Land market analysis method and system based on large model

Abstract

The invention provides a land market analysis method and system based on a large model. The method comprises the steps of constructing a neural network model, inputting land market related data, performing hidden layer feature extraction and ReLU activation function processing, and outputting land price prediction and market trend classification. The model is trained by using historical data, parameters are optimized by using a back propagation and cross entropy loss function and a gradient descent method, the learning rate is 0.001-0.1, and iteration is performed until the model is converged. The land price and the market trend can be effectively predicted, and the accuracy and the efficiency of land market analysis are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of deep learning and big data analysis, and more specifically, to a land market analysis method and system based on a large model. Background Art

[0002] Existing land market analysis technologies mainly rely on traditional statistical methods and machine learning techniques, which have certain limitations when processing large-scale, multi-source data sets. Traditional methods often require a lot of manual participation in land resource monitoring, land use planning and decision support, land market analysis and risk management, resulting in low efficiency and high costs. In addition, traditional technologies rely on expert experience in feature extraction and pattern recognition, which may not fully capture the complexity and dynamic changes of land market data. Deep learning technology, especially convolutional neural networks (CNNs), has shown significant advantages in the automatic classification of land cover in remote sensing images. It can automatically learn and extract features from large-scale data, and improve the automation and accuracy of classification. However, these technologies have not been widely used in land market analysis, especially in the automated analysis and prediction of land market trends.

[0003] In the process of implementing the embodiments of the present invention, the inventors found that there are at least the following problems or defects in the prior art: the prior art has limited processing capabilities for big data in land market analysis and cannot fully utilize multi-source data such as land transaction records and price changes; it relies on manual experience in feature extraction and trend prediction, and lacks automation and intelligence; and it is insufficient in the accuracy and real-time nature of land market trend prediction. Summary of the invention

[0004] The present invention provides a land market analysis method and system based on a large model.

[0005] In a first aspect of the present invention, a land market analysis method based on a large model is provided, comprising: A large land market analysis model based on deep learning is constructed. The model adopts a neural network architecture and includes an input layer, multiple hidden layers, and an output layer. The input layer is used to receive land market related data, including land transaction price, land area, land use, geographical location information, surrounding infrastructure information, and land transfer time information. The hidden layer extracts and transforms the input data, and uses an activation function to perform nonlinear transformation on the data. The activation function uses the ReLU function. The model is trained using historical land market data. The back propagation algorithm is used in the training process. The loss function between the predicted results and the actual results is calculated, and the model parameters are optimized based on the loss function. The loss function uses the cross entropy loss function. The gradient descent algorithm is used to adjust the model parameters during the training process and set the learning rate. , iterate the training until the model converges. The convergence condition is that the change in the loss function value is less than the preset threshold or the maximum number of iterations is reached .

[0006] Furthermore, the method comprises a data preprocessing step: Data cleaning: remove outliers and duplicate data by setting data value range rules and data logic verification rules; for land transaction prices, if the price exceeds the average historical transaction price of the same type of land in the area, If the value is greater than times the standard deviation, it is considered an outlier and will be removed. Data standardization uses the Z-score standardization method to convert data of different dimensions into the same dimension.

[0007] Furthermore, when building a large land market analysis model based on deep learning: 1. Number of hidden layers in the neural network Determined according to the complexity of the data; 2. Number of neurons in each hidden layer Calculated using empirical formula; 3. When initializing the model, the model parameters are randomly initialized, and the parameter value range is set according to the characteristics of the neuron connection weights.

[0008] Furthermore, when training the model using historical land market data: The historical land market data is The ratio is divided into training set and validation set, where represents the proportion of training set, Indicates the proportion of the validation set; During the training process, each training After batches, the model is evaluated using the validation set; Evaluation indicators include accuracy in is the number of true positives, is the number of true negatives, is the number of false positives, is the number of false negatives; recall value in ; Adjust model hyperparameters based on evaluation indicators, including learning rate , the number of hidden layers , the number of neurons in each hidden layer .

[0009] Furthermore, after completing the model training, the trained model is used to analyze and predict the land market data, including: The land market data to be analyzed is input into the trained model. The data must first be converted into a format acceptable to the model through data preprocessing steps; The model performs forward propagation calculations on the input data, and obtains the prediction results through the input layer, hidden layer, and output layer in sequence; For the land price forecast, if the error between the forecast result and the actual value exceeds the preset error range, the data will be marked as suspicious data and its relevant information will be recorded for subsequent manual verification. The preset error range is set according to the fluctuation characteristics of land market prices, for example, the error rate does not exceed .

