A data asset evaluation method and system based on multi-level analysis

Abstract

The present invention provides a data asset evaluation method and system based on multi-level analysis, which relates to the technical field of data processing. The method comprises: setting an initial population of a genetic algorithm, each individual represents a data processing parameter combination, wherein the data processing parameter combination is a combination of weight coefficients in a data application score; calculating the fitness of the individual according to the data application score; selecting the corresponding individual to enter the next generation according to the fitness of the individual; performing crossover and mutation operations on the selected individual to generate a new individual, repeating the selection, crossover and mutation operations until a termination condition is met to obtain the corresponding optimized parameters; and calculating a regional index according to the quality analysis score and the final data application score. The present invention can improve processing accuracy.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a data asset evaluation method and system based on multi-level analysis. Background Art

[0002] In the practice of parking data asset processing, some traditional methods have some limitations, which affect the accurate processing and efficient utilization of the value of data assets.

[0003] Specifically, some traditional methods mainly focus on data quality processing, such as data completeness, accuracy and consistency, but ignore the effect and potential of data in practical applications. For example, high-quality data has great value in processing parking space allocation, but this is not fully reflected in some traditional processing systems.

[0004] In addition, some traditional methods have not fully considered the multi-dimensional characteristics of data, including key factors such as frequency of use, coverage, and real-time nature of data. These factors are crucial to the ability of data processing to quickly respond to market changes and support real-time decision-making. Therefore, even if some data is slightly flawed in quality, it can still bring commercial value if it has the characteristics of high-frequency updates, wide coverage, and real-time feedback. However, the lack of traditional processing methods in this regard has led to many valuable data resources being underestimated or ignored. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a data asset evaluation method and system based on multi-level analysis, which can improve processing accuracy.

[0006] In order to solve the above technical problems, the technical solution of the present invention is as follows:

[0007] In a first aspect, a data asset evaluation method based on multi-level analysis is provided, the method comprising:

[0008] Collect parking data, including parking lot location information, parking space usage, parking time, and parking fees;

[0009] Preprocessing the collected parking data to obtain preprocessed parking data; performing quality analysis on the preprocessed parking data based on preset data quality analysis indicators to obtain a quality analysis score;

[0010] Based on preset data analysis indicators, the pre-processed parking data is scored for application to obtain a data application score, wherein the data analysis indicators include the frequency of data use, the coverage of data, and the real-time nature of data;

[0011] The initial population of the genetic algorithm is set, and each individual represents a data processing parameter combination, wherein the data processing parameter combination is a combination of weight coefficients in the data application score; the fitness of the individual is calculated according to the data application score; the corresponding individual is selected to enter the next generation according to the fitness of the individual; the selected individual is crossover and mutation operation is performed to generate a new individual, and the selection, crossover and mutation operation are repeated until the termination condition is met to obtain the corresponding optimized parameters; the final data application score is recalculated using the optimized parameters; the regional index is calculated according to the quality analysis score and the final data application score;

[0012] The parking revenue is calculated based on the regional index, combined with the parking lot location information, parking space usage, parking time and parking fees; the parking space allocation strategy is dynamically processed based on the parking revenue.

[0013] Furthermore, the quality analysis score The calculation formula is:

[0014] ;

[0015] in, Indicates The number of error records for each indicator; Indicates The total number of records for each indicator; Indicates The number of recorded data points for a metric; Indicates The number of data points to be recorded for each indicator; Indicates The number of contradictory records for each indicator; Indicates The total number of records related to the indicator; Indicates The time-effectiveness decay rate of each indicator; Indicates The data delay time of each indicator; and They are The mean and standard deviation of each indicator; is the total number of data quality analysis indicators; Indicates the index value of the indicator; represents the base of natural logarithms; , , , and Represents the weight coefficient.

[0016] Furthermore, data application scores The calculation formula is:

[0017] ;

[0018] in, Indicates the number of The number of times the data was queried. Indicates the total duration of the statistical time period. Indicates the maximum possible number of queries; Indicates the number of covered parking lots, Indicates the total number of parking lots; represents the average data delay time, Indicates the maximum acceptable delay time; Indicates the query count index value; Indicates the total number of queries; , and Represents the weight coefficient.

[0019] Furthermore, the specific calculation formula corresponding to the fitness is:

[0020] ;

[0021] in, Representative The fitness value of individual data is the weight coefficient; Indicates Individuals in the data application evaluation Scores on the dimensions; is the total number of dimensions of data application evaluation; is an index; is a control coefficient; is an index from 1 to m, used to traverse the dimensions of data application evaluation; It is a specific index used to identify an individual in the genetic algorithm.

