A real estate data visualization display method and system based on GIS technology

Abstract

The application provides a real estate data visualization display method and system based on GIS technology, which comprises the following steps: collecting, cleaning, standardizing, mapping and aligning multiple-source real estate data such as registration, transaction, planning approval, justice, tax, sensing terminal and remote sensing image, establishing a space-time joint index with space unit ID and time slice ID as the primary key, and extracting a feature set and a target attribute set for modeling; constructing a structured causal model in the space-time database, establishing a causal query index through a structure learning algorithm and parameter training; receiving a scenario parameter set, parsing it into an intervention instruction, executing a do-intervention operation on the causal model, collecting user interaction behavior and external observation data, triggering online calibration of the model and mapping rules when the error exceeds a threshold, updating the causal query index and the mapping threshold and supporting version rollback, and forming a data, model, display, feedback closed-loop link.

Description

Technical Field

[0001] This invention relates to the field of computer data visualization technology, and in particular to a method and system for visualizing real estate data based on GIS technology. Background Technology

[0002] Existing real estate information management systems are typically based on Geographic Information System (GIS) technology. By overlaying land parcel, building, and ownership information on electronic maps, they enable spatial queries, statistical analysis, and graphical displays. Such systems can intuitively reflect the geographical location, use, and area of ​​real estate and have been widely used in urban planning, land supervision, and real estate transaction management.

[0003] However, with the expansion of urban construction and frequent changes in policy regulation, real estate data exhibits characteristics of multi-source heterogeneity, strong time-varying nature, and complex correlations. Traditional GIS systems mostly adopt static mapping methods, which only support the display of current status information and simple attribute queries, and have the following shortcomings:

[0004] 1. Traditional GIS systems are not suitable for simulating the dynamic evolution of real estate value, use, or spatial layout under conditions of planning adjustments, policy interventions, or market changes;

[0005] 2. Traditional GIS system visualization methods are mostly passive displays, lacking computational models based on causal relationships or logical inferences. They cannot explain the causes of changes in real estate, nor can they predict potential impacts based on hypothetical scenarios.

[0006] Therefore, we propose a method and system for visualizing real estate data based on GIS technology.

[0007] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0008] The purpose of this invention is to address the shortcomings of existing technologies by providing a method and system for visualizing real estate data based on GIS technology, thereby resolving the technical problems mentioned in the background section.

[0009] To achieve the above objectives, the present invention provides the following technical solution:

[0010] A method for visualizing real estate data based on GIS technology includes the following steps:

[0011] S1. Clean, deduplicate, unify the format and align the original real estate data from multiple sources with entities, establish a real estate spatiotemporal database, construct a unified identification system and spatiotemporal index for spatial unit sets and their events, and form a causal modeling feature set and target attribute set.

[0012] S2. Determine the structural relationships between causal variables, train and determine the structured causal model and model parameters, establish a causal query index for performing causal queries in the real estate spatiotemporal database, and provide the distribution of target attributes under different intervention conditions without changing the data sampling distribution.

[0013] S3. Convert the scenario parameter set into intervention instructions that can be executed on the structured causal model, calculate scenario outcome data based on the intervention instructions, and calculate baseline outcome data under the condition of no intervention within the same time window;

[0014] S4. Perform differential operations on the scenario result data and baseline result data in the spatial unit and time slice dimensions to obtain a differential tensor. Based on the preset differential-to-visual coding mapping rules, convert the differential tensor into a visual coding parameter set and write the differential tensor and the visual coding parameter set into the spatiotemporal index.

[0015] S5. The visualization engine generates a scenario visualization view on the GIS map according to the set of visual coding parameters. Within the user's view range, it performs hierarchical expansion and detailed association display of spatially significant units. It also provides a linkage comparison between the current view and the scenario view on the same interface, expressing the spatial changes of real estate under intervention conditions through differentiation.

[0016] S6. Record user interaction behavior in the scenario visualization view and posterior observation data from external systems, calculate the error index between scenario result data and actual observation, perform online calibration of structured causal model and differential to visual coding mapping rules based on error index, and write back the calibrated model parameters and mapping rules to causal query index and spatiotemporal index for subsequent scenario visualization display.

[0017] S1 specifically includes: acquiring original data on real estate registration, transactions, planning approvals, judicial matters, taxation, sensor terminals, and remote sensing images; identifying the source and collection time of the data source; and forming a set of data to be processed.

[0018] The data set to be processed is cleaned, deduplicated, and formatted. The name, code, coordinates, timestamp, and ownership fields are normalized and mapped to form a set of data entries with a consistent structure.

[0019] A unified identification system is established based on land parcels, individual buildings and interior units to complete entity alignment and relationship alignment, and to map historical change records into a time-oriented event sequence.

[0020] A spatiotemporal database of real estate is constructed under a geographic coordinate system. Spatial and temporal indexes are established for spatial unit sets and event sequences, respectively, and a spatiotemporal composite index is generated to support spatial queries over time.

[0021] Extract the feature set and target attribute set for causal modeling from a data entry set with consistent structure. Associate the extraction results with the spatiotemporal joint index and store them to obtain the feature view for subsequent modeling.

[0022] S2 specifically includes: determining the set of causal variables based on the feature set and the target attribute set, combined with real estate business constraints, to form a set of causal candidate relationships;

[0023] Structural learning and constraint verification are performed on the causal candidate relation set to determine the directed acyclic graph structure of the structured causal model and to record the parent-child relationships that are accessible for intervention.

[0024] Historical samples from a spatiotemporal database are used to learn the parameters of a structured causal model, generating a set of model parameters, and a stable range of parameter values ​​is determined through cross-validation.

[0025] A causal query index is established in the spatiotemporal database, and the directed acyclic graph structure, model parameter set and variable encoding rules are persisted in the form of index entries to support subsequent causal queries and intervention calculations.

[0026] Consistency checks are performed on the structured causal model, and edges or parameters that do not meet statistical significance are marked as low-confidence entries. A fallback strategy is set in the causal query index to ensure query stability.

