors year by year; The panel model module performs a preliminary screening of the uncultivated index and uses an F test and / or a t test and / or a Hausman test to establish different types of panel models and obtain different significant influencing factors; the preliminary screening method includes screening based on the correlation between independent variables, screening based on the correlation between the independent variables and the uncultivated rate, and removing variables that are weakly correlated with the uncultivated rate based on the screening based on the correlation between the independent variables and the uncultivated rate; and removing insignificant variables based on the screening based on the correlation between the independent variables and the uncultivated rate. The spatial model module first uses five spatial weight matrices, namely, a spatial adjacency weight matrix, a spatial inverse distance weight matrix, a square of a spatial inverse distance weight matrix, a distance-based spatial adjacency matrix, and an economic distance matrix, to sequentially perform a spatial correlation test on the uncultivated rate; if the test passes, a spatial model is established; if not, no spatial model is required; The spatiotemporal geographic weighted regression model module constructs an influencing factor system including multiple primary indicators and multiple secondary indicators, uses panel data of uncultivated rates in target areas and target years, screens highly correlated indicators through correlation coefficients, and uses variance inflation factors to screen indicator variables. A spatiotemporal geographic weighted regression model is established and solved to obtain the coefficients of each indicator in the target area. The comprehensive analysis module organizes the results of the spatiotemporal geographic weighted regression model of the target area, screens significant features, and combines the results of each model solution to determine the degree of influence and the positive or negative impact of the analysis results of each significant indicator, ultimately obtaining the results of significant influencing factors of non-cultivation in the target area; The panel model prediction module constructs different types of regression models based on the change characteristics of the respective variables to make a preliminary prediction of the uncultivated rate for the next year. After completing the independent variable prediction, based on the solution results of the panel model and the obtained independent variable prediction values, a time series prediction model is constructed based on the panel model parameter estimation results to further predict the uncultivated rate corresponding to each administrative district in the next year. The time t prediction module divides regional types through Euclidean clustering and hierarchical clustering; for all cities in each category, a dynamic prediction model for uncultivated rate based on the time dimension is established in combination with the regression prediction mechanism and regression prediction is performed. The regression models used include linear regression model, quadratic regression model, exponential regression model, logarithmic regression model, exponential decay model and trend extrapolation; a prediction model that matches each city category is selected to obtain the corresponding accurate uncultivated rate prediction value.

[0008] In one embodiment, in the geographic detector model module, the formulas for differentiation and factor detection are as follows: (1-1) in, represents the i-th factor The uncultivated rate of cultivated land in year t The explanatory power of is in the range [0,1]. The larger the value is, the stronger the explanatory power of the i-th factor is on the uncultivated land rate in year t; Represents the classification of the i-th factor in year t, which is divided into If the factor is a continuous variable, it needs to be discretized; N represents the total number of samples in the entire region, and the total number of samples in each year is the same. represents the number of samples corresponding to the hth category of the i-th factor in the t-th year; represents the variance of the uncultivated land rate in the entire region in year t, represents the variance of the uncultivated land rate corresponding to the hth category of the i-th factor in the t-th year; Will After a simple transformation, the following non-central F distribution is satisfied:

[0009] ; in, is the noncentral parameter, is the mean value of the uncultivated land rate corresponding to the hth category of the i-th factor in the t-th year; Verifiable Is it significant? At the significance level of 0.1, if , the result is considered significant, that is, the factor is the influencing factor of uncultivated arable land.

[0010] In one embodiment, the panel model module establishes different types of panel models through F test and / or t test and / or Hausman test, specifically including: mixed model, fixed effect model and random effect model.

[0011] In one embodiment, the spatiotemporal geographic weighted regression model module constitutes a spatiotemporal three-dimensional coordinate , the model expression is as follows:

[0012] Where n is the number of sample points, r is the number of independent variables, 、 、 are the longitude, latitude and time of the i-th sample point respectively, is the dependent variable of the i-th sample point, is the kth independent variable of the i-th sample point, is the intercept term, is the regression coefficient of the kth independent variable of the ith sample point, is the random error of the i-th sample point that is independent and normally distributed.