[0010] Furthermore, the model prediction results are visualized, showing the changing trend of land price prediction values ​​over time, the comparison of land prices in different regions, the distribution of land market trend prediction results, etc. in the form of charts. The chart types include line charts, bar charts and pie charts. During the display process, different time granularities are selected according to user needs.

[0011] Further, a land market analysis report is generated based on the model prediction results, the report includes a summary of the current land market status, the summary content covers the overall price level of the current land market, the distribution of land prices for different uses, and the activity of the regional land market; Detailed analysis of forecast results, including detailed interpretation of land price forecast values ​​and analysis of reasons for land market trend forecast results; Investment advice based on the analysis results provides land investors with investment directions, risk warnings, etc. according to land price trends and market trends. Investment advice is classified according to different investment scales and risk preferences.

[0012] Furthermore, the method further comprises: Regularly collect new land market data, process the new data according to the data preprocessing steps, and use the new data to incrementally train the trained model to keep the model adaptable to changes in the land market; The incremental training process uses the same algorithm and hyperparameter settings as the initial training, or fine-tunes the hyperparameters based on the characteristics of the new data; The model update cycle is set according to the frequency of land market data updates, such as updating the model every quarter or every six months.

[0013] Furthermore, during the model training and use process, the model training log is recorded, including information such as training time, training data source, changes in loss function value during training, model evaluation index results, etc.; the input data, prediction results, prediction time, etc. during the model prediction process are recorded; in order to facilitate traceability analysis and troubleshooting of model performance, and provide data support for model optimization.

[0014] In a second aspect of the present invention, a land market analysis system based on a large model is provided, comprising: Data collection module, used to collect data related to the land market; A data preprocessing module is used to preprocess the collected data; Model building module, used to build a large land market analysis model based on deep learning; Model training module, used to train the constructed model using historical land market data; The analysis and prediction module is used to analyze and predict land market data using the trained model; The result display module is used to visualize the model prediction results; Report generation module, used to generate land market analysis reports based on model prediction results; Model update module, used to regularly collect new data and update the model; The logging module is used to record relevant information during model training and prediction.

[0015] The above embodiments of the present invention have at least the following beneficial effects: the large land market analysis model constructed using deep learning technology can improve the accuracy and efficiency of land market analysis. Through the neural network architecture, the model can automatically extract features from a large amount of land market data and perform nonlinear transformations, thereby more accurately predicting land prices and market trends. This method can reduce dependence on expert experience and make the analysis process more objective and scientific.

[0016] In addition, this method uses historical land market data to train the model and uses the back propagation algorithm to optimize the model parameters, so that the model can adapt to market changes and continuously improve the prediction accuracy. The iterative training of the model until convergence can ensure the stability and reliability of the model in practical applications. In this way, more accurate market analysis results can be provided to land investors and decision makers, helping them make more reasonable investment and planning decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood by reading the following detailed description with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, in which: Figure 1 A schematic diagram of a flow chart of a land market analysis method based on a large model provided by an embodiment of the present invention; Figure 2 A schematic diagram of the structure of a land market analysis system based on a large model provided by an embodiment of the present invention; Figure 3 The schematic diagram schematically shows the structure of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0018] The principles and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and implement the present invention, and are not intended to limit the scope of the present invention in any way. On the contrary, these embodiments are provided to make the present invention more thorough and complete, and to fully convey the scope of the present invention to those skilled in the art.

[0019] Those skilled in the art know that the embodiments of the present invention can be implemented as a system, device, apparatus, method or computer program product. Therefore, the present invention can be specifically implemented in the following forms, namely: complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0020] It should be noted that any number of elements in the drawings is for illustration rather than limitation, and any naming is only for distinction and does not have any limiting meaning.