[0022] Further, crossover and mutation operations are performed on the selected individuals to generate new individuals, and the selection, crossover and mutation operations are repeated until the termination condition is met to obtain the corresponding optimized parameters, including:

[0023] Roulette wheel selection is used to select a portion of individuals from the current population to enter the next generation;

[0024] For the selected individual pairs, a multi-point crossover operation is performed at a certain crossover rate to generate new individuals by exchanging some genes between individuals;

[0025] For the new individuals generated after crossover, bit flipping operations are performed at a certain mutation rate;

[0026] The new individuals generated after multi-point crossover and bit flipping are combined with some individuals in the original population to form a new population. Check whether the termination condition is met. If the termination condition is met, stop the iteration and output the corresponding individuals in the current population as the optimized parameter combination.

[0027] Furthermore, the calculation formula of the regional index is:

[0028] ;

[0029] in, is the regional index; The quality analysis score is The weight coefficient of each indicator; The quality analysis score is The score of each indicator; The quality analysis score is The index coefficient of an indicator; The data application score is The weight coefficient of each dimension; The data application score is The scores of the dimensions; The data application score is The exponential coefficient of the dimension.

[0030] Furthermore, after calculating the regional index based on the quality analysis score and the final data application score, it also includes:

[0031] Extract key features related to parking demand from historical parking data, including timestamps, weather conditions, holiday information, and nearby events;

[0032] The key features are converted into codes, and the neural network model is trained using the historical parking data as a training set. The parameters of the neural network model are optimized through cross-validation and grid search to obtain the trained neural network model.

[0033] Use the trained neural network model to run simulations to predict parking demand in different scenarios to obtain simulation results, including predicted demand and demand fluctuation trends;

[0034] Compare simulation results under different scenarios with actual operation data to obtain analysis results;

[0035] According to the analysis results, the risk warning threshold is set, and the warning is triggered when the simulation results exceed the threshold.

[0036] In the second aspect, a data asset evaluation system based on multi-level analysis includes:

[0037] An acquisition module is used to collect parking data, wherein the parking data includes parking lot location information, parking space usage, parking time, and parking fees;

[0038] The quality analysis module is used for the data quality analysis module, and is used for preprocessing the collected parking data to obtain preprocessed parking data; performing quality analysis on the preprocessed parking data based on preset data quality analysis indicators to obtain a quality analysis score;

[0039] A data application processing module, used to perform application scoring on the pre-processed parking data based on preset data analysis indicators to obtain a data application score, wherein the data analysis indicators include the frequency of data use, the coverage of the data, and the real-time nature of the data;

[0040] The calculation module is used to set the initial population of the genetic algorithm, each individual represents a data processing parameter combination, wherein the data processing parameter combination is a combination of weight coefficients in the data application score; the fitness of the individual is calculated according to the data application score; the corresponding individual is selected to enter the next generation according to the fitness of the individual; the selected individual is crossover and mutation operations are performed to generate new individuals, and the selection, crossover and mutation operations are repeated until the termination condition is met to obtain the corresponding optimized parameters; the final data application score is recalculated using the optimized parameters; the regional index is calculated according to the quality analysis score and the final data application score; the parking revenue is calculated according to the regional index in combination with the parking lot location information, parking space usage, parking time and parking fee; and the parking space allocation strategy is dynamically processed according to the parking revenue.

[0041] According to a third aspect, a computing device includes:

[0042] one or more processors;

[0043] The storage device is used to store one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the method described.

[0044] In a fourth aspect, a computer-readable storage medium stores a program, and when the program is executed by a processor, the method described is implemented.

[0045] The above scheme of the present invention includes at least the following beneficial effects.

[0046] By comprehensively considering the quality of the data and the application effect, this method can more comprehensively evaluate the value of parking data, which not only helps to identify potential problems in the data, but also more accurately reflects the performance of the data in actual applications.

[0047] The genetic algorithm is used to optimize the weight coefficient combination of the data application score, making the data processing more accurate. Through continuous iteration and optimization, this method can find the weight coefficient combination that best suits the current data characteristics, thereby improving the accuracy of data processing.

[0048] Based on the regional index and combined with various parking lot operation data, this method can calculate a more scientific parking revenue, thus providing a strong basis for decision-making for parking lot managers. This helps managers allocate parking resources more reasonably and improve the overall operation efficiency of parking lots.

[0049] By analyzing parking data and calculating parking revenue in real time, this method can dynamically handle the allocation strategy of parking spaces, which can not only effectively alleviate the problem of parking difficulties, but also flexibly adjust the use of parking spaces according to market demand, thereby maximizing the revenue of parking lots. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 It is a flowchart of a data asset evaluation method based on multi-level analysis provided by an embodiment of the present invention.