[0013] The present invention provides a method for processing and analyzing arable land uncultivated data, comprising the following steps: Step 1: Analyze the geographic detector module to obtain relevant indicators of the uncultivated rate, collect indicator data and preprocess the data, and use the equal interval method, quantile method, and systematic clustering method to classify the indicator type of one indicator of uncultivated land in the target area corresponding to one group of data; use the systematic clustering method to classify the indicator type of one indicator of uncultivated land in the target area corresponding to multiple groups of data; after classifying the different methods of independent variables, calculate the q value and p value of each independent variable in turn according to the geographic detector method; and determine the factors affecting the uncultivated land each year and their explanatory power based on the p value and / or q value; Step 2: Establish an influencing factor index system and pre-process the index data through panel model module analysis; after processing some index data by proportion, perform preliminary index screening through four methods, and then establish different types of panel models through F test and / or t test and / or Hausman test; after processing some index data by area, perform preliminary index screening through four methods, and then establish different types of panel models through F test and / or t test and / or Hausman test; the four methods are screening based on the correlation between independent variables; screening based on the correlation between independent variables and uncultivated rate; removing variables with weak correlation with uncultivated rate based on screening based on the correlation between independent variables and uncultivated rate; removing insignificant variables based on screening based on the correlation between independent variables and uncultivated rate; Step 3: Obtain panel data on the uncultivated land rate and influencing factors for the target region and target year intervals through spatial panel module analysis. Preprocess the sample data to construct a spatial adjacency weight matrix, a spatial inverse distance weight matrix, the square of the spatial inverse distance weight matrix, a distance-based spatial adjacency matrix, and an economic distance matrix. Perform spatial autocorrelation tests on the uncultivated land rate for the target region and target year intervals, and analyze the test results. Step 4: Obtain panel data on the target year-specific uncultivated rate and influencing factors in the target region through spatiotemporal geographic weighted regression analysis; construct an influencing factor indicator system consisting of multiple primary and secondary indicators; screen indicator variables based on the variance inflation factor; construct a spatiotemporal geographic weighted regression model; and obtain the coefficients of each indicator in the target region based on the established spatiotemporal geographic weighted regression model. Step 5: Analyze and organize the results of the spatiotemporal geographic weighted regression model through the comprehensive analysis module, and extract p-value and q-value data from the relevant calculation results. At the same time, integrate the output results of the panel model, spatial model, and spatiotemporal geographic weighted regression model, and record the impact direction and degree of each indicator in detail. Systematically organize the integrated data and screen out the significant indicator results from the summary table based on the rounding-up strategy. The screened significant influencing indicators are sorted in descending order according to their q-values. Finally, the results of the significant influencing factors of uncultivated land in the target area are obtained. Step 6, uncultivated land prediction analysis; based on the prediction of uncultivated land rate of the panel model, after completing the prediction of independent variables, with the help of the solution results of the panel model and combined with the obtained independent variable prediction values, predict the uncultivated land rate corresponding to each administrative district in the next year; based on the prediction of uncultivated land rate at time t, the regional types are divided by Euclidean clustering and hierarchical clustering; for all cities in each category, a dynamic prediction model of uncultivated land rate based on the time dimension is established in combination with the regression prediction mechanism and regression prediction is performed; the regression models used include linear regression model, quadratic regression model, exponential regression model, logarithmic regression model, exponential decay model and trend extrapolation; a prediction model that matches each city category is selected to obtain an accurate uncultivated land rate prediction value; corresponding evaluation criteria are constructed to measure the accuracy of the prediction results by calculating the absolute error and relative error of each uncultivated land rate prediction value.

[0014] In one embodiment, the comprehensive analysis module analysis specifically includes: Step 5.1: Organize the results of the spatiotemporal geographic weighted regression model. Based on the pre-defined three-level target area division rules, systematically merge the significant indicators from each year using the spatiotemporal geographic weighted regression model output data. Ultimately, organize and form an independent and complete set of solution result data for each target area. First, accurately count the number of three-level target areas and filter the significant indicators based on the screening threshold. The filtered indicators and their related data are orderly arranged and recorded according to the pre-defined data format and arrangement rules. Finally, construct a spatiotemporal geographic weighted regression model solution result table. Step 5.2: Summarize the model solution results, including: Step 5.2.1: Extract the p-value and q-value data from the relevant calculation results. At the same time, comprehensively integrate the output results of the panel model, spatial model, and spatiotemporal geographic weighted regression model; record the impact direction and impact degree of each indicator in detail; and construct a summary table of the results of the influencing factor model solution based on the integrated data. Step 5.2.2: This table summarizes the results of each model and related indicator information; based on the rounding-up strategy, filter out the significant indicator results from the summary table; Step 5.2.3: Arrange the selected significant influencing indicators in descending order according to their q-values. At the same time, classify them into positive influencing factor sets and negative influencing factor sets based on the impact direction of each indicator. In step 5.2.4, continue to use the rounding-up strategy to calculate the new screening threshold and screen the indicators with significant impact; In step 5.2.5, only the q-value results are used as the screening basis to screen out the indicators that have a key impact on the target variable; Step 5.2.6: Systematically organize the results obtained from the screening and analysis of the above steps to finally obtain the results of the significant influencing factors of non-cultivation in the target area.

[0015] In one embodiment, the step 6 of uncultivated land prediction analysis specifically includes: Step 6.1: Predicting the uncultivated rate based on the panel model. First, predict the independent variables. Based on the changing characteristics of each variable, construct different types of regression models to achieve a preliminary prediction of the uncultivated rate for the next year. Based on the independent variable predictions, the panel model results, combined with the obtained independent variable predictions, further predict the uncultivated rate for each administrative district in the next year. Step 6.2, prediction of the uncultivated rate based on time t, includes: Step 6.2.1: transform the three-year uncultivated data of each city into ,in, represents the uncultivated rate of the i-th city in the t-th year, represents the uncultivated rate in the ith city in year t+1, represents the uncultivated rate in the ith city in year t+2; Step 6.2.2: City cluster analysis operation, using Euclidean clustering for classification; hierarchical clustering is also used to select clusters to be merged by minimizing the objective function; Step 6.2.3: Regression prediction. Perform regression prediction for all cities in each category. The regression models used include linear regression, quadratic regression, exponential regression, logarithmic regression, exponential decay, and trend extrapolation. Through these operations, six predicted values ​​of uncultivated rates are obtained for each city. Step 6.2.4, model selection, select a suitable prediction model for each city category to obtain an accurate prediction value of the uncultivated rate; Step 6.3, establish evaluation criteria, construct corresponding evaluation criteria, and measure the accuracy of the prediction results by calculating the absolute error and relative error of each uncultivated rate prediction value.