[0021] Reference below Figure 1 , Figure 1 The following is a flow chart of a land market analysis method based on a large model provided by an embodiment of the present invention. Figure 1 As shown, a land market analysis method 100 based on a large model includes: Step 101, construct a large land market analysis model based on deep learning, which adopts a neural network architecture and includes an input layer, multiple hidden layers and an output layer; the input layer is used to receive land market related data, and the land market related data includes land transaction price, land area, land use, geographical location information, surrounding infrastructure information and land transfer time information; the hidden layer extracts and transforms the input data, and uses an activation function to perform nonlinear transformation on the data, and the activation function uses the ReLU function: in represents the activation function, Represents the value input to the activation function; the output layer is used to output the land market analysis results, including the land price forecast value and the land market trend classification results, where the land market trend classification results include rising trend, stable trend and falling trend; Step 102, the model is trained using historical land market data. The training process uses a back propagation algorithm to calculate the loss function between the predicted results and the actual results, and optimize the model parameters based on the loss function. The loss function uses a cross entropy loss function: in, represents the number of samples, represents the true label, Represents the model prediction value. During the training process, the gradient descent algorithm is used to adjust the model parameters and set the learning rate , whose value range is , iterate the training until the model converges. The convergence condition is that the change in the loss function value is less than the preset threshold or the maximum number of iterations is reached , The value range is .

[0022] It should be noted that this implementation is based on building a large land market analysis model based on deep learning. The model adopts a neural network architecture, including an input layer, multiple hidden layers, and an output layer. Here, the neural network architecture refers to a computing model that imitates the way the human brain processes information. It receives data through the input layer, extracts and converts features in the hidden layer, and finally gives the analysis results in the output layer. The land market-related data received by the input layer include land transaction prices, land areas, etc. These data are the basic information for analyzing the land market.

[0023] Specifically, the input layer of the model is designed to receive the above-mentioned land market-related data, which go through preprocessing steps such as data cleaning and standardization to improve the efficiency and accuracy of model training.

[0024] More specifically, the number of hidden layers and the number of neurons in each hidden layer are set according to the complexity of the data and the performance requirements of the model. For example, the number of hidden layers may be between 3 and 10, and the number of neurons is calculated using a specific empirical formula. The activation function uses the ReLU function, which is a commonly used activation function because it can speed up the training of neural networks and reduce the amount of calculation.

[0025] Preferably, the model parameters are initialized using a random initialization method, and the parameter value range is set according to the characteristics of the neuron connection weights. For example, the weight value range may be between -6 and 6, while the bias value range is between -1 and 1. This parameter initialization method helps to break the symmetry, increase the diversity of the model, and thus improve the generalization ability of the model.

[0026] Furthermore, the gradient descent algorithm and learning rate setting used in the training process are also key factors in whether the model can converge successfully. The learning rate ranges from 0.001 to 0.1, and the learning rate within this range can balance the speed and stability of model training.

[0027] In some embodiments, the method comprises a data preprocessing step: Data cleaning: remove outliers and duplicate data by setting data value range rules and data logic verification rules; for land transaction prices, if the price exceeds the average historical transaction price of the same type of land in the area, times the standard deviation, it is considered an outlier. The value range is , and remove it; Data standardization uses the Z-score standardization method to convert data of different dimensions into the same dimension. The calculation formula is: in is the original data, is the data mean, is the standard deviation of the data for subsequent analysis.

[0028] It should be noted that this implementation involves a data preprocessing step, which is the process of cleaning and standardizing data before deep learning model training. Data cleaning refers to removing outliers and duplicate data in the data by setting data value range rules and data logic verification rules to ensure data quality. Data standardization converts data of different dimensions into the same dimension to facilitate subsequent analysis. Here, outliers refer to data points that are beyond the normal range, such as land transaction prices that exceed a certain multiple standard deviation of the historical transaction price average.

[0029] Specifically, during the data cleaning process, if the land transaction price exceeds the average historical transaction price of the same type of land in the region by 2 to 5 times the standard deviation, it is considered an outlier and removed. The standard deviation here is a statistical concept used to measure the degree of dispersion of a set of values. Data standardization uses the Z-score standardization method, and the calculation formula is: in is the original data, is the data mean, is the standard deviation of the data.

[0030] Preferably, the data preprocessing step also includes processing of missing values, such as filling missing data by interpolation or using methods such as mean and median. In addition, the data can be normalized according to the needs of the model, such as scaling the data to the range of [0, 1] or [-1, 1] to accelerate the convergence of the model and improve the performance of the model. These steps can further improve the quality of the data and provide more accurate and reliable input for model training.