[0051] Figure 2 It is a schematic diagram of a data asset evaluation system based on multi-level analysis provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0052] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0053] like Figure 1 As shown, an embodiment of the present invention proposes a data asset evaluation method based on multi-level analysis, the method comprising the following steps:

[0054] Step 11, collecting parking data, the parking data including parking lot location information, parking space usage, parking time, and parking fees;

[0055] Step 12: preprocessing the collected parking data to obtain preprocessed parking data; performing quality analysis on the preprocessed parking data based on a preset data quality analysis index to obtain a quality analysis score;

[0056] Step 13, based on preset data analysis indicators, the pre-processed parking data is scored for application to obtain a data application score, wherein the data analysis indicators include the frequency of data use, the coverage of the data, and the real-time nature of the data;

[0057] Step 14, set the initial population of the genetic algorithm, each individual represents a data processing parameter combination, where the data processing parameter combination is the weight coefficient in the data application score , , combination; calculate the fitness of the individual according to the data application score; select the corresponding individual to enter the next generation according to the individual fitness; perform crossover and mutation operations on the selected individuals to generate new individuals, repeat the selection, crossover and mutation operations until the termination condition is met to obtain the corresponding optimized parameters; use the optimized parameters to recalculate the final data application score; calculate the regional index according to the quality analysis score and the final data application score; calculate the parking revenue according to the regional index, combined with the parking lot location information, parking space usage, parking time and parking fee; dynamically process the parking space allocation strategy according to the parking revenue.

[0058] In the embodiment of the present invention, step 11, comprehensive collection of parking data provides a solid foundation for subsequent analysis and evaluation. These data cover key aspects of parking lot operations, such as location information, parking space usage, parking time and fees, thereby ensuring the accuracy and comprehensiveness of the evaluation. Step 12, data preprocessing can clean and organize the original data, eliminate outliers and noise, and improve data quality. Scoring based on preset data quality analysis indicators can quantitatively understand the integrity, accuracy and consistency of the data. Step 13, scoring the data through preset data analysis indicators can deeply understand the potential and value of the data in practical applications. These indicators include frequency of use, coverage and real-time, which together reflect the ability of data in supporting decision-making, optimizing operations, etc. Step 14, using genetic algorithms to optimize the combination of data processing parameters, it is possible to find the weight coefficient that best suits the current data characteristics, thereby improving the accuracy of data processing, combining the quality analysis score and the data application score to calculate the regional index, and providing a scientific decision-making basis for parking lot managers. Further, dynamically processing the parking space allocation strategy based on parking revenue helps to improve the operational efficiency and profitability of the parking lot.

[0059] In the embodiment of the present invention, the above step 11 collects parking data, and the parking data includes parking lot location information, parking space usage, parking time, and parking fees, and may include:

[0060] Determine the source of the data:

[0061] Parking management system, which is the main data source, usually contains key information such as parking space usage, parking time and cost.

[0062] Sensor records: including parking space occupancy sensors, entry / exit gate sensors, etc., used to monitor the status of parking spaces in real time.

[0063] Mobile payment platform: provides payment records of parking fees, including payment time, amount, license plate number and other information.

[0064] Create a spreadsheet or database template to record parking data in a standardized manner. The template contains the following fields: parking lot ID, parking space number, parking start time, parking end time, parking duration, parking fee, license plate number, payment method, etc.

[0065] Automated Data Collection:

[0066] API interface connection: Establish API connection with parking management system, mobile payment platform, etc. to achieve automatic synchronization and update of data. Set up timer scripts to regularly grab the latest data from each system to ensure the real-time nature of the data.

[0067] For data that cannot be collected through automated means (such as specific events and abnormal situations), on-site staff will record them manually. Information about car owners' parking experience and cost perception can be collected through online or offline questionnaires. Data collected from different sources will be imported into a unified data storage platform (such as a database). Data cleaning will be performed, including removing duplicate records, correcting erroneous data, and processing missing values, to ensure the accuracy and completeness of the data. Preliminary verification will be performed on the cleaned data to ensure that the data quality meets the requirements of subsequent analysis.

[0068] In the embodiment of the present invention, in the above step 12, the quality analysis score The calculation formula is:

[0069] ;

[0070] in, Indicates The number of error records for each indicator; Indicates The total number of records for each indicator; Indicates The number of recorded data points for a metric; Indicates The number of data points to be recorded for each indicator; Indicates The number of contradictory records for each indicator; Indicates The total number of records related to the indicator; Indicates The time-effectiveness decay rate of each indicator; Indicates The data delay time of each indicator; and They are The mean and standard deviation of each indicator; is the total number of data quality analysis indicators; Indicates the index value of the indicator; represents the base of natural logarithms; , , , and Represents the weight coefficient.

[0071] In the embodiment of the present invention, the collected raw parking data is cleaned to remove invalid, erroneous or duplicate records to ensure the accuracy of the data. The data is standardized to conform to specific formats and specifications to facilitate subsequent data quality analysis. The data is checked for missing values, and records containing missing values ​​are filled or deleted according to actual conditions.

[0072] Data quality analysis:

[0073] In order to comprehensively evaluate the data quality, several evaluation indicators are defined, including the number of error records. , Total number of records , the number of data points recorded , the number of data points to be recorded , Number of conflicting records , Total number of related records , Time-dependent decay rate , Data delay time wait.