[0016] The system and method for processing and analyzing arable land uncultivated data provided by the present invention have at least the following beneficial effects compared with the prior art: Establish a unified spatiotemporal data warehouse, integrate multi-source heterogeneous data indicators such as land use, climate environment, production conditions, and natural factors, and through the construction of a multi-dimensional analysis framework and the organic integration of a five-dimensional model, form a comprehensive analysis method that integrates influencing factor diagnosis, spatial differentiation analysis, and time series trend prediction. The analysis results will be more effective, comprehensive, and comprehensive. Compared with traditional analysis methods, the factor identification accuracy is significantly improved, the spatial prediction accuracy is significantly improved, and the time series prediction error is significantly reduced. The geographic detector model provides a more detailed and comprehensive classification of indicators, including primary and secondary indicators, resulting in more detailed and effective analysis results. The panel model automatically generates differentiated variable combinations based on the specific circumstances of different regions, significantly reducing the model's prediction error rate and significantly improving the pertinence of regional analysis. For the first time, the spatial model uses five spatial weight matrices, simultaneously considering three dimensions, including adjacency and geography, to avoid single-value test errors and significantly improve effectiveness and credibility. The accuracy of spatial autocorrelation detection has been significantly improved, effectively avoiding test bias. By combining geographic spatial coordinates with time, temporal and spatial factors are integrated to provide a more accurate analysis of variables in different regions. The prediction model first predicts the independent variable and then predicts the independent variable through the dependent variable. By changing the time scale and the independent variable, the prediction is enriched and the lack of data is avoided. A bidirectional prediction model for time series and spatial pattern is established. By dynamically adjusting the time window length and spatial grid accuracy, the problem of prediction bias caused by missing data is resolved, and data utilization is significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The present invention will be described in more detail below based on embodiments and with reference to the accompanying drawings, wherein: Figure 1 Shows a complete technical solution flow diagram of the present invention; Figure 2 A schematic diagram of the panel model selection process of the present invention is shown; Figure 3 Shows the geographic detector module flow chart of the present invention; Figure 4 The panel model module flow chart of the present invention is shown; Figure 5 The spatial panel module and spatiotemporal geographic weighted regression module processes of the present invention are shown. DETAILED DESCRIPTION

[0018] The present invention will be further described below with reference to the accompanying drawings.

[0019] The present invention provides a universal uncultivated rate influencing factor analysis and prediction system and method.

[0020] In one embodiment, Figure 1 As shown, the analysis model of factors affecting the uncultivated rate includes the following modules: 1. Geographic detector model module: First, collect uncultivated indicator data and preprocess missing values. Classification is implemented for different data types: single-indicator single-group data uses the equal interval method, quantile method, and systematic clustering method to form 9 classification combinations; single-indicator multi-group data only uses the systematic clustering method for classification. Based on the geographic detector principle, the q value and p value of each variable are calculated, and the influencing factor is determined through two key factor screening mechanisms: one is to sort by q value when p < 0.1; the other is to sort directly by q value to obtain the influencing factors and their explanatory power year by year; 2. Panel Model Module: Based on pre-processed data from geographic detectors, two data conversion modes (proportional data / area data) were designed to address strong correlations and multicollinearity issues among indicators. Four variable screening strategies were also implemented: 1. Correlation coefficient method between independent variables; 2. Correlation coefficient method between independent variables and uncultivated rate; 3. Elimination of weakly correlated variables based on Strategy 2; 4. Elimination of insignificant regression variables based on Strategy 2. The time-fixed-effect model was determined using F-tests, t-tests, and Hausman tests to identify different significant influencing factors. 3. Spatial Model Module: Innovatively constructs five types of spatial weight matrices (spatial adjacency, inverse distance, inverse distance squared, distance-based spatial adjacency, and economic distance matrix), simultaneously considering the three dimensions of adjacency, geographic distance, and economic connections, enabling multi-dimensional testing of the spatial effects of uncultivated land. 4. Spatial-temporal geographic weighted regression model module: This module constructs an influencing factor system consisting of five primary indicators and 42 secondary indicators. It uses correlation coefficients to screen highly correlated indicators and uses variance inflation factors to eliminate multicollinearity. It then analyzes the spatial distribution characteristics and influencing mechanisms of the uncultivated rate from a spatiotemporal perspective. 5. Comprehensive Analysis: The results of the spatiotemporal geographic weighted regression model for the third-level target areas under the first-level target areas are collated. Significant features are screened according to the screening criteria. Then, the results of different models are combined to determine the degree of impact and the positive or negative impact of each significant indicator, and finally a comprehensive conclusion is drawn.