[0031] In some embodiments, when building a large land market analysis model based on deep learning: 1. Number of hidden layers in the neural network Determined according to the complexity of the data. The value range is ; 2. Number of neurons in each hidden layer Using empirical formula Calculated, among which is the number of neurons in the input layer, is the number of neurons in the output layer; 3. When the model is initialized, the model parameters are randomly initialized, and the parameter value range is set according to the characteristics of the neuron connection weights, such as weight The value range is , bias The value range is .

[0032] It should be noted that when building a large land market analysis model based on deep learning, it involves determining the number of hidden layers and neurons in the neural network, as well as the initialization method of the model parameters. Here, the hidden layer refers to the layer between the input layer and the output layer of the neural network, which is responsible for extracting and converting the features of the input data. The number of neurons refers to the number of processing units in each layer. The setting of these parameters has a direct impact on the performance of the model.

[0033] Specifically, the number of hidden layers can be determined according to the complexity of the data, and its value range is 3 to 10 layers. The number of neurons in each hidden layer can be calculated by an empirical formula: in is the number of neurons in the input layer, is the number of neurons in the output layer.

[0034] More specifically, when the model is initialized, model parameters such as weights and biases are randomly initialized, and the parameter value range is set according to the characteristics of the neuron connection weights. For example, the weight value range is arrive , the bias value range is from -1 to 1.

[0035] Preferably, in addition to the above-mentioned parameter settings and initialization methods, other factors can be considered to further refine the model building process. For example, different activation functions, such as sigmoid or tanh, can be selected according to specific application scenarios and data characteristics to adapt to different nonlinear problems.

[0036] Furthermore, different parameter initialization strategies can be used, such as He initialization or Xavier initialization, which can vary according to the activation function to improve the efficiency and effect of model training. Through these refined operation steps or alternatives, the performance and adaptability of the model can be further improved.

[0037] In some embodiments, when training a model using historical land market data: The historical land market data is The ratio is divided into training set and validation set, where represents the proportion of training set, represents the proportion of the validation set, The value range is ; During the training process, each training After batches, the model is evaluated using the validation set. The value range is ; Evaluation indicators include accuracy in is the number of true positives, is the number of true negatives, is the number of false positives, is the number of false negatives; recall value in ; Adjust model hyperparameters based on evaluation indicators, including learning rate , the number of hidden layers , the number of neurons in each hidden layer .

[0038] It should be noted that this implementation involves the process of training the model using historical land market data, which includes key steps such as data set division, model evaluation, and hyperparameter adjustment. Here, the training set and validation set refer to the subsets of historical data used for model training and evaluation, respectively. The training set is used in the learning process of the model, while the validation set is used to evaluate the performance of the model during the training process to prevent the model from overfitting.

[0039] Specifically, the historical land market data was divided into a training set and a validation set in a ratio of 7:3, with the training set accounting for 70% and the validation set accounting for 30%. During the training process, the validation set was used to evaluate the model after every 5 batches of training.

[0040] More specifically, the evaluation metrics include precision and recall. Precision refers to the ratio of the number of samples correctly predicted by the model to the total number of samples, and recall refers to the ratio of the number of samples successfully predicted by the model to the number of samples that are actually positive. These evaluation metrics help adjust the model's hyperparameters, such as the learning rate, the number of hidden layers, and the number of neurons in each hidden layer, to achieve optimal model performance.

[0041] Preferably, the model evaluation process can also include the F1 score as an evaluation indicator. The F1 score is the harmonic mean of the precision and recall rate. It comprehensively considers the accuracy and recall rate of the model and provides a more comprehensive model performance measurement.

[0042] Furthermore, hyperparameter tuning can be done through automated techniques such as grid search or random search, which can find the optimal parameter combination within the preset hyperparameter space, thereby improving the efficiency and effectiveness of model training. Through these detailed operation steps or alternative solutions, the quality and efficiency of model training can be further improved.

[0043] In some embodiments, after the model training is completed, the trained model is used to analyze and predict the land market data, including: The land market data to be analyzed is input into the trained model. The data must first be converted into a format acceptable to the model through data preprocessing steps; The model performs forward propagation calculations on the input data, and obtains the prediction results through the input layer, hidden layer, and output layer in sequence; For the land price forecast, if the error between the forecast result and the actual value exceeds the preset error range, the data will be marked as suspicious data and its relevant information will be recorded for subsequent manual verification. The preset error range is set according to the fluctuation characteristics of land market prices, for example, the error rate does not exceed , The value range is .