[0074] The quality analysis score is calculated using the given formula This formula takes into account the accuracy, completeness, consistency, timeliness and other aspects of the data, and uses the weight coefficient , , , and to adjust the importance of different evaluation dimensions.

[0075] Calculation process:

[0076] For each evaluation metric, calculate its components, such as the error recording rate , Record completeness rate , conflict record rate Etc. Considering the timeliness of data, the timeliness decay rate and data delay time Calculate the timeliness factor The probability density function of the normal distribution is introduced to evaluate the stability of the data, where and They are The weighted sum of the above parts is calculated using the given formula to obtain the final quality analysis score Q.

[0077] The quality of parking data can be quantitatively evaluated based on the calculated Q value. A higher Q value means better data quality, while a lower Q value indicates that there are quality issues with the data. By analyzing the scores of each evaluation indicator, specific data quality issues and improvement directions can be further identified. This step helps ensure the accuracy and reliability of parking data by combining multiple evaluation dimensions and quantitative indicators.

[0078] In the embodiment of the present invention, in the above step 13, the data application score The calculation formula is:

[0079] ;

[0080] in, Indicates the number of The number of times the data was queried. Indicates the total duration of the statistical time period. Indicates the maximum possible number of queries; Indicates the number of covered parking lots, Indicates the total number of parking lots; represents the average data delay time, Indicates the maximum acceptable delay time; Indicates the query count index value; Indicates the total number of queries; , and Represents the weight coefficient.

[0081] In the embodiment of the present invention, the core indicators of data application evaluation are determined, including the frequency of data use, the coverage of data, and the real-time nature of data. Reasonable weight coefficients are set for these indicators according to actual business needs and scenarios. , and .

[0082] Data usage frequency assessment:

[0083] Statistics for a specific time period The total number of times the data was queried .

[0084] Calculate the maximum possible number of queries within the time period, which is usually based on the system's maximum query capacity or a preset query frequency cap.

[0085] By formula Get the normalized value of the frequency of data usage.

[0086] Determine the number of parking lots that have been covered; get the total number of parking lots, which can be the number of all parking lots in the area. Calculate the coverage ratio of the data. Measure the average latency of the data, that is, the average time from the generation of data to the receipt of the system and available for query. Set the maximum acceptable data latency, which is determined by business needs. Data exceeding this time may be considered not real-time. Calculate the real-time score of the data, where lower latency results in higher real-time scores.

[0087] Comprehensive Assessment:

[0088] Using the scores of the above three evaluation indicators and their respective weight coefficients, the formula Calculate the comprehensive data application score A.

[0089] This step comprehensively considers multiple key dimensions of data in practical applications, including frequency of use, coverage, and real-time performance, so as to more comprehensively evaluate the application effect of data. Through specific calculation formulas and quantitative indicators, the data application score is more objective, accurate, and comparable, which helps to find bottlenecks and optimization directions in data applications. The introduction of weight coefficients allows the evaluation focus to be adjusted according to different business scenarios and needs, making the evaluation results more in line with actual conditions and decision-making needs.

[0090] In a preferred embodiment of the present invention, in the above step 14, the specific calculation formula corresponding to the fitness is:

[0091] ;

[0092] in, Representative The fitness value of individual data is the weight coefficient; Indicates Individuals in the data application evaluation Scores on the dimensions; is the total number of dimensions of data application evaluation; is an index; is a control coefficient; is an index from 1 to m, used to traverse the dimensions of data application evaluation; It is a specific index used to identify an individual in the genetic algorithm.

[0093] Perform crossover and mutation operations on the selected individuals to generate new individuals, and repeat the selection, crossover and mutation operations until the termination condition is met to obtain the corresponding optimized parameters, including:

[0094] Roulette wheel selection is used to select a portion of individuals from the current population to enter the next generation;

[0095] For the selected individual pairs, a multi-point crossover operation is performed at a certain crossover rate to generate new individuals by exchanging some genes between individuals;

[0096] For the new individuals generated after crossover, bit flipping operations are performed at a certain mutation rate;

[0097] The new individuals generated after multi-point crossover and bit flipping are combined with some individuals in the original population to form a new population. Check whether the termination condition is met. If the termination condition is met, stop the iteration and output the corresponding individuals in the current population as the optimized parameter combination.

[0098] In the embodiment of the present invention, the weight coefficients of data quality analysis and data application evaluation are optimized by genetic algorithm, so that the data processing is more accurate. The automated genetic algorithm iteration process can quickly find the optimal weight coefficient combination, thereby improving the evaluation efficiency. Through the quantitative evaluation method and the optimized weight coefficient, the objectivity and flexibility of data processing are increased, so that the evaluation results can better reflect the actual value of the data.

[0099] When the above step 14 is applied specifically, it specifically includes the following steps:

[0100] An initial population is randomly generated, which consists of multiple individuals, each of which represents a possible combination of data processing parameters. The parameter combination here mainly refers to the weight coefficient in the data application score, which will be used for subsequent data value calculations. Each individual can be represented by a vector, and each element of the vector corresponds to a weight coefficient.