[0021] Secondly, the uncultivated rate prediction module includes the following modules: Panel model prediction module: Integrates the independent variable prediction screening mechanism and the prediction mechanism, and builds a time series prediction model based on the panel model parameter estimation results. Time t prediction module: regional types are divided through cluster analysis mechanism, and a dynamic prediction model for uncultivated rate based on time dimension is established in combination with regression prediction mechanism. In one embodiment, each module is described in detail as follows: Geographic detector is a statistical method used to detect spatial heterogeneity and reveal the driving force behind it. Its basic idea is that if the independent variable has a significant impact on the dependent variable, then the spatial distribution of the independent variable and the dependent variable should be consistent; Different administrative regions have different uncultivated rates. The economic levels, demographic factors, production conditions, etc. corresponding to different administrative regions also have certain differences. The independent variables and dependent variables have their own spatial distribution characteristics. Therefore, geographic detectors are used to find factors that affect the uncultivated rate of cultivated land.

[0022] In one embodiment, Figure 2 As shown, it includes the panel model module, and the panel model mainly includes the mixed model, fixed effect model and random effect model.

[0023] In one embodiment, the method includes a spatial model module. For the spatial model, five spatial weight matrices, namely, a spatial adjacency weight matrix, a spatial inverse distance weight matrix, a square of a spatial inverse distance weight matrix, a distance-based spatial adjacency matrix, and an economic distance matrix, are used to sequentially perform spatial correlation tests on the uncultivated rate. The spatial model refers to the introduction of spatial factors into traditional econometrics. Before establishing the spatial model, it is necessary to establish a spatial weight matrix, followed by a spatial autocorrelation test. If the test passes, the spatial model is established. If not, no spatial model is required. The global Moran index is usually used to reflect the spatial aggregation of the entire study area. For the spatial model with fixed effects, the within-group deviation transformation method is first used to remove the individual or time effect, and then the maximum likelihood estimation is performed; for the spatial model with random effects, the generalized deviation transformation is first performed, and then the maximum likelihood estimation is performed.

[0024] In one embodiment, a comprehensive analysis module is included. Since the spatiotemporal geographic weighted model is analyzed year by year, and the factors affecting the uncultivated cultivated land in the three-level target area are obtained, it is impossible to compare the results with other models. Therefore, a data interface standardization operation is required. If a variable significantly affects the uncultivated rate of a certain administrative district in any year, then this variable is considered to be an influencing factor of the uncultivated cultivated land in the administrative district. After sorting, the solution results of each three-level target area of ​​the spatiotemporal geographic weighted model can be obtained. Secondly, the significant indicators that affect more than half or more are screened out, and the solution screening results of the spatiotemporal geographic weighted model are obtained and incorporated into the comprehensive analysis framework. Finally, based on the p-value and q-value analysis of the geographic detector model, the indicators in the influencing factor model that are considered to be significant in more than half of the modules are screened one by one, sorted according to the q-value, and classified according to their impact direction, and finally the comprehensive analysis results of the uncultivated influencing factors are obtained.

[0025] In one embodiment, including an uncultivated land prediction module and an uncultivated land prediction module of a panel model, in order to effectively solve the problem of predicting the uncultivated land rate of cultivated land, this application adopts a panel model to perform relevant prediction operations. In this process, the prediction of independent variables is first carried out. According to the change characteristics presented by each variable, different types of regression models are constructed in a targeted manner to achieve a preliminary prediction of the uncultivated land rate in the next year. On the basis of completing the independent variable prediction, with the help of the solution results of the panel model and combined with the obtained independent variable prediction values, the uncultivated land rate corresponding to each administrative district in the next year can be further predicted.

[0026] In one embodiment, a module for predicting the uncultivated crop rate based on time t is included. When performing this prediction, Euclidean clustering is first used to accurately measure the differences between cities. Regression predictions are then performed for all cities in each category using six regression models: linear, quadratic, exponential, logarithmic, exponential decay, and trend extrapolation. This yields six predicted values ​​for the uncultivated crop rate for each city. Based on this, a suitable prediction model is carefully selected for each city category to obtain a correspondingly accurate predicted value for the uncultivated crop rate.

[0027] In one embodiment, an uncultivated factor analysis module is included, such as Figure 3 As shown, the first step is the analysis of the geographic detector module, step 1.1 construction of the influencing factor indicator system and data preprocessing: Step 1.1.1 Based on the current research status of factors affecting uncultivated arable land at home and abroad, this application constructs an index system of factors affecting uncultivated arable land from five aspects: economic level, demographic factors, production conditions, policy regulation, and natural factors. Among them, population mobility refers to the difference between the number of registered residents at the end of the year and the number of permanent residents at the end of the year; the proportion of employees in the primary industry refers to the ratio of employees in the primary industry to the total number of employed people; total grain output per unit area refers to the ratio of total grain output to total grain planting area; distance from the city center refers to the driving distance from the city center of each prefecture-level city to the city center of the provincial capital; Step 1.1.2 Collect indicator data. By collecting and organizing data, we obtained data for a total of 50 indicators, and all of these 50 indicators are continuous variables, so we first need to discretize the independent variables. Based on the data type of the indicators, the 50 indicators can be divided into two categories. One category is that one indicator corresponds to only one set of data, such as per capita GDP, rural employment, and cultivated land water resource coverage. This category has a total of 40 indicators. The other category is that one indicator corresponds to multiple sets of data, such as cultivated land grade (excellent / high / medium / low), soil texture (loam / clay / sandy soil), and soil biodiversity (rich / average / not rich). This category has a total of 10 indicators. Step 1.1.3 Data preprocessing: For missing data, use the interpolation method to fill in all missing values.