[0044] It should be noted that this embodiment describes in detail the process of using a trained model to analyze and predict land market data. The forward propagation calculation involved here refers to the process in which the model obtains the prediction result based on the input data through each layer of the neural network until the output layer. The preset error range refers to the maximum error ratio allowed when predicting land prices. For example, if the preset error range is 5%, it means that the difference between the predicted price and the actual price should not exceed 5% of the actual price.

[0045] Specifically, the land market data to be analyzed must first undergo data preprocessing steps, including data cleaning and standardization, to convert it into a format acceptable to the model. The model performs forward propagation calculations on the input data and obtains prediction results through the input layer, hidden layer, and output layer in sequence.

[0046] More specifically, for the land price forecast, if the error between the forecast result and the actual value exceeds the preset error range, the data will be marked as suspicious data and its relevant information will be recorded for subsequent manual verification. The preset error range is set according to the fluctuation characteristics of land market prices, for example, the error rate does not exceed 5% to 20%.

[0047] Preferably, the processing of the model prediction results can be further refined into the following steps: First, the error between the model prediction results and the actual value can be determined by calculating the absolute difference or relative difference between the predicted value and the actual value. Second, for the prediction results that exceed the preset error range, an automated marking system can be used to classify these data separately for subsequent manual review and analysis.

[0048] Furthermore, an alarm mechanism can be set up so that when the prediction error exceeds a certain threshold, the system will automatically notify the analyst so that timely measures can be taken. These detailed steps or alternatives can improve the accuracy and reliability of model predictions and ensure the quality and efficiency of land market analysis.

[0049] In some embodiments, the model prediction results are visualized in the form of charts showing the changing trends of land price prediction values ​​over time, comparisons of land prices in different regions, and the distribution of land market trend prediction results. Chart types include line charts, bar charts, and pie charts. During the display process, different time granularities are selected for display according to user needs.

[0050] It should be noted that this embodiment describes the process of visually displaying the model prediction results, including displaying the trend of land price prediction values ​​over time, the comparison of land prices in different regions, the distribution of land market trend prediction results, etc. in the form of charts. Here, visual display refers to the intuitive presentation of data and analysis results through graphical means, so that users can understand and analyze them more easily. Common chart types include line charts, bar charts, and pie charts, which are suitable for displaying time series data, comparative data, and percentage data, respectively.

[0051] Specifically, visualization can be achieved through the following steps: First, select the appropriate chart type for display based on the land price forecast value and market trend classification results output by the model. For example, use a line chart to show the trend of land prices over time, a bar chart to compare land prices in different regions, and a pie chart to show the distribution of market trend forecast results. Second, select different time granularities for display according to user needs, such as day, month, quarter, or year, to meet the analysis needs of different users. In addition, interactive functions can be set to allow users to filter specific data and time periods as needed, and switch between different chart types for comparative analysis.

[0052] Preferably, the visualization system can be further refined into the following aspects: First, the system can provide customized color and style options so that users can adjust the appearance of the chart according to their personal preferences. Second, the system can integrate more chart types, such as scatter plots, area charts, and heat maps, to provide a richer way of displaying data.

[0053] Furthermore, the system can also provide data export functions, allowing users to export charts and data into common file formats, such as CSV, Excel or PDF, to facilitate further analysis and report making. Through these detailed operation steps or alternative solutions, the user experience can be improved, making the land market analysis results more intuitive and easy to understand.

[0054] In some embodiments, a land market analysis report is generated based on the model prediction results, the report including a summary of the current land market status, the summary covering the overall price level of the current land market, the distribution of land prices for different uses, and the activity of the regional land market; Detailed analysis of forecast results, including detailed interpretation of land price forecast values ​​and analysis of reasons for land market trend forecast results; Investment advice based on the analysis results provides land investors with investment directions, risk warnings, etc. according to land price trends and market trends. Investment advice is classified according to different investment scales and risk preferences.

[0055] It should be noted that this embodiment involves the process of generating a land market analysis report based on the model prediction results. Here, the land market analysis report refers to a detailed document that summarizes the current status of the land market, a detailed analysis of the prediction results, and investment recommendations based on the analysis results.