[0101] The fitness of each individual is calculated according to the preset fitness function (such as the above formula). The fitness reflects the pros and cons of the parameter combination represented by the individual in data processing. Then, a part of the individuals are selected from the current population to enter the next generation using the roulette wheel (also known as proportional selection) method. Specifically, the probability of each individual being selected is determined according to the proportion of its fitness in the total fitness. Individuals with high fitness have a high probability of being selected. The selected individuals are randomly paired into multiple pairs of individuals. For each pair of individuals, a multi-point crossover operation is performed at a certain crossover rate (such as 0.7 or 0.8). Specifically, multiple crossover points are randomly selected in the vector representation of the individual, and then some genes (i.e., weight coefficients) after these points are exchanged. In this way, each newly generated individual incorporates some of the characteristics of its parent individuals.

[0102] For the new individuals generated after crossover, a bit flipping operation is performed at a certain mutation rate (such as 0.01 or 0.001). Specifically, a position in the individual vector is randomly selected and the gene value (weight coefficient) on it is flipped or replaced with another random value. This helps to increase the diversity of the population and prevent the algorithm from converging to a local optimal solution too early.

[0103] Combine the new individuals generated after multi-point crossover and bit flipping with some excellent individuals in the original population (such as the individuals with the highest fitness) to form a new population. Check whether the new population meets the preset termination conditions. These conditions may include reaching the maximum number of iterations. If the termination conditions are met, stop the iteration and output the individual with the highest fitness in the current population as the optimized parameter combination. These parameter combinations will be used for subsequent data processing.

[0104] In a preferred embodiment of the present invention, the calculation formula of the regional index is:

[0105] ;

[0106] in, is the regional index; The quality analysis score is The weight coefficient of each indicator; The quality analysis score is The score of each indicator; The quality analysis score is The index coefficient of an indicator; The data application score is The weight coefficient of each dimension; The data application score is The scores of the dimensions; The data application score is The exponential coefficient of the dimension.

[0107] In the embodiment of the present invention, the formula comprehensively considers multiple dimensions of data quality (such as accuracy, completeness, etc.) and data application (such as frequency of use, coverage, etc.), ensuring a comprehensive assessment of the value of parking data. and They are the weight coefficients of each indicator in the data quality and data application scores, reflecting the importance of these indicators in the overall evaluation. and They are the scores of each indicator in data quality and data application assessment, representing the performance in these specific aspects. and are exponential coefficients that may be used to adjust the influence of each indicator score, allowing a nonlinear reflection of the relationship between the indicator score and the value. This formula can more accurately evaluate the value of parking data by taking multiple dimensions into account.

[0108] In a preferred embodiment of the present invention, after calculating the regional index according to the quality analysis score and the final data application score, the method further comprises:

[0109] Extract key features related to parking demand from historical parking data, including timestamps, weather conditions, holiday information, and nearby events, specifically including: collecting a large amount of historical parking data, which is usually stored in a database and contains various operational information of the parking lot; from these data, select features closely related to parking demand, such as timestamps (reflecting parking periods, such as weekdays, weekends, peak hours, etc.), weather conditions (affecting travel and parking choices), holiday information (changes in parking demand on special dates), and nearby events (such as large-scale concerts, sports events, etc. that may lead to a surge in parking demand); clean and preprocess the selected features, such as removing outliers, filling missing values, and standardizing data, to ensure data quality and consistency.

[0110] Encode and convert key features, use historical parking data as a training set, train the neural network model, and optimize the parameters of the neural network model through cross-validation and grid search to obtain the trained neural network model. Specifically, for non-numeric features, such as weather conditions or holiday types, use encoding techniques (such as one-hot encoding, label encoding, etc.) to convert them into numerical data that the model can process. Convert the processed data into a format suitable for neural network model input, such as tensor or data frame. According to the complexity of the problem, build a suitable neural network model, which can include a long short-term memory network (LSTM) to learn the complex relationship between parking demand and key features. Divide the historical parking data into a training set, a validation set, and a test set, where the training set is used for model training, the validation set is used to adjust model parameters, and the test set is used to evaluate the performance of the model.

[0111] The neural network model is trained using the training set, and the weights and biases of the model are adjusted through the back propagation algorithm to minimize the error between the predicted value and the true value, including:

[0112] The connection weights between each neuron in the neural network and the bias of each neuron are given initial values, which are determined using a specific initialization strategy, such as He initialization or Xavier initialization.

[0113] For each sample in the training set, it is first input into the neural network. The input data passes through each layer of the neural network, undergoes nonlinear transformation through activation functions (such as ReLU), and propagates forward layer by layer until the final output of the network is obtained. This output is the model's predicted value for the current input sample.