[0027] S3 specifically includes: receiving a scenario parameter set, which includes intervention descriptions of changes in planning conditions, addition of public facilities, or adjustment of restrictive policies, and completing parameter legality verification and unit standardization;

[0028] Based on the scenario parameter set and causal query index, the intervention description is converted into intervention instructions that can be executed on the structured causal model, and the intervention variables and their value methods are clearly defined;

[0029] Within a unified time window, scenario outcome data are calculated based on a structured causal model and intervention instructions to obtain the target attribute distribution and spatiotemporal location results under the intervention conditions.

[0030] Under conditions consistent with the unified time window, baseline result data are calculated without intervention to obtain the target attribute distribution and spatiotemporal positioning results under the current conditions;

[0031] Align the scenario results data with the baseline results data according to spatial units and time slices to generate a set of data pairs to be differentiated, and record the corresponding relationship in the spatiotemporal joint index.

[0032] S4 specifically includes: performing difference operations on the paired data set, calculating the difference tensor of the target attribute by spatial unit and time slice, and obtaining the difference result used to represent the magnitude and direction of change;

[0033] According to the preset differential-to-visual encoding mapping rules, the differential tensor is mapped to a set of visual encoding parameters, which includes layer selection, symbolization style, layer expansion level and annotation priority.

[0034] Threshold pruning and conflict resolution are performed on the visual encoding parameter set to ensure that layer on / off, symbolization style and hierarchical expansion level do not contradict each other within the same view range;

[0035] Write the difference tensor and the set of visually encoded parameters into the spatiotemporal joint index, establish index entries from spatial unit identifiers to the set of visually encoded parameters, and form a parameter view that can be directly consumed by the visualization engine.

[0036] Set significance and update flags for the difference results, and record the generation time and data source for subsequent view refresh and incremental update.

[0037] S5 specifically includes: the visualization engine initiates a query to the spatiotemporal joint index based on the user's view range to obtain the differential tensor and visual encoding parameter set corresponding to the view range;

[0038] The visualization engine generates scene visualization views on the GIS map according to the set of visual coding parameters, and enables hierarchical expansion and detailed association display for spatial units with significant differences;

[0039] The current status view and the scenario view are generated and compared on the same interface to maintain spatial positioning consistency, and the spatial changes of real estate under intervention conditions are expressed by difference.

[0040] Incremental rendering and caching are performed on layers, labels, and hierarchical expansion within the user's view range to ensure continuity and stability during view scrolling and zooming;

[0041] The generated results of the scenario visualization view and the query parameters are recorded as view session entries for subsequent feedback collection and online calibration.

[0042] S6 specifically includes: recording user interaction behavior in the context visualization view, including click, pause, comparison and back operations, and establishing association with the corresponding spatial unit identifier;

[0043] Receive posterior observation data from external systems, verify and align the posterior observation data in the same spatial unit and time slice dimension to form an aligned data set for evaluation;

[0044] Error indices between scenario results data and actual observations are calculated based on the aligned dataset. These error indices include deviation and matching metrics, and an error assessment report is generated.

[0045] Based on the error assessment report, the structured causal model and the differential-to-visual coding mapping rules are calibrated online, and the model parameter set and mapping threshold are updated to ensure that subsequent calculations and displays are consistent with actual observations.

[0046] The calibrated model parameter set, mapping threshold, and related tags are written back to the causal query index and spatiotemporal joint index, triggering version updates of the corresponding index entries for subsequent scenario visualization.

[0047] A real estate data visualization system based on GIS technology includes:

[0048] The data modeling and indexing module is used to collect original real estate data from registration, transactions, planning approval, judicial, tax, sensor terminals and remote sensing images, clean the data, unify the format and align the entities, and establish a set of spatial units and its event identification system.

[0049] The causal model construction module is used to determine the structural relationship between causal variables based on the feature set and the target attribute set, establish a structured causal model through structural learning algorithms and parameter training, and generate a causal query index.

[0050] The scenario intervention and counterfactual inference module is used to receive a scenario parameter set and perform legality verification, parse the scenario parameter set into intervention instructions, perform do-intervention operations on the structured causal model, and calculate scenario outcome data under the intervention conditions.

[0051] The differential mapping and visual encoding generation module is used to perform differential operations on the scenario result data and baseline result data in the spatial unit and time slice dimensions, generate differential tensors, and generate corresponding visual encoding parameter sets according to preset differential-to-visual encoding mapping rules.

[0052] The visualization view generation and linkage display module is used by the visualization engine to generate a scene visualization view on the GIS map according to the set of visual encoding parameters, and to perform layer expansion and detail display on spatial units with significant differences within the user's view window.

[0053] The feedback acquisition and online calibration module is used to collect user interaction behavior and posterior observation data from external systems, calculate the error index between scenario result data and observation values, and trigger online calibration of the structured causal model and visual coding mapping rules when the error exceeds a preset threshold.

[0054] The beneficial effects of this invention are as follows:

[0055] This invention solves the problems of multiple sources, heterogeneous formats, and inconsistent time resolution of real estate data by constructing a spatiotemporal joint index with spatial unit ID and time slice ID as the primary keys. It realizes the unified integration and efficient retrieval of multi-source data such as registration, transaction, planning, taxation, and judicial data, and provides a stable data foundation for subsequent modeling and visualization.

[0056] This invention establishes a structured causal model in a spatiotemporal database and determines the logical dependencies between variables through structure learning and parameter training. It can simulate different scenarios such as planning adjustments, policy interventions, or market fluctuations without changing the original data distribution, and perform "do-intervention" operations to obtain counterfactual results, thereby achieving interpretable deduction of changes in real estate attributes.

[0057] This invention obtains the difference tensor between the scenario result and the baseline result through differential calculation, and introduces a saliency determination and visual encoding mapping function to automatically convert the magnitude and direction of change into layer identifiers, symbol styles, hierarchical expansion levels and annotation priorities, thereby realizing automated mapping from data calculation to visualization rendering and significantly improving the intuitiveness and operability of the display.