[0056] Specifically, the land market analysis report first includes a summary of the current status of the land market, which covers the overall price level of the current land market, the distribution of land prices for different uses, and the activity of the regional land market. Secondly, the report contains a detailed interpretation of the land price forecast and an analysis of the reasons for the land market trend forecast results, which helps readers deeply understand the logic behind the forecast and possible influencing factors. Finally, the report provides investment recommendations based on the analysis results. These recommendations provide investment directions and risk warnings for land investors based on land price trends and market trends, and are classified according to different investment scales and risk preferences.

[0057] Preferably, the land market analysis report can be further refined into the following aspects: First, the report can include more charts and visualization elements to intuitively display the data analysis results and enhance the readability and persuasiveness of the report. Second, the report can provide more case studies or historical data analysis to demonstrate the accuracy and practicality of the model prediction.

[0058] Furthermore, the report can also include a special section to discuss the possible impact of external factors such as policy changes and economic trends on the land market, thereby providing a more comprehensive market analysis. Through these detailed operational steps or alternatives, the professionalism and practicality of the report can be improved, and better services can be provided to land investors and decision makers.

[0059] In some embodiments, the method further comprises: Regularly collect new land market data, process the new data according to the data preprocessing steps, and use the new data to incrementally train the trained model to keep the model adaptable to changes in the land market; The incremental training process uses the same algorithm and hyperparameter settings as the initial training, or fine-tunes the hyperparameters based on the characteristics of the new data; The model update cycle is set according to the frequency of land market data updates, such as updating the model every quarter or every six months.

[0060] It should be noted that this embodiment describes a land market analysis method based on a large model, which includes regularly collecting new land market data and using these data to incrementally train the trained model to maintain the model's adaptability to changes in the land market. Here, incremental training refers to the process of updating and optimizing the model by introducing new data based on the original model, rather than training a completely new model from scratch.

[0061] Specifically, the same algorithm and hyperparameter settings as those for initial training can be used in the incremental training process, or the hyperparameters can be fine-tuned according to the characteristics of the new data. The model update cycle can be set according to the frequency of land market data updates, such as updating the model once a quarter or every six months. In terms of parameter settings, the model can be set to automatically trigger the incremental training process after receiving a certain amount of new data, ensuring that the model can reflect the latest market trends in a timely manner.

[0062] Preferably, the specific steps of incremental training may include the following aspects: first, the newly collected land market data is subjected to the same preprocessing steps as before to ensure data consistency. Second, before incremental training, the current state of the model can be backed up so that it can be restored to the previous state when the incremental training effect is not good.

[0063] Furthermore, an evaluation mechanism can be set up to evaluate whether the performance of the model has improved after each incremental training. If there is no improvement or the improvement is not obvious, the hyperparameters can be further adjusted or the quality of the data can be reconsidered. Through these detailed operation steps or alternative solutions, the adaptability and accuracy of the model can be improved, ensuring the timeliness and effectiveness of the land market analysis results.

[0064] In some embodiments, during model training and use, the model training log is recorded, including information such as training time, training data source, changes in loss function value during training, model evaluation index results, etc.; the input data, prediction results, prediction time, etc. during the model prediction process are recorded; so as to facilitate traceability analysis and troubleshooting of model performance, and provide data support for model optimization.

[0065] It should be noted that this implementation involves recording relevant information during model training and use, so as to facilitate traceability analysis and troubleshooting of model performance, and provide data support for model optimization. Here, model training logs refer to all relevant information recorded during model training, including but not limited to training time, training data source, loss function value changes, and model evaluation indicator results. These logs are crucial for understanding model behavior and guiding future model improvements.

[0066] Specifically, the model training log includes key parameters and results of the training process, such as the loss value, accuracy, recall rate and other indicators of each epoch, as well as the performance of the model on the validation set. These indicators help evaluate the performance of the model at different training stages and guide the adjustment of hyperparameters. At the same time, the log should also record the input data, prediction results and prediction time of the model prediction process, so as to conduct a retrospective analysis and find out the possible problems when the model performs poorly.

[0067] Preferably, the log records during model training and prediction can be further refined into the following aspects: First, an automated log management system can be introduced, which can collect and store data during training and prediction in real time for subsequent analysis. Second, the logs can contain more contextual information, such as model version, training environment configuration, etc., which helps to quickly locate problems after model deployment.