[0114] Compare the model's predicted value with the true value of the sample, and use the loss function to calculate the difference between the two, that is, the error. This error reflects the accuracy of the model's prediction of the sample under the current parameters. The calculation formula of the loss function is:

[0115] ;

[0116] in, Represents the complex binary cross entropy loss, which reflects the difference between the probability distribution predicted by the model and the true label. Is the true label of the sample. For a binary classification problem, it usually takes a value of 0 or 1, indicating the category to which the sample belongs. It is the probability that the model predicts that the sample is a positive class (for example, the label is 1). This probability value is calculated by the output layer of the neural network (such as the sigmoid function) and its value range is between 0 and 1. and These two items represent the logarithmic probabilities of the model predicting the positive and negative classes, respectively. The logarithmic function is used to convert the probability value into a loss value, so that when the predicted probability is close to the true label, the loss value is small; and when the predicted probability is far from the true label, the loss value is large. and These two factors increase the nonlinearity of the log loss. They increase the rate of change of the loss when the predicted probability is close to 0 or 1, thus encouraging the model to be more cautious when making very certain or uncertain predictions. This term is in the form of a Gaussian function, which predicts the probability It reaches its maximum value when it approaches 0.5. This term is used as part of the denominator to adjust the sensitivity of the loss. When the predicted probability is close to 0.5 (that is, the model is less certain about the category), this term will increase, making the loss more sensitive to predictions in this area.

[0117] This is the calculated cross entropy loss value, which reflects the difference between the model's prediction and the true label. During training, the goal is to minimize this loss value by adjusting the model's parameters. When the model's prediction is very accurate, the cross entropy loss will be close to 0; when the prediction is inaccurate, the loss value will increase.

[0118] The network weights and biases are adjusted according to the error. Starting from the output layer, the error gradient is calculated backward layer by layer, that is, the partial derivative of the error with respect to the weights and biases of each layer. These gradients indicate how each parameter should be adjusted in order to reduce the error.

[0119] Use gradient descent to update the weights and biases of the network. By continuously updating the parameters, the model gradually learns how to better map input to output, thereby reducing the prediction error.

[0120] Repeat the above steps for multiple iterations of the entire training set until the model's performance reaches the preset standard or no longer improves significantly. In each iteration, mini-batches of data are used to speed up the training process and reduce the consumption of computing resources.

[0121] Use cross-validation techniques (such as k-fold cross-validation) to evaluate the performance of the model under different data partitions to reduce the risk of overfitting and underfitting. Use grid search or random search methods to find the corresponding model parameter combination in the preset parameter space, such as learning rate, batch size, number of hidden layer nodes, etc. According to the results of cross-validation and parameter search, select the model with corresponding performance as the final prediction model.

[0122] Use the trained neural network model to simulate and run, predict parking demand in different scenarios, and obtain simulation results, including predicted demand and demand fluctuation trends, including: defining different prediction scenarios, such as normal working days, weekends, holidays, special weather conditions or nearby large-scale events, etc. Generate corresponding feature input data for each scenario, including timestamp, weather conditions, holiday information, etc. Input the generated feature data into the trained neural network model, perform simulation runs, and obtain parking demand prediction results in different scenarios.

[0123] Compare the simulation results under different scenarios with the actual operation data to obtain the analysis results, including: collecting the actual operation data corresponding to the simulation scenario, and performing necessary sorting and processing; comparing and analyzing the simulation results with the actual operation data, and calculating indicators such as prediction error and accuracy to evaluate the prediction performance of the model.

[0124] According to the analysis results, the risk warning threshold is set, and the warning is triggered when the simulation result exceeds the threshold. Specifically, based on the results of the comparative analysis, combined with the actual operation needs and risk tolerance of the parking lot, a reasonable risk warning threshold is set. These thresholds can include the upper limit of parking demand, the fluctuation range of demand, etc. When the simulation result exceeds the set threshold, the corresponding warning mechanism is triggered, such as sending an alarm email, SMS notification, etc., so that managers can take timely measures to deal with potential risks and problems.

[0125] In the embodiment of the present invention, by extracting key features from historical parking data and encoding these features, various factors affecting parking demand can be captured more accurately. The trained neural network model can accurately predict parking demand based on these factors, thereby helping managers to better understand future parking demand. Accurate parking demand prediction enables parking lot managers to make reasonable resource allocation decisions in advance, such as adjusting the number of parking spaces, etc., to meet parking needs in different scenarios and improve the operational efficiency and customer satisfaction of parking lots. By comparing simulation results with actual operating data, managers can promptly discover potential risks and problems, such as abnormal fluctuations in parking demand and uneven parking space utilization. After setting the risk warning threshold, when the simulation result exceeds the threshold, an early warning can be triggered, so that managers can take timely measures to respond and reduce operational risks. The rapid prediction capability of the neural network model can help managers obtain simulation results for a large number of scenarios in a short period of time, thereby accelerating the decision-making process and improving decision-making efficiency. The entire process makes full use of parking data, realizes full-chain management from data evaluation to data application, promotes the transformation of parking lots to a data-driven management model, and improves the level of refined management. By simulating and predicting parking demand in different scenarios, parking lots can make adaptive adjustments more quickly when faced with emergencies or abnormal situations, thereby enhancing the system's resilience and risk resistance.