[0058] This invention generates simultaneous and linked current status and scenario views on the same GIS interface, supporting differential overlay, highlighting significant areas, synchronous zooming, and cross-view alignment operations. This allows users to directly compare spatial differences and structural changes of real estate under different scenarios, improving decision-makers' perception accuracy of the effects of planning adjustments or policies.

[0059] This invention collects user interaction behaviors (clicks, dwell times, comparisons, selections, etc.) and posterior observation data provided by external systems to calculate the error index between the predicted results and the actual results. When the error exceeds a threshold, it automatically triggers online calibration of the causal model and visual mapping rules. The system stability is ensured through version management and rollback mechanisms, thereby achieving continuous model optimization and dynamic adaptation of visualization results.

[0060] This invention can not only intuitively reflect the current status of real estate, but also perform interpretable calculations and counterfactual analysis for different intervention factors, realizing the transformation from "static display" to "intelligent prediction and verification", providing technical support with reasoning and feedback learning capabilities for business scenarios such as real estate supervision, planning evaluation, and market analysis. Attached Figure Description

[0061] Figure 1 This is a schematic diagram of a real estate data visualization method based on GIS technology according to the present invention;

[0062] Figure 2This is a schematic diagram of the framework of a real estate data visualization system based on GIS technology according to the present invention. Detailed Implementation

[0063] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0064] Example 1: As Figure 1 As shown, this embodiment provides a method and system for visualizing real estate data based on GIS technology, including the following steps:

[0065] S1. Clean, deduplicate, unify the format and align the original real estate data from multiple sources with entities, establish a real estate spatiotemporal database, construct a unified identification system and spatiotemporal index for spatial unit sets and their events, and form a causal modeling feature set and target attribute set.

[0066] S2. Determine the structural relationships between causal variables, train and determine the structured causal model and model parameters, establish a causal query index for performing causal queries in the real estate spatiotemporal database, and provide the distribution of target attributes under different intervention conditions without changing the data sampling distribution.

[0067] S3. Convert the scenario parameter set into intervention instructions that can be executed on the structured causal model, calculate scenario outcome data based on the intervention instructions, and calculate baseline outcome data under the condition of no intervention within the same time window;

[0068] S4. Perform differential operations on the scenario result data and baseline result data in the spatial unit and time slice dimensions to obtain a differential tensor. Based on the preset differential-to-visual coding mapping rules, convert the differential tensor into a visual coding parameter set and write the differential tensor and the visual coding parameter set into the spatiotemporal index.

[0069] S5. The visualization engine generates a scenario visualization view on the GIS map according to the set of visual coding parameters. Within the user's view range, it performs hierarchical expansion and detailed association display of spatially significant units. It also provides a linkage comparison between the current view and the scenario view on the same interface, expressing the spatial changes of real estate under intervention conditions through differentiation.

[0070] S6. Record user interaction behavior in the scenario visualization view and posterior observation data from external systems, calculate the error index between scenario result data and actual observation, perform online calibration of structured causal model and differential to visual coding mapping rules based on error index, and write back the calibrated model parameters and mapping rules to causal query index and spatiotemporal index for subsequent scenario visualization display.

[0071] S1 specifically includes the following sub-steps:

[0072] S110. Multi-source real estate data collection: Collect original real estate data from multiple sources, including registration, transactions, planning approvals, judicial data, tax data, sensor terminals, and remote sensing images; periodically pull data through REST API and batch processing interface, and record the source identifier and collection time identifier of each data point.

[0073] Perform anomaly detection on all collected items, including: whether the timestamp exceeds the tolerance range; whether the coordinate offset exceeds the preset threshold (e.g., 50m); and whether the field missing rate exceeds the tolerance. If any detection result is abnormal, mark the item as invalid and remove it.

[0074] This ultimately forms a structured set of data to be processed, ensuring the stability and consistency of subsequent processing.

[0075] S120. Data Cleaning and Standardization Mapping: Perform field standardization and structure cleaning on the dataset to be processed, including mapping fields such as codes, coordinates, timestamps, and ownership to a unified format, such as mapping the original "land parcel number" to the standard field ID_X;

[0076] Duplicate entries are deduplicated using a timestamp-first rule, and the latest valid record is selected:

[0077]

[0078] in This represents the data object or input sample to be evaluated; Represents the candidate mapping function, parameter set, or visualization scheme; Indicates the input object The following are candidate solutions The calculated performance indicators or scores; This indicates that among all candidate solutions, the one that makes Index of the scheme that takes the maximum value.

[0079] Therefore, the function This represents the process of automatically selecting the optimal solution from multiple candidate solutions, i.e., keeping the solution that makes the evaluation function... The largest candidate strategy is selected to ensure that the optimal mapping or parameter configuration is always used during system execution; numerical normalization and encoding consistency verification are performed to remove entries that do not conform to the mapping rules; the cleaning result forms a set of standard data entries with unified field format and redundancy elimination.

[0080] S130, Real Estate Entity Alignment and Event Sequence Construction: Aggregate standard data entries according to land parcels, individual buildings, and interior units.

[0081] A spatial overlay algorithm (based on polygon overlap ≥ 0.8), unique identifier primary key matching, and fuzzy string matching are combined to align entries from different sources.

[0082] Based on entity alignment, historical timestamp information is extracted, arranged in ascending order of time, and a time-directed event sequence is constructed.

[0083] When there are time conflicts or multiple overlapping records, the "latest valid priority + stable time order" strategy is adopted to resolve the conflicts, ensuring the uniqueness of the event sequence and the correctness of the temporal logic.

[0084] S140. Spatiotemporal Index Establishment and Primary Key Design: Establish spatial and temporal indexes for spatial unit sets and event sequences respectively under the geographic coordinate system.

[0085] The spatial index uses an R tree structure to support fast spatial range queries, while the temporal index is divided into fixed time slices (default 24 hours per time slice):

[0086]

[0087] in Indicates the time slice number; Indicates the timestamp of the data record; Indicates the time segmentation interval, preferably 24 hours (i.e., one calendar day); symbol This indicates a floor operation; the system uses "spatial unit ID" and "time slice ID" as a combined primary key to establish a spatiotemporal composite index table, and records the corresponding geographical location, time slice label and associated event ID in each index entry, thereby realizing efficient spatiotemporal query and rapid location of real estate data.