[0068] Furthermore, you can also consider using visualization tools to display log data to make performance trends and potential problems more intuitive. Through these detailed operation steps or alternative solutions, you can improve the efficiency and effectiveness of model management and ensure continuous optimization and performance improvement of the model.

[0069] The above-mentioned embodiments of the present invention have the following beneficial effects: The method described in the present invention can perform more accurate analysis and prediction of the land market through deep learning and big data analysis technology. This method can automatically extract the characteristics of land market-related data, and improve the accuracy of land price prediction and market trend classification through nonlinear transformation of input data. At the same time, by optimizing the model parameters through the back propagation algorithm and the gradient descent method, the convergence of the model during the training process and the reliability of the prediction results can be ensured. In addition, the method can also improve the data quality through the data preprocessing step, provide more accurate input for the model, and further improve the accuracy of the analysis results.

[0070] In addition, the method described in the present invention also includes incremental training and regular updating of the model, which can ensure that the model can adapt to the latest changes in the land market and maintain the timeliness and adaptability of the analysis results. By recording detailed information during model training and prediction, the model performance can be traced and troubleshooted, while providing data support for further optimization of the model. This method can also generate land market analysis reports to provide investment directions and risk warnings for land investors, thereby helping them make more informed investment decisions.

[0071] like Figure 2 As shown, a land market analysis system 200 based on a large model in some embodiments includes: The data collection module 201 is used to collect data related to the land market; the data related to the land market include land transaction price, land area, etc. These data are basic information for analyzing the land market and lay the foundation for subsequent data analysis; The data preprocessing module 202 is used to preprocess the collected data, such as data cleaning and standardization, which aims to improve the efficiency and accuracy of subsequent model training, lay a foundation for subsequent further analysis, and further improve the quality of the data; A model building module 203 is used to build a large land market analysis model based on deep learning; A model training module 204 is used to train the constructed model using historical land market data; The analysis and prediction module 205 is used to analyze and predict the land market data using the trained model; A result display module 206 is used to visualize the model prediction results; A report generation module 207, used to generate a land market analysis report based on the model prediction results; A model updating module 208, for periodically collecting new data and updating the model; The logging module 209 is used to record relevant information during the model training and prediction process.

[0072] It is understandable that the modules and references in the land market analysis system 200 based on the large model are Figure 1 Therefore, the operations, features and beneficial effects described above for the land market analysis method based on a large model are also applicable to the land market analysis system 200 based on a large model and the modules contained therein, and will not be repeated here.

[0073] Reference below Figure 3 , which shows a schematic diagram of a structure 300 of an electronic device suitable for implementing some embodiments of the present invention. The electronic devices in some embodiments of the present invention may include, but are not limited to, mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 3 The terminal device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0074] like Figure 3As shown, the electronic device 300 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 to a random access memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the electronic device 300 are also stored. The processing device 301, the ROM 302, and the RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0075] Typically, the following devices may be connected to the I / O interface 305: an input device 306 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 307 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 308 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 309. The communication device 309 may allow the electronic device 300 to communicate with other devices wirelessly or by wire to exchange data. Figure 3 The electronic device 300 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead. Figure 3 Each block shown in the figure may represent one device, or may represent multiple devices as required.

[0076] Furthermore, the storage medium of the embodiment of the present application stores program instructions that can implement all the above methods, wherein the program instructions can be stored in the above storage medium in the form of a software product, including several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) or a processor to execute all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or terminal devices such as a computer, a server, a mobile phone, and a tablet.

[0077] The above descriptions are only some preferred embodiments of the present invention and an explanation of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present invention is not limited to the technical solutions formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above inventive concept. For example, the above features are replaced with (but not limited to) technical features with similar functions disclosed in the embodiments of the present invention.

Claims

1. A land market analysis method based on a large model, characterized in that: The following steps are involved: A large land market analysis model based on deep learning is constructed. The model adopts a neural network architecture and includes an input layer, multiple hidden layers, and an output layer. The input layer is used to receive land market related data, including land transaction price, land area, land use, geographical location information, surrounding infrastructure information, and land transfer time information. The hidden layer extracts and transforms the input data, and uses an activation function to perform nonlinear transformation on the data. The activation function uses the ReLU function. The model is trained using historical land market data. The back propagation algorithm is used in the training process. The loss function between the predicted results and the actual results is calculated, and the model parameters are optimized based on the loss function. The loss function uses the cross entropy loss function. The gradient descent algorithm is used to adjust the model parameters during the training process and set the learning rate. , iterate the training until the model converges. The convergence condition is that the change in the loss function value is less than the preset threshold or the maximum number of iterations is reached .