[0126] In the embodiment of the present invention, the parking revenue is calculated based on the regional index, combined with the parking lot location information, parking space usage, parking time and parking fee; and the parking space allocation strategy is dynamically processed based on the parking revenue, which may include:

[0127] Get the regional index calculated previously, which reflects the relative profitability of different regions.

[0128] According to the layout and characteristics of the parking lot, the parking lot is divided into several areas. These areas can be divided based on physical location, type of parking space (such as ordinary parking space, disabled parking space, electric vehicle charging parking space, etc.), frequency of use or other relevant factors. For each demarcated area, a corresponding profit index value is assigned to the parking spaces in the area according to the previously calculated regional index. If an area is evaluated as a high-profit area, all parking spaces in the area will share a relatively high profit index value. Within each area, the profit index of a single parking space can also be fine-tuned according to the specific conditions of the parking space (such as proximity to the entrance and exit, whether the view is wide, whether it is convenient to park, etc.). For example, a parking space close to the entrance or exit of the parking lot may have a higher use value due to convenience, so a slightly higher profit index value can be assigned. The use of each parking space is tracked in real time, including parking time, parking fees, etc. When a vehicle leaves the parking space, the weighted parking revenue is calculated based on the profit index of the parking space and the actual parking fee. The specific calculation method can be to multiply the actual parking fee by the profit index value of the parking space.

[0129] For example, suppose a parking lot is divided into three areas, A, B, and C, where area A is a high-yield area, area B is a medium-yield area, and area C is a low-yield area. The parking spaces in each area will be assigned a basic profit index value, such as 1.2 for area A, 1.0 for area B, and 0.8 for area C. In area A, there is a parking space close to the entrance. Due to its convenience, we can fine-tune its profit index to 1.3. When a car parks in this parking space and pays, we multiply the actual parking fee by 1.3 to calculate the weighted profit of the parking space. In this way, the actual profit capacity of each parking space can be more accurately reflected, and the parking space allocation strategy can be dynamically adjusted based on these data.

[0130] Based on the actual revenue data of each parking space, identify high-revenue and low-revenue parking spaces, and analyze the changing trends of parking space revenue in different time periods, such as the difference between weekdays and weekends, daytime and nighttime. Prioritize the allocation of idle high-revenue parking spaces to vehicles entering the parking lot to maximize revenue. During peak hours, parking fees can be adjusted dynamically, such as increasing fees for high-revenue parking spaces to reflect their higher value. For low-revenue parking spaces that have been idle for a long time, consider reducing fees or conducting promotional activities to attract more car owners to use them. The formulated dynamic allocation strategy is implemented through the parking management system. The execution of the strategy is monitored through real-time data, including key indicators such as parking space turnover rate, average parking time and actual revenue. The effectiveness of the strategy is regularly evaluated, and necessary adjustments and optimizations are made based on feedback data.

[0131] like Figure 2As shown, an embodiment of the present invention further provides a data asset evaluation system based on multi-level analysis, comprising:

[0132] An acquisition module is used to collect parking data, wherein the parking data includes parking lot location information, parking space usage, parking time, and parking fees;

[0133] The quality analysis module is used for the data quality analysis module, and is used for preprocessing the collected parking data to obtain preprocessed parking data; performing quality analysis on the preprocessed parking data based on preset data quality analysis indicators to obtain a quality analysis score;

[0134] A data application processing module, used to perform application scoring on the pre-processed parking data based on preset data analysis indicators to obtain a data application score, wherein the data analysis indicators include the frequency of data use, the coverage of the data, and the real-time nature of the data;

[0135] The calculation module is used to set the initial population of the genetic algorithm, each individual represents a data processing parameter combination, wherein the data processing parameter combination is a combination of weight coefficients in the data application score; the fitness of the individual is calculated according to the data application score; the corresponding individual is selected to enter the next generation according to the fitness of the individual; the selected individual is crossover and mutation operations are performed to generate new individuals, and the selection, crossover and mutation operations are repeated until the termination condition is met to obtain the corresponding optimized parameters; the final data application score is recalculated using the optimized parameters; the regional index is calculated according to the quality analysis score and the final data application score; the parking revenue is calculated according to the regional index in combination with the parking lot location information, parking space usage, parking time and parking fee; and the parking space allocation strategy is dynamically processed according to the parking revenue.