[0088] S150. Feature set and target attribute set extraction: Extract the feature set and target attribute set that can be used for subsequent causal modeling from the standard data entries.

[0089] The feature set includes numerical fields (such as building area and land price per unit area), enumerated fields (such as planned use and ownership type), and time series fields (such as transaction frequency curve).

[0090] The target attribute set includes land price, plot ratio, probability of litigation risk, number of ownership dispute records, etc.

[0091] The feature extraction process is achieved through field filtering and statistical aggregation, forming a feature vector for each spatial unit at each time slice:

[0092]

[0093] Where i represents the spatial unit ID; t represents the time slice ID; The characteristic quantities corresponding to the spatial unit and time slice include numerical fields (such as building area and land price per unit area), enumerated fields (such as planned use and ownership category), and time series fields (such as transaction frequency curve).

[0094] The above formula describes the multidimensional feature aggregation vector under each spatial unit i and time slice t. During feature extraction, the system normalizes multiple original records through field filtering and statistical aggregation operations, ensuring that each spatiotemporal unit can form a unique feature vector. The feature vector is then linked to the spatiotemporal joint index in a one-to-one or one-to-many manner for use in subsequent causal modeling.

[0095] S2 specifically includes the following sub-steps:

[0096] S210. Determining the set of causal variables and generating candidate relation sets: Extracting the set of causal variables from the feature set and the target attribute set. And generate a set of causal candidate relations based on domain knowledge and data dependencies. .

[0097] The causal candidate relation set is obtained by fusing the following three types of information:

[0098] Spatial adjacency relationships (such as the history of ownership changes of adjacent land parcels).

[0099] Temporal dependencies (such as the lagged effect of transaction frequency on price changes);

[0100] Business logic prior (e.g., changes in planned land use affect land unit price).

[0101] For all candidate relation record variable pairs and time lag parameters This forms the initial set of directed edges.

[0102] in This represents the i-th causal variable, derived from fields in the feature set and the target attribute set; n represents the total number of variables. This represents the complete set of causal variables, used for subsequent structure learning and causal edge relationship identification.

[0103] S220. Causal Structure Learning and Directed Acyclic Graph Construction: Based on the initial candidate relations, a structure learning algorithm is used to determine the final causal directed acyclic graph (DAG).

[0104] Specifically, it includes:

[0105] Constraint-based PC algorithms or score-based GES algorithms are used to determine the existence of directed edges between variables; for time-dependent relationships, lag parameters are incorporated. Temporal consistency checks are performed, and edges that do not meet the temporal constraints are removed; the Bayesian Information Criterion (BIC) is used to score the candidate graph structures, and the structure with the best score is retained; finally, the causal structure is obtained. ,in It is a set of causal directed edges; if there is no solution due to structural conflict, it reverts to the previous feasible graph structure version and records the revert flag.

[0106] S230. Causal Model Parameter Training and Stability Interval Determination: Utilizing Historical Samples from a Spatiotemporal Database The parameters of the conditional probability distribution corresponding to each edge in the causal graph structure are trained.

[0107] For continuous variables, use linear or Gaussian regression:

[0108]

[0109] in Represent the target variable; It represents the set of its parent nodes (i.e., the set of direct dependent variables); The mean is variance is The Gaussian distribution is used to describe The probability distribution characteristics given a parent node.

[0110] For discrete variables, the system employs a multinomial distribution model and calculates parameters based on the maximum likelihood estimation method; the trained model parameters are denoted as... Its stability interval was evaluated using cross-validation. When a parameter estimate is detected to be outside the stable range, the system determines that the parameter is an anomalous parameter and automatically triggers model resampling or structural backoff operations to ensure the convergence and stability of the causal model.

[0111] S240. Causal Query Index Construction and Storage: Write the causal graph structure and parameter information into the causal query index.

[0112] The entry format for a causal query index is as follows: VarID: causal variable identifier; PaList: list of parent node identifiers; Theta: parameter vector or conditional probability table; Lag: lag parameter. EdgeConf: Edge confidence value; RollbackFlag: Rollback status flag; TimeStamp: Index update time.

[0113] The index adopts a key-value storage structure with (VarID, TimeStamp) as the primary key, which can quickly support subsequent scenario intervention and counterfactual inference. When the causal model structure changes, the causal query index is automatically updated, and the old version is automatically archived to support historical backtracking.

[0114] S250, Causal Model Consistency Verification and Rollback Strategy: After completing the causal query index construction, a consistency verification is performed, including:

[0115] Topological directed acyclicity check to ensure no cyclic paths; causal strength threshold check to remove low-confidence edges (e.g., EdgeConf < 0.6); parameter interpretability check to ensure the direction of hysteresis relations is consistent with the domain logic; when any check fails, a rollback strategy is triggered: rollback is prioritized to the structure version that passed the most recent check.

[0116] If there is no version to roll back, then a minimal skeleton graph structure is used (preserving the most basic business logic edges); the consistency verification results are written back to the causal query index in the form of tags for subsequent intervention command parsing and inference calls.

[0117] S3 specifically includes the following sub-steps:

[0118] S310. Scenario Parameter Set Reception and Validity Verification: Receive scenario parameter set S, which is represented in structured JSON format and includes the following fields:

[0119] target_var: The target variable for intervention (such as planned use, number of public facilities, tax rate coefficient);

[0120] intervention_type: Intervention type (numerical increase / decrease, category replacement, range scaling);

[0121] intervention_value: Intervention value or intervention range;

[0122] spatial_scope: Spatial scope (list of parcel IDs or polygonal region).

[0123] temporal_scope: The time scope (start time slice and end time slice).

[0124] The sources of scenario parameter sets include: manual input; automated strategy recommendation engines; and external business system APIs. Upon receiving a scenario parameter set, a validity check is performed, including field integrity check, value range check, and whether the intervention target exists in the causal query index. If any check fails, the parameter set is marked as invalid, triggering a rollback mechanism and returning the check log.