[0028] Step 1.2: One indicator corresponds to a set of data discretization: Step 1.2.1 Classify continuous indicator data using the equal interval method (divided into 3 / 4 / 5 categories); The data of a group of data corresponding to an indicator of uncultivated land in the target area are classified using the equal interval method with classification into 3, 4 and 5 categories. The classification results of the uncultivated land indicators in the target area using the equal interval method are presented. The classification results of the equal interval method are greatly affected by extreme values, which may reduce the accuracy and reliability of the results of the geographic detector model. Step 1.2.2 Classify indicator data using the quantile method (divided into 3 / 4 / 5 categories); The quantile method, which is classified into 3, 4, and 5 categories, is used to classify the indicator type corresponding to a group of data related to an indicator of uncultivated land in the target area. The classification results of the uncultivated land indicators in the target area using the quantile method are displayed. The classification results of the quantile method are not affected by extreme values, and the number of samples in each category is relatively uniform, and the variance of a certain category will not be zero. However, this classification method cannot well reflect the distribution characteristics of the data. Step 1.2.3 Classify indicator data using the hierarchical clustering method (divided into 3 / 4 / 5 categories); The systematic clustering method classified into 3, 4 and 5 categories is used to classify the data of the indicator type corresponding to a set of data for an indicator related to uncultivated land in the target area, and the classification results of the uncultivated land indicators in various regions in 2022 using the quantile method are displayed. The systematic clustering method makes the intra-class differences after classification small and the inter-class differences large, but like the equal interval method, some categories only correspond to one sample, which may reduce the accuracy and reliability of the geographic detector model results.

[0029] Step 1.3: Discretize data where one indicator corresponds to multiple sets of data; The systematic clustering method with classification into 3, 4 and 5 categories is used to classify the data of indicator types corresponding to multiple groups of data for one indicator of uncultivated land in the target area, and the classification results of the uncultivated land indicators in the target area using the quantile method are displayed.

[0030] Step 1.4: Geographic detector model construction: After classifying the independent variables using different methods, the q and p values ​​of each independent variable are calculated in turn according to the geographic detector principle. After classifying the continuous indicator data using the equal interval method, the classified data is substituted into the geographic detector model, and the q and p values ​​of each independent variable are calculated in turn, and the solution results of some variables in the target area are obtained. After classifying the continuous indicator data using the quantile method, the classified data is substituted into the geographic detector model, and the q and p values ​​of each independent variable are calculated in turn, and the solution results of some variables in the target area are obtained. After classifying the continuous indicator data using the system clustering method, the classified data is substituted into the geographic detector model, and the q and p values ​​of each independent variable are calculated in turn, and the solution results of some variables in the target area are obtained.

[0031] Step 1.5 Identify the key factors that affect the uncultivated rate; Step 1.5.1 Determine the geodetector results based on the p-value and q-value: According to the principle of geographic detectors, after solving the data of the target year in sequence, the results obtained by different classification methods are compared and sorted. If there is more than one classification method that makes the p-value corresponding to a variable in a certain year less than 0.1, the classification method with the largest q-value is selected. The factors affecting the uncultivated land in each year and their explanatory power are determined by sorting according to the q-value corresponding to the p-value less than 0.1. Step 1.5.2 Determine the result based only on the q value: If the q value is not tested for significance, it still has physical meaning. Therefore, we only focus on the q value. When q > 0.5, we select the classification method with the largest q value. We then sort by q value to determine the factors that influence the uncultivated land in each year and their explanatory power.

[0032] In one embodiment, Figure 4 As shown, it includes panel model module analysis, step 2.1 construction of influencing factor indicator system and data preprocessing.

[0033] Step 2.2: Process some indicator data according to proportion and area; Step 2.2.1 Process some indicator data proportionally. Seven indicators—the proportion of cultivated land within the cultivated land protection target, cultivated land water resource coverage, cultivated land urban and rural road coverage, cultivated land rural road coverage, cultivated land rural residential area coverage, cultivated land urban agglomeration area coverage, and soil pH—are expressed using their corresponding proportions of cultivated land area. Subjectively filter out some similar indicators, and then construct an indicator system for influencing factors. Step 2.2.2: Process some indicator data by area. Seven indicators—the proportion of cultivated land within the cultivated land protection target, cultivated land water resource coverage, cultivated land urban and rural road coverage, cultivated land rural road coverage, cultivated land rural residential area coverage, cultivated land urban agglomeration area coverage, and soil pH—are expressed using their corresponding cultivated land area. Subjectively screen out some similar indicators, and then construct an indicator system for influencing factors.

[0034] Step 2.3: Build the model based on the indicators of the proportions corresponding to the factors.

[0035] After partial indicator data are processed proportionally, the indicators are initially screened in the indicator system, and different types of panel models are established through various tests.