[0136] It should be noted that the system is a system corresponding to the above method, and all implementation methods in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

[0137] The embodiment of the present invention further provides a computing device, comprising: a processor, a memory storing a computer program, wherein when the computer program is executed by the processor, the method described above is executed. All implementations in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

[0138] The embodiment of the present invention also provides a computer-readable storage medium storing instructions, which, when executed on a computer, enable the computer to execute the method described above. All implementations in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

[0139] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A data asset evaluation method based on multi-level analysis, characterized in that: The method comprises: Collect parking data, including parking lot location information, parking space usage, parking time, and parking fees; Preprocessing the collected parking data to obtain preprocessed parking data; performing quality analysis on the preprocessed parking data based on preset data quality analysis indicators to obtain a quality analysis score; Based on preset data analysis indicators, the pre-processed parking data is scored for application to obtain a data application score, wherein the data analysis indicators include the frequency of data use, the coverage of data, and the real-time nature of data; The initial population of the genetic algorithm is set, and each individual represents a data processing parameter combination, wherein the data processing parameter combination is a combination of weight coefficients in the data application score; the fitness of the individual is calculated according to the data application score; the corresponding individual is selected to enter the next generation according to the fitness of the individual; the selected individual is crossover and mutation operation is performed to generate a new individual, and the selection, crossover and mutation operation are repeated until the termination condition is met to obtain the corresponding optimized parameters; the final data application score is recalculated using the optimized parameters; the regional index is calculated according to the quality analysis score and the final data application score; Calculate parking revenue based on the regional index, combined with parking lot location information, parking space usage, parking time and parking fees; dynamically process parking space allocation strategies based on parking revenue; The specific calculation formula corresponding to fitness is: ； in, Representative The fitness value of each individual data; is the exponential coefficient of the jth dimension in the data application score; Indicates is the score of an individual on the jth dimension of data application evaluation; m is the total number of data application evaluation dimensions; is a control coefficient; j is an index from 1 to m, used to traverse the dimensions of data application evaluation; Is an index used to identify an individual in the genetic algorithm.

2. A data asset evaluation method based on multi-level analysis according to claim 1, characterized in that: Perform crossover and mutation operations on the selected individuals to generate new individuals, and repeat the selection, crossover and mutation operations until the termination condition is met to obtain the corresponding optimized parameters, including: Roulette wheel selection is used to select a portion of individuals from the current population to enter the next generation; For the selected individual pairs, a multi-point crossover operation is performed to generate new individuals by exchanging some genes between the individuals; For the new individuals generated after crossover, perform bit flipping operation; The new individuals generated after multi-point crossover and bit flipping are combined with some individuals in the original population to form a new population. Check whether the termination condition is met. If the termination condition is met, stop the iteration and output the corresponding individuals in the current population as the optimized parameter combination.

3. A data asset evaluation method based on multi-level analysis according to claim 2, characterized in that: The calculation formula of regional index is: ; Where V is the regional index; is the weight coefficient of the i-th indicator in the quality analysis score; is the score of the ith indicator in the quality analysis score; is the exponential coefficient of the i-th indicator in the quality analysis score; is the weight coefficient of the jth dimension in the data application score; is the score of the jth dimension in the data application score; is the exponential coefficient of the j-th dimension in the data application score.

4. A data asset evaluation method based on multi-level analysis according to claim 3, characterized in that: Based on the quality analysis score and the final data application score, the regional index is calculated, which also includes: Extract key features related to parking demand from historical parking data, including timestamps, weather conditions, holiday information, and nearby events; The key features are converted into codes, and the neural network model is trained using the historical parking data as a training set. The parameters of the neural network model are optimized through cross-validation and grid search to obtain the trained neural network model. Use the trained neural network model to run simulations to predict parking demand in different scenarios to obtain simulation results, including predicted demand and demand fluctuation trends; Compare simulation results under different scenarios with actual operation data to obtain analysis results; According to the analysis results, the risk warning threshold is set, and the warning is triggered when the simulation results exceed the threshold.

5. A data asset evaluation system based on multi-level analysis, characterized in that: Applied to the method according to any one of claims 1 to 4, comprising: An acquisition module is used to collect parking data, wherein the parking data includes parking lot location information, parking space usage, parking time, and parking fees; The quality analysis module is used for the data quality analysis module, and is used for preprocessing the collected parking data to obtain preprocessed parking data; performing quality analysis on the preprocessed parking data based on preset data quality analysis indicators to obtain a quality analysis score; A data application processing module, used to perform application scoring on the pre-processed parking data based on preset data analysis indicators to obtain a data application score, wherein the data analysis indicators include the frequency of data use, the coverage of the data, and the real-time nature of the data; The calculation module is used to set the initial population of the genetic algorithm, each individual represents a data processing parameter combination, wherein the data processing parameter combination is a combination of weight coefficients in the data application score; the fitness of the individual is calculated according to the data application score; the corresponding individual is selected to enter the next generation according to the fitness of the individual; the selected individuals are crossover and mutation operations are performed to generate new individuals, and the selection, crossover and mutation operations are repeated until the termination condition is met to obtain the corresponding optimized parameters; the final data application score is recalculated using the optimized parameters; the regional index is calculated according to the quality analysis score and the final data application score; the parking revenue is calculated according