[0125] S320, Intervention Instruction Parsing and Intervention Node Mapping: Intervention instructions are parsed from the valid scenario parameter set S to generate intervention instructions. Its structure is as follows:

[0126]

[0127] in To intervene in the target variable, For intervention operations (such as adding values ​​or replacing categories). Where is the spatial range of influence, and T is the temporal interval of influence.

[0128] The parsing process includes: matching the corresponding target variable in the causal query index. Interventional edges; intervention values ​​are assigned based on intervention type. Convert the code into an opcode that the model can recognize; verify the parent-child relationship of the intervention target to ensure that the intervention location exists in the directed acyclic graph and is an unambiguous path; if several pre-targets do not match or paths conflict, trigger the intervention rollback strategy to roll back to the most recent effective intervention configuration.

[0129] S330, Scenario Outcome Data Calculation (Intervention Implementation): Based on Intervention Instructions Perform do-intervention operations on a structured causal model:

[0130]

[0131] Where Y represents the target attribute set; Indicates the causal variable to which intervention was applied; For fixed values ​​set in the intervention instructions; symbol This refers to the exogenous intervention in a causal model, i.e., cutting off Connect the parent node to the edge and force it to be set to a value. .

[0132] The execution process includes:

[0133] In the cause-effect graph The parent node connection is broken, and its value is set to 0. ;

[0134] The intervention effect is propagated from top to bottom along the directed acyclic graph, and the conditional distribution of downstream variables is calculated sequentially according to topological sorting; Monte Carlo sampling is used to perform multiple inferences on each spatial unit i and time slice t:

[0135]

[0136] Where M is the number of samples. This is a causal inference function; used to calculate the causal inference function in the context of intervention instructions. The target output value is as follows; This represents the set of input variables after intervention in the m-th sampling. When the sampling variance exceeds the preset threshold, the system automatically increases the number of samplings M or reverts to the mean estimation mode to ensure the stability of the results.

[0137] S340, Baseline Outcome Data Calculation (No Intervention Applied): Within the Same Time Window and spatial range Within, without applying any intervention, execute:

[0138]

[0139] That is, calculating the baseline result values ​​of the target attribute set according to the original causal model:

[0140]

[0141] in Let be the original feature vector of spatial unit i; This is the prediction function for the original causal model; after calculation, the scenario outcome data and baseline outcome data are indexed and aligned to ensure... One-to-one correspondence.

[0142] S350, Data Alignment and Anomaly Rollback Mechanism: [Regarding...] and Perform alignment operations at the spatial unit and time slice levels to establish the correspondence:

[0143]

[0144] in T is a set of spatial units; T is a set of time slices; This represents a one-to-one pairing set of intervention results and baseline results, used for subsequent differencing calculations. When the alignment process encounters issues such as missing spatial units, missing time slices, or an empty set of intervention results, an automatic rollback strategy is triggered, including: rolling back to the previous valid scenario calculation result; if no valid record is found, using the model baseline value as the default value; ultimately generating an aligned set of paired data. And write it into the spatiotemporal joint index for difference calculation.

[0145] S4 specifically includes the following sub-steps:

[0146] S410. Calculation and significance determination of difference tensors:

[0147] Scenario outcome data Compared with baseline results data Perform the difference calculation between spatial unit i and time slice t to obtain a two-dimensional difference tensor:

[0148]

[0149] Where: i is the spatial unit ID; t is the time slice ID; This represents the change in the target attribute.

[0150] Further significance determination is performed on the difference tensor:

[0151]

[0152] in The significance threshold for the difference is set by the system based on historical distribution or configuration strategy; when... When, it indicates that the intervention had a significant impact on the spatial unit at time slice t; when... At that time, the change was considered to be within the range of fluctuation and not statistically significant; significance markers Used for subsequent layer expansion and annotation priority determination.

[0153] S420. Mapping from Difference Tensors to Visual Coding Parameter Sets: Mapping difference tensors and saliency labels to visual coding parameter sets. The mapping rule is defined as follows:

[0154]

[0155] in This is a visual encoding mapping function, and the mapping result includes the following fields: This is a set of visual rendering parameters used to control layer rendering styles, color gradients, transparency, and other attributes.

[0156] LayerFlag: Determines the layer based on the difference attribute category; SymbolStyle: Based on... The sign and size determine the color gradient and transparency; LODLevel (Level of Deployment): based on the salience marker. Determine the expansion level; LabelPriority (mark priority): based on... Prioritize based on threshold grading.

[0157] For example, symbolization styles can use linear mapping:

[0158]

[0159] in This is the upper limit threshold for mapping, used to limit the maximum visible transparency; when Approaching or exceeding When the element is completely opaque, it is not transparent at all; when... When the elements are small, they are semi-transparent to show subtle changes.

[0160] S430, threshold pruning, conflict resolution, and mapping stability control: for all spatial units of Perform global pruning and conflict resolution:

[0161] when Below At that time, Marked as "Skip Rendering";

[0162] If multiple layers conflict in the same spatial location, they are sorted according to LabelPriority, with priority given to retaining parameter entries with higher saliency.

[0163] Apply smoothing strategies (such as moving averages) to the time-slice successive difference results to eliminate rendering flicker.

[0164] If an error occurs during the trimming or resolution process (such as empty mapping parameters or unresolved conflicts), the system will automatically revert to the default rendering template to ensure system availability.

[0165] S440, Differential and Encoding Parameter Index Writing: Write the differential tensor Significant markers With visual encoding parameter set Write the spatiotemporal composite index according to the spatial unit ID and time slice ID. The index structure is as follows:

[0166] Key: (i,t); DeltaValue: Significance: VisualParam: UpdateFlag: Parameter update flag; Version: Index version number.

[0167] Atomic operations are used during writes to ensure index consistency and traceability of rendering results. If a write fails, the system automatically rolls back to the previous version of the index entry.

[0168] S450, Status Marking and Incremental Update Mechanism: Perform status marking on the written differential tensors and visual coding parameter sets:

[0169] when When an item is marked as "active", it will be forcibly loaded during subsequent rendering.

[0170] when When this happens, the entry will be marked as "lazy" and will only be loaded when the user zooms in or filters.