[0036] Method 1 for initial screening of indicators: Screening based on the correlation between independent variables; Step 2.3.1: Screen based on the correlation between the independent variables. If the correlation coefficient is greater than 0.7, it means that the correlation between the two indicators is strong. Only one of the indicators that has a greater impact on the uncultivated land rate is retained. The correlation coefficient graph of the remaining indicators after screening is shown. Step 2.3.2: Perform an F test and a t test. The F test determines whether to use a mixed or fixed-effects model. The F test results show that only the p-value for the time effect is less than 0.05, indicating that a time effect exists, but no individual effect exists. A t test is used to test the significance of each time effect. The corresponding p-values ​​are generally less than 0.05, indicating that a time effect exists. A Hausman test is used to determine whether the effect is a fixed or random effect. Step 2.3.3 Determine the type of panel model to be established. The F test and Hausman test indicate that a time-fixed-effect model should be used.

[0037] Step 2.3.4: Solve the model and obtain the results.

[0038] Method 2 for initial screening of indicators: screening based on the correlation between the independent variable and the uncultivated rate; Step 2.3.1: For screening based on the correlation between the independent variable and the uncultivated rate, when the correlation between two factors is greater than 0.7, their correlation coefficients with the uncultivated rate can be calculated, and the factors with the higher correlation coefficients with the uncultivated rate can be retained. The correlation coefficient graph of the remaining indicators after screening; Step 2.3.2: Perform an F test and a t test. The F test determines whether to use a mixed or fixed-effect model. The F test results show that only the p-value for the time effect is less than 0.05, indicating that a time effect exists, but no individual effect exists. A t test is used to test the significance of each time effect. The corresponding p-values ​​are generally less than 0.05, indicating that a time effect exists. Step 2.3.3 Use the Hausman test to determine whether it is a fixed effect or a random effect; Step 2.3.4 Determine the type of panel model to be established. The F test and Hausman test indicate that a time-fixed-effect model should be used. Step 2.3.5 Solve the model and obtain the results.

[0039] Since the above two methods obtain fewer significant variables affecting uncultivated land, we consider comprehensively screening variables using strong correlation coefficients between independent variables and weak correlation coefficients with the independent variable uncultivated land rate.

[0040] Method 3 for initial screening of indicators: Based on method 2, remove insignificant variables one by one.

[0041] Step 2.3.1: Among the indicators with a correlation coefficient greater than 0.7 with the uncultivated rate, retain one that has a greater impact on the uncultivated rate of cultivated land, and then remove the variables with insignificant regression coefficients in the regression results. In steps 2.3.2-2.3.5, perform F-tests and Hausman tests to determine the panel model to use and solve the model. After each solution, remove the variable with the largest p-value until all remaining variables are significant, and then return the regression results.

[0042] Step 2.4: Build a model based on the area indicators corresponding to the factors.

[0043] First, we use part of the obtained indicator data to construct the corresponding data visualization diagram. Secondly, according to the corresponding image area in the image, combined with the indicator system and actual needs, we use four methods to preliminarily screen the indicators in the established indicator system. Finally, through various testing methods, we establish different types of corresponding panel models.

[0044] Method 1 for initial screening of indicators: Screening based on the correlation between independent variables; Step 2.4.1: Screen based on the correlation between the independent variables. If the correlation coefficient is greater than 0.7, it means that the correlation between the two indicators is strong. Only one of the indicators that has a greater impact on the uncultivated land rate is retained. The correlation coefficient graph of the remaining indicators after screening is shown. Step 2.4.2: Perform an F test and a t test. The F test determines whether to use a mixed or fixed-effects model. The F test results show that only the p-value for the time effect is less than 0.05, indicating that a time effect exists, but no individual effect exists. A t test is used to test the significance of each time effect. The corresponding p-values ​​are generally less than 0.05, indicating that a time effect exists. Step 2.4.2: Perform the Hausman test. Then, use the Hausman test to determine whether the effect is fixed or random. The result is p < 0.05 at a significance level of 0.05, so a fixed-effect model should be used. Step 2.4.3 Determine the type of panel model to be established. The F test and Hausman test indicate that a time-fixed-effect model should be used. Step 2.4.4 Solve the model and obtain the results.

[0045] Method 2 for initial screening of indicators: screening based on the correlation between the independent variable and the uncultivated rate; Step 2.4.1: For screening based on the correlation between the independent variable and the non-cultivation rate, when the correlation between two factors is greater than 0.7, their correlation coefficients with the non-cultivation rate can also be calculated, and the factors with higher correlation coefficients with the non-cultivation rate are retained; Step 2.4.2: Perform an F test and a t test. The F test determines whether to use a mixed or fixed-effects model. The F test results show that only the p-value for the time effect is less than 0.05, indicating that a time effect exists, but no individual effect exists. A t test is used to test the significance of each time effect. The corresponding p-values ​​are generally less than 0.05, indicating that a time effect exists. Step 2.4.3: Perform the Hausman test. Then, use the Hausman test to determine whether the effect is fixed or random. The result is p < 0.05 at a significance level of 0.05, so a fixed-effect model should be used. Step 2.4.4 Determine the type of panel model to be established. The F test and Hausman test indicate that a time-fixed-effect model should be used. Step 2.4.5 Solve the model and obtain the results.

[0046] Since the above two methods obtain fewer significant variables affecting uncultivated land, we consider comprehensively screening variables using strong correlation coefficients between independent variables and weak correlation coefficients with the independent variable uncultivated land rate.