2. The method according to claim 1, characterized in that The method comprises the steps of data preprocessing: Data cleaning: remove outliers and duplicate data by setting data value range rules and data logic verification rules; for land transaction prices, if the price exceeds the average historical transaction price of the same type of land in the area, If the value is greater than times the standard deviation, it is considered an outlier and will be removed. Data standardization uses the Z-score standardization method to convert data of different dimensions into the same dimension.

3. The method according to claim 1, characterized in that: When building a large land market analysis model based on deep learning, The number of hidden layers in the neural network Determined according to the complexity of the data; The number of neurons in each hidden layer Calculated using empirical formula; When the model is initialized, the model parameters are initialized randomly, and the parameter value range is set according to the characteristics of the neuron connection weights.

4. The method according to claim 1, characterized in that: When training the model using historical land market data, The historical land market data is The ratio is divided into training set and validation set, where represents the proportion of training set, Indicates the proportion of the validation set; During the training process, each training After batches, the model is evaluated using the validation set; The evaluation indicators include accuracy, the formula is: in, is the number of true positives, is the number of true negatives, is the number of false positives, is the number of false negatives; the recall formula is: The value formula is: in, ; Adjust model hyperparameters based on evaluation indicators, including learning rate , the number of hidden layers , the number of neurons in each hidden layer .

5. The method according to claim 1, characterized in that After completing the model training, use the trained model to analyze and predict the land market data, including: The land market data to be analyzed is input into the trained model, and the data is converted into a format acceptable to the model through data preprocessing steps; The model performs forward propagation calculations on the input data, and obtains the prediction results through the input layer, hidden layer, and output layer in sequence; For the land price forecast, if the error between the forecast result and the actual value exceeds the preset error range, the data will be marked as suspicious data and its relevant information will be recorded for subsequent manual verification. The preset error range is set according to the fluctuation characteristics of land market prices, and the error rate does not exceed .

6. The method according to claim 1, characterized in that The model prediction results are visualized, showing the changing trend of land price prediction values ​​over time, the comparison of land prices in different regions, the distribution of land market trend prediction results, etc. in the form of charts. The chart types include line charts, bar charts and pie charts. During the display process, different time granularities are selected according to user needs.

7. The method according to claim 1, characterized in that Generate a land market analysis report based on the model prediction results, the report includes a summary of the current land market status, including the overall price level of the current land market, the distribution of land prices for different uses, and the activity of the regional land market; Detailed analysis of forecast results, including detailed interpretation of land price forecast values ​​and analysis of reasons for land market trend forecast results; Investment advice based on the analysis results provides land investors with investment directions and risk warnings according to land price trends and market trends. Investment advice is classified according to different investment scales and risk preferences.

8. The method according to claim 1, characterized in that The method further comprises: Regularly collect new land market data, process the new data according to the data preprocessing steps, and use the new data to incrementally train the trained model to keep the model adaptable to changes in the land market; The incremental training process uses the same algorithm and hyperparameter settings as the initial training, or adjusts the hyperparameters based on the characteristics of the new data; The model update cycle is set according to the frequency of land market data updates.

9. The method according to claim 1, characterized in that: During model training and use, record the model training log, including training time, training data source, loss function value changes during training, model evaluation index results, and other information; Record the input data, prediction results, prediction time and other information during the model prediction process; to facilitate traceability analysis and troubleshooting of model performance, and provide data support for model optimization.

10. A land market analysis system based on a large model, characterized in that: include: Data collection module, used to collect data related to the land market; A data preprocessing module is used to preprocess the collected data; Model building module, used to build a large land market analysis model based on deep learning; Model training module, used to train the constructed model using historical land market data; The analysis and prediction module is used to analyze and predict land market data using the trained model; The result display module is used to visualize the model prediction results; Report generation module, used to generate land market analysis reports based on model prediction results; Model update module, used to regularly collect new data and update the model; The logging module is used to record relevant information during model training and prediction.

Images