[0171] when When the update frequency exceeds the threshold, an incremental update mechanism is triggered, which only refreshes the changed items to avoid a full redraw.

[0172] Status markers and incremental update information are synchronously recorded in the index update log for subsequent view scheduling and performance optimization.

[0173] S5 specifically includes the following sub-steps:

[0174] S510, Visualization Engine Query and Rendering Range Determination: When a user browses, zooms, or pans in the GIS interface, the system determines the range based on the geographical boundaries of the current viewport. and time slice interval Generate a query request.

[0175] The visualization engine initiates a query to the spatiotemporal composite index:

[0176]

[0177] Return the corresponding difference tensor and visual encoding parameter set .

[0178] If no entries are returned during the query process, the default layer is loaded (reverting to the template) to ensure view integrity.

[0179] S520. Layer rendering of the scenario visualization view: The visualization engine generates a scenario visualization view on the GIS map based on the query results. The rendering process includes:

[0180] Press LayerFlag to select all Assign to a predefined layer;

[0181] Render the color, transparency, boundary thickness, and gradient direction of spatial unit i according to SymbolStyle;

[0182] Perform layer expansion based on LODLevel, and load vector or tile data of different resolutions;

[0183] Control the visibility of annotation text using LabelPriority to avoid overlapping annotations;

[0184] When the difference significance indicator At that time, force the rendering of highlight effects.

[0185] Overall rendering follows:

[0186]

[0187] in For rendering functions, A graphical entity representing the corresponding spatial unit i and time slice t; The geographic boundaries of the currently visible window; This function represents the current time slice interval; it enables dynamic drawing and real-time updating of spatial features within the viewport.

[0188] S530, Linked Comparison of Current Status View and Scenario View: The system generates simultaneously on the same interface:

[0189] Current Status View :based on ;

[0190] Scenario View :based on .

[0191] Both maintain complete consistency in their spatial reference frame, time axis, and scaling level.

[0192] The system establishes a mapping during the rendering process:

[0193]

[0194] in This is a view linkage mapping function used to ensure that when a user performs interactive operations such as mouse hovering, clicking, or selecting in the current view, the corresponding spatial unit in the context view is simultaneously highlighted and positioned, thereby achieving linkage comparison.

[0195] Users can perform differential observation using sliders, split screens, or mixed overlays.

[0196] when Exceeding the significance threshold When the system automatically highlights the corresponding areas between the two views, it provides an intuitive magnification of the differences.

[0197] S540, Incremental Rendering and View Figure 1 Consistency control: To avoid lag caused by a full refresh, the system adopts an incremental rendering mechanism.

[0198] Only refresh when the user view changes. or Entries with a status of "Active" or "Updated":

[0199]

[0200] in To update the flag, it is set to 1 when the data in spatial cell i or time slice t changes significantly; This represents the minimum dataset that needs to be redrawn in the current frame; the system only... The rendering operation is performed on the entries in the layer, while the content of the unupdated layer is cached and reused, thereby significantly reducing the rendering load.

[0201] In addition, the system records the layer check hash value before and after rendering. and ,when When, it indicates that the two renderings are consistent, maintaining the current view state; when When this happens, the system automatically rolls back to the previous frame's rendering snapshot to prevent screen flickering or layer misalignment. This verification mechanism ensures the stability and consistency of the view rendering process.

[0202] S550, View Session Recording and Interaction State Synchronization: After rendering, the system records the session entries for this scenario visualization view, including:

[0203] Current window range With time interval ; A snapshot of the loaded layer ID and the set of visual encoding parameters of the mapping;

[0204] User interaction trajectories (panning, zooming, clicking, split-screen operations, etc.); statistical summary of the difference significance distribution.

[0205] All session entries are written to the session log index for later use in model calibration, scenario playback, or difference demonstration. When a view fails to render, has missing parameters, or terminates abnormally, the system rolls back to the rendering snapshot corresponding to the most recent session entry to ensure view recovery and reproduction capabilities.

[0206] S6 specifically includes the following sub-steps:

[0207] S610. User Interaction Data Collection and Association: During the interaction process in the contextual visualization view, the system records the user's interaction behavior log in real time. Interaction events include, but are not limited to:

[0208] Click; Hover; SplitView; Selection; Zoom / Pan.

[0209] Each interaction event It includes the following fields: EventType: interaction type; SpatialRef: corresponding spatial unit IDi; TemporalRef: corresponding time slice IDt; EventValue: interaction intensity or dwell time; TimeStamp: event occurrence timestamp.

[0210] All interaction events are stored in the interaction log queue. and in the spatiotemporal joint index The items are associated one-to-one, providing a reference for behavioral weights for subsequent model calibration.

[0211] S620, External Observation Data Reception and Alignment: The system periodically receives posterior observation data from external systems. This includes actual transaction prices, construction progress, and the effectiveness of policy implementation.

[0212] After receiving the observation data, data cleaning and temporal-spatial alignment operations are performed to align the external observations with... according to correspond:

[0213]

[0214] in This represents the set of matching relationships between observation results and model predictions. If observation data is missing or the spatiotemporal resolution is inconsistent, the system uses interpolation or nearest neighbor rules to automatically fill in the missing data and marks the "source of the missing data" in the index.

[0215] S630, Error Index Calculation and Calibration Trigger Decision: For each spatial cell i and time slice t, calculate the absolute error between the predicted value and the observed value:

[0216]

[0217] in These are measured observations collected from an external system. These are the scenario predictions of the model under the same space and time. This represents the prediction bias of the spatiotemporal unit.

[0218] Calculate the overall average error:

[0219]

[0220] Where N is the total number of spatial units and T is the number of time slices; when At that time, online calibration of the causal model and visual encoding mapping rules is triggered; The preset error threshold can be dynamically adjusted based on historical fluctuation ranges.

[0221] S640, Online Calibration of Structured Causal Model and Mapping Rules: When calibration is triggered, the system adjusts the structured causal model parameters according to the error distribution. With differential-visual coding mapping threshold Adjustments need to be made. Calibration consists of two parts:

[0222] Model parameter tuning: For variable nodes with high error Recalculate the conditional probability parameters:

[0223]

[0224] in For learning rate, For the error gradient, update the Theta field of the corresponding causal query index.