[0047] Method 3 for preliminary screening of indicators: Based on method 2, insignificant variables are removed one by one; Step 2.4.1: Among the indicators with a correlation coefficient greater than 0.7 with the uncultivated rate, retain one that has a greater impact on the uncultivated rate of cultivated land, and then remove the variables with insignificant regression coefficients in the regression results. In steps 2.4.2-2.4.5, perform F-tests and Hausman tests to determine the panel model to use and solve the model. After each solution, remove the variable with the largest p-value until all remaining variables are significant. Step 2.5 Identify the key factors affecting the uncultivated rate.

[0048] A panel model was established to select significant variables for uncultivated land. It was concluded that in order to retain more variables while obtaining better results, proportional data should be used. After preliminary screening based on the correlation between the independent variables and the uncultivated rate, the variables that were not significant and had the largest p-values ​​were removed in turn.

[0049] In one embodiment, Figure 5 As shown in the flowchart of the spatial panel module and the spatiotemporal geographic weighted regression module, the spatial panel module analysis includes: Step 3.1 Sample selection: Obtain panel data on non-quantitative rates and influencing factors for the target region and target year interval from relevant departments or official data websites; Step 3.2 Data Preprocessing: Preprocess the sample data. For example, in the indicator system used by the geographic detector model, the altitude of cultivated land and the slope of cultivated land are both indicators that correspond to multiple sets of data. Before solving the panel model, these multiple sets of data need to be merged into a single set of data. Multiply the median of each interval by the proportion of cultivated land area corresponding to that interval, and then add the results of different intervals. Step 3.3 Construct five spatial weight matrices: the spatial adjacency weight matrix, the spatial inverse distance weight matrix, the square of the spatial inverse distance weight matrix, the distance-based spatial adjacency matrix, and the economic distance matrix. Step 3.4: Spatial autocorrelation test: Use the five spatial weight matrices to conduct spatial autocorrelation tests on the uncultivated land rates in the target regions and target years. Step 3.5: Analyze test results. When using a spatial adjacency matrix, if the P value of the global Moran's index for the uncultivated rate in the target year is less than 0.05, but the P values ​​of the global Moran's index for the other weight matrices and years are all greater than 0.05, then there is no spatial autocorrelation.

[0050] In one embodiment, the analysis is performed by a spatiotemporal geographically weighted regression module; Step 4.1 Sample selection: Obtain panel data on non-quantitative rates and influencing factors for the target region and target year from relevant departments or official data websites; Step 4.2 Data Preprocessing: Preprocess the sample data. For example, in the indicator system used by the geographic detector model, the altitude of cultivated land and the slope of cultivated land are both indicators that correspond to multiple sets of data. Before solving the panel model, these multiple sets of data need to be merged into a single set of data. Multiply the median of each interval by the proportion of cultivated land area corresponding to that interval, and then add the results of different intervals. Step 4.3: Construct an indicator system and screen influencing factors; Step 4.3.1 Preliminary construction of the influencing factor indicator system; The influencing factor index system initially constructed by this model includes 5 first-level indicators and 24 second-level indicators; Step 4.3.2: Screen influencing factors based on the correlation coefficient between the indicator variable and the dependent variable; In order to reduce the impact of multicollinearity, the correlation between indicator variables should be reduced, and variables with strong correlation between indicators should be eliminated. When the correlation between two indicator variables is greater than 0.7, the correlation coefficient between each indicator variable and the uncultivated rate is calculated, and the indicator variable with the larger correlation coefficient with the uncultivated rate is retained, thereby reducing multicollinearity. Plot the correlation coefficient graph between the uncultivated rate and each indicator variable and between each indicator variable; when the correlation coefficient between two indicators is greater than 0.7, the indicator variable with the larger correlation coefficient with the uncultivated rate is retained, and the rest are deleted; Step 4.3.3: Screen based on the variance inflation factor between indicator variables; Step 4.4: Preparation of spatiotemporal geographically weighted regression model; Step 4.5: Establish a spatiotemporal geographically weighted regression model; The spatiotemporal geographic weighted regression model incorporates the time and space dimensions into the regression model, fully considering the time and space factors, and making more accurate analysis of variables in different regions. The model combines the original geographic space coordinates with time to form a spatiotemporal three-dimensional coordinate system. .

Claims

1. A system for processing and analyzing arable land data, characterized in that: It includes the module for analyzing factors affecting the uncultivated rate and the module for predicting the uncultivated rate; The uncultivated rate influencing factor analysis module includes a geographic detector model module, a panel model module, a spatial model module, a spatiotemporal geographic weighted regression model module and a comprehensive analysis module; The uncultivated rate prediction module includes a panel model prediction module and a time t prediction module; The geographic detector model module collects uncultivated data and pre-processes missing values; classifies different data types; calculates the q value and p value of each variable based on the geographic detector, determines the influencing factor, and obtains the influencing factors year by year; The panel model module performs a preliminary screening of the uncultivated data and uses the F test and / or t test and / or Hausman test to establish different types of panel models and obtain different significant influencing factors; The spatial model module uses a spatial adjacency weight matrix, a spatial inverse distance weight matrix, a square of a spatial inverse distance weight matrix, a distance-based spatial adjacency matrix, and an economic distance matrix to sequentially perform a spatial correlation test on the uncultivated rate; if the test passes, a spatial model is established; if not, no spatial model is established; The spatiotemporal geographic weighted regression model module constructs an influencing factor system including multiple primary indicators and multiple secondary indicators, uses panel data of uncultivated rates in target areas and target years, screens related indicators through correlation coefficients, and uses variance inflation factors to screen indicator variables. The spatiotemporal geographic weighted regression model is established and solved to obtain the coefficients of each indicator in the target area. The comprehensive analysis module organizes the results of the spatiotemporal geographic weighted regression model of the target area, screens significant features, and combines the results of each model solution to determine the degree of influence of each significant indicator based on the analysis results, ultimately obtaining the results of significant influencing factors of non-cultivation in the target area; The panel model prediction module constructs different types of regression models based on the change characteristics of the respective variables to make a preliminary prediction of the uncultivated rate for the next year; After completing the independent variable prediction, according to the solution results of the panel model and combined with the obtained independent variable prediction values, a time series prediction model is constructed based on the panel model parameter estimation results to further predict the corresponding uncultivated rate of each region in the next year; The time t prediction module divides regional types through Euclidean clustering and hierarchical clustering; for all cities in each category, a dynamic prediction model of uncultivated rate based on the time dimension is established in combination with the regression prediction mechanism, and regression prediction is performed. The regression models used include linear regression model, quadratic regression model, exponential regression model, logarithmic regression model, exponential decay model and trend extrapolation; a prediction model that matches each city category is selected to obtain the corresponding uncultivated rate prediction value.

2. The uncultivated land data processing and analysis system according to claim 1, characterized in that: Different kinds of panel models, including mixed models, fixed-effect models, and random-effect models.

3. The system for processing and analyzing cultivated land data according to claim 1, wherein: The spatiotemporal three-dimensional coordinates in the spatiotemporal geographic weighted regression model module are , the expression of the spatiotemporal geographically weighted regression model is as follows: ； Where n is the number of sample points, r is the number of independent variables, 、 、 are the longitude, latitude and time of the i-th sample point respectively, is the dependent variable of the i-th sample point, is the kth independent variable of the i-th sample point, is the intercept term, is the regression coefficient of the kth independent variable of the ith sample point, is the random error of the i-th sample point that is independent and normally distributed.

4. A method for processing and analyzing arable land uncultivated data, characterized in that: The system for processing and analyzing arable land data according to any one of claims 1 to 3 comprises the following steps: Step 1: Analyze and collect relevant indicator data on the uncultivated land rate through the geographic detector module, preprocess the indicator data, and classify the indicator types of the uncultivated land data in the target area using the equal interval method, quantile method, and systematic clustering method; calculate the q value and p value of each independent variable in turn according to the geographic detector method; and determine the factors affecting the uncultivated land in each year and their explanatory power based on the p value and / or q value; Step 2: Establish an influencing factor index system through panel model module analysis, and establish different types of panel models through F test and / or t test and / or Hausman test; Step 3: Obtain panel data on the uncultivated land rate and influencing factors for the target region and target year through spatial model module analysis, and perform spatial autocorrelation tests on the uncultivated land rate for the target region and target year using the spatial adjacency weight matrix. Step 4: Obtain panel data on the uncultivated rate and influencing factors in the target region and target year through spatiotemporal geographic weighted regression analysis; construct an influencing factor indicator system consisting of multiple primary and secondary indicators; screen indicator variables based on the variance inflation factor; and construct a spatiotemporal geographic weighted regression model to obtain the coefficients of each indicator in the target region. Step 5: Analyze and organize the output data of the spatiotemporal geographic weighted regression model through the comprehensive analysis module, and extract the p-value and q-value data from them. At the same time, integrate the output results of the panel model, spatial model, and spatiotemporal geographic weighted regression model, and record the impact direction and degree of each indicator in detail. Systematically organize the integrated data and screen out the significant indicator results based on the rounding-up strategy. The screened significant indicators are sorted in descending order according to their q-values. Finally, the results of the significant influencing factors of uncultivated land in the target area are obtained. Step 6: Uncultivated crop forecast analysis. Based on the panel model, the uncultivated crop rate forecast is performed. After completing the independent variable forecast, the panel model solution is used to combine the obtained independent variable forecast values ​​to predict the uncultivated crop rate for each administrative district in the next year. Based on the uncultivated crop rate forecast at time t, regional types are divided through Euclidean clustering and hierarchical clustering. For all cities in each category, a dynamic prediction model of uncultivated rate based on the time dimension is established in combination with the regression prediction mechanism, and regression prediction is performed; corresponding evaluation criteria are constructed, and the accuracy of the prediction results is measured by calculating the absolute error and relative error of each uncultivated rate prediction value.

5. The method for processing and analyzing uncultivated land data according to claim 4, characterized in that: For the result data output by the spatiotemporal geographic weighted regression model, the significance indicators in each year are systematically merged according to the pre-set three-level target area division rules, and finally an independent and complete solution result data set is formed for each target area. First, the number of three-level target areas is accurately counted, and the significance indicators obtained are screened based on the screening threshold. The screened indicators and their related data are arranged and recorded in order according to the preset data format and arrangement rules, and finally a spatiotemporal geographic weighted regression model solution result table is constructed.

Images