[0225] Mapping threshold adjustment: Adjust the difference significance threshold according to the error frequency distribution. :

[0226]

[0227] in For adjustment coefficients, The relative proportion of high-error regions is determined; the new threshold is written to the mapping rule index; during the calibration process, all modifications are first written to a temporary buffer, and then written to the formal index after passing the consistency check, to ensure that the front-end rendering service is not interrupted.

[0228] S650, Index Version Control and Rollback Mechanism: After calibration is complete, the system generates a new index version number Ver_new and writes the updated content to:

[0229] Causal Query Index (Updated) );

[0230] Mapping rule index (updated) );

[0231] Spatiotemporal joint index (updating difference significance flags and status flags).

[0232] The system retains the previous version, Ver_old, as a security snapshot. If it detects [something] in a subsequent cycle... Abnormal fluctuations (exceeding) If the value is 0, it will automatically roll back.

[0233]

[0234] Ensure the system remains stable and available at all times; at the same time, write rollback events and error distributions to the calibration log for subsequent model optimization and strategy adjustment.

[0235] Example 2:Figure 2 As shown, this embodiment provides a real estate data visualization system based on GIS technology, including:

[0236] The data modeling and indexing module is used to collect original real estate data from multiple sources, such as registration, transaction, planning approval, judicial, tax, sensor terminals and remote sensing images. The module cleans the data, unifies the format and aligns the entities, establishes a set of spatial units and its event identification system, constructs a spatiotemporal joint index based on spatial unit ID and time slice ID, and extracts the feature set and target attribute set for modeling.

[0237] The causal model construction module is used to determine the structural relationship between causal variables based on the feature set and the target attribute set, establish a structured causal model through structural learning algorithm and parameter training, and generate a causal query index to support the calculation of the target attribute distribution under different intervention conditions without changing the data sampling distribution.

[0238] The scenario intervention and counterfactual inference module is used to receive scenario parameter sets and perform legality verification, parse the scenario parameter sets into intervention instructions, perform do-intervention operations on the structured causal model, calculate scenario outcome data under intervention conditions, calculate baseline outcome data under non-intervention conditions within the same time window and spatial range, and perform spatiotemporal alignment of the two sets of results to generate a differential data pairing set.

[0239] The differential mapping and visual coding generation module is used to perform differential operations on the scenario result data and baseline result data in the spatial unit and time slice dimensions to generate a differential tensor, and generate a corresponding visual coding parameter set according to the preset differential-to-visual coding mapping rules, and write the differential tensor and the visual coding parameter set into the spatiotemporal joint index.

[0240] The visualization view generation and linkage display module is used by the visualization engine to generate a scene visualization view on the GIS map according to the set of visual encoding parameters. It performs layer expansion and detail display on spatial units with significant differences within the user's view window, and realizes linkage comparison and incremental rendering of the current view and the scene view on the same interface.

[0241] The feedback acquisition and online calibration module is used to collect user interaction behavior and posterior observation data from external systems, calculate the error index between scenario result data and observation values, and trigger online calibration of the structured causal model and visual encoding mapping rules when the error exceeds a preset threshold. It also updates the parameter versions of the causal query index and spatiotemporal index and supports version rollback, thereby realizing adaptive optimization of the model and dynamic correction of visualization effects.

[0242] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0243] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0244] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for visualizing real estate data based on GIS technology, characterized in that, Includes the following steps: S1. Clean, deduplicate, unify the format and align the original real estate data from multiple sources with entities, establish a real estate spatiotemporal database, construct a unified identification system and spatiotemporal index for spatial unit sets and their events, and form a causal modeling feature set and target attribute set. S2. Determine the structural relationships between causal variables, train and determine the structured causal model and model parameters, establish a causal query index for performing causal queries in the real estate spatiotemporal database, and provide the distribution of target attributes under different intervention conditions without changing the data sampling distribution. S3. Convert the scenario parameter set into intervention instructions that can be executed on the structured causal model, calculate scenario outcome data based on the intervention instructions, and calculate baseline outcome data under the condition of no intervention within the same time window; S4. Perform differential operations on the scenario result data and baseline result data in the spatial unit and time slice dimensions to obtain a differential tensor. Based on the preset differential-to-visual coding mapping rules, convert the differential tensor into a visual coding parameter set and write the differential tensor and the visual coding parameter set into the spatiotemporal index. S5. The visualization engine generates a scenario visualization view on the GIS map according to the set of visual coding parameters. Within the user's view range, it performs hierarchical expansion and detailed association display of spatially significant units. It also provides a linkage comparison between the current view and the scenario view on the same interface, expressing the spatial changes of real estate under intervention conditions through differentiation.

2. The method for visualizing real estate data based on GIS technology according to claim 1, characterized in that, It also includes S6, which records user interaction behavior in the scenario visualization view and posterior observation data from external systems, calculates the error index between scenario result data and actual observations, performs online calibration of structured causal models and differential-to-visual coding mapping rules based on the error index, and writes back the calibrated model parameters and mapping rules to the causal query index and spatiotemporal index for subsequent scenario visualization display.

3. The method for visualizing real estate data based on GIS technology according to claim 1, characterized in that, S1 specifically includes: Acquire original data on real estate registration, transactions, planning approvals, judicial matters, taxation, sensor terminals, and remote sensing images; identify the source and collection time of the data sources to form a set of data to be processed. The data set to be processed is cleaned, deduplicated, and formatted. The name, code, coordinates, timestamp, and ownership fields are normalized and mapped to form a set of data entries with a consistent structure. A unified identification system is established based on land parcels, individual buildings and interior units to complete entity alignment and relationship alignment, and to map historical change records into a time-oriented event sequence. A spatiotemporal database of real estate is constructed under a geographic coordinate system. Spatial and temporal indexes are established for spatial unit sets and event sequences, respectively, and a spatiotemporal composite index is generated to support spatial queries over time. Extract the feature set and target attribute set for causal modeling from a data entry set with consistent structure. Associate the extraction results with the spatiotemporal joint index and store them to obtain the feature view for subsequent modeling.

4. The method for visualizing real estate data based on GIS technology according to claim 1, characterized in that, S2 specifically includes: Based on the feature set and the target attribute set, and combined with real estate business constraints, the causal variable set is determined, forming a causal candidate relationship set; Structural learning and constraint verification are performed on the causal candidate relation set to determine the directed acyclic graph structure of the structured causal model and to record the parent-child relationships that are accessible for intervention. Historical samples from a spatiotemporal database are used to learn the parameters of a structured causal model, generating a set of model parameters, and a stable range of parameter values ​​is determined through cross-validation. A causal query index is established in the spatiotemporal database, and the directed acyclic graph structure, model parameter set and variable encoding rules are persisted in the form of index entries to support subsequent causal queries and intervention calculations. Consistency checks are performed on the structured causal model, and edges or parameters that do not meet statistical significance are marked as low-confidence entries. A fallback strategy is set in the causal query index to ensure query stability.

5. The method for visualizing real estate data based on GIS technology according to claim 1, characterized in that, S3 specifically includes: Receive a scenario parameter set, which includes intervention descriptions of changes in planning conditions, addition of public facilities, or adjustment of restrictive policies, and complete parameter legality verification and unit standardization; Based on the scenario parameter set and causal query index, the intervention description is converted into intervention instructions that can be executed on the structured causal model, and the intervention variables and their value methods are clearly defined; Within a unified time window, scenario outcome data are calculated based on a structured causal model and intervention instructions to obtain the target attribute distribution and spatiotemporal location results under the intervention conditions. Under conditions consistent with the unified time window, baseline result data are calculated without intervention to obtain the target attribute distribution and spatiotemporal positioning results under the current conditions; Align the scenario results data with the baseline results data according to spatial units and time slices to generate a set of data pairs to be differentiated, and record the corresponding relationship in the spatiotemporal joint index.

6. The method for visualizing real estate data based on GIS technology according to claim 1, characterized in that, S4 specifically includes: Perform difference operations on the paired data set, calculate the difference tensors of the target attributes by spatial unit and time slice, and obtain the difference results used to represent the magnitude and direction of change; According to the preset differential-to-visual encoding mapping rules, the differential tensor is mapped to a set of visual encoding parameters, which includes layer selection, symbolization style, layer expansion level and annotation priority. Threshold pruning and conflict resolution are performed on the visual encoding parameter set to ensure that layer on / off, symbolization style and hierarchical expansion level do not contradict each other within the same view range; Write the difference tensor and the set of visually encoded parameters into the spatiotemporal joint index, establish index entries from spatial unit identifiers to the set of visually encoded parameters, and form a parameter view that can be directly consumed by the visualization engine. Set significance and update flags for the difference results, and record the generation time and data source for subsequent view refresh and incremental update.

7. The method for visualizing real estate data based on GIS technology according to claim 1, characterized in that, S5 specifically includes: The visualization engine initiates a query to the spatiotemporal joint index based on the user's view range to obtain the differential tensor and visual encoding parameter set corresponding to the view range; The visualization engine generates scene visualization views on the GIS map according to the set of visual coding parameters, and enables hierarchical expansion and detailed association display for spatial units with significant differences; The current status view and the scenario view are generated and compared on the same interface to maintain spatial positioning consistency, and the spatial changes of real estate under intervention conditions are expressed by difference. Incremental rendering and caching are performed on layers, labels, and hierarchical expansion within the user's view range to ensure continuity and stability during view scrolling and zooming; The generated results of the scenario visualization view and the query parameters are recorded as view session entries for subsequent feedback collection and online calibration.

8. The method for visualizing real estate data based on GIS technology according to claim 2, characterized in that, S6 specifically includes: Record user interactions in the context visualization view, including clicks, pauses, comparisons, and back actions, and associate them with the corresponding spatial unit identifiers; Receive posterior observation data from external systems, verify and align the posterior observation data in the same spatial unit and time slice dimension to form an aligned data set for evaluation; Error indices between scenario results data and actual observations are calculated based on the aligned dataset. These error indices include deviation and matching metrics, and an error assessment report is generated. Based on the error assessment report, the structured causal model and the differential-to-visual coding mapping rules are calibrated online, and the model parameter set and mapping threshold are updated to ensure that subsequent calculations and displays are consistent with actual observations. The calibrated model parameter set, mapping threshold, and related tags are written back to the causal query index and spatiotemporal joint index, triggering version updates of the corresponding index entries for subsequent scenario visualization.

9. A real estate data visualization system based on GIS technology, employing the real estate data visualization method based on GIS technology as described in any one of claims 1-8, characterized in that, include: The data modeling and indexing module is used to collect original real estate data from registration, transactions, planning approval, judicial, tax, sensor terminals and remote sensing images, clean the data, unify the format and align the entities, and establish a set of spatial units and its event identification system. The causal model construction module is used to determine the structural relationship between causal variables based on the feature set and the target attribute set, establish a structured causal model through structural learning algorithms and parameter training, and generate a causal query index. The scenario intervention and counterfactual inference module is used to receive a scenario parameter set and perform legality verification, parse the scenario parameter set into intervention instructions, perform do-intervention operations on the structured causal model, and calculate scenario outcome data under the intervention conditions. The differential mapping and visual encoding generation module is used to perform differential operations on the scenario result data and baseline result data in the spatial unit and time slice dimensions, generate differential tensors, and generate corresponding visual encoding parameter sets according to preset differential-to-visual encoding mapping rules. The visualization view generation and linkage display module is used by the visualization engine to generate a scene visualization view on the GIS map according to the set of visual encoding parameters, and to perform layer expansion and detail display on spatial units with significant differences within the user's view window. The feedback acquisition and online calibration module is used to collect user interaction behavior and posterior observation data from external systems, calculate the error index between scenario result data and observation values, and trigger online calibration of the structured causal model and visual coding mapping rules when the error exceeds a preset threshold.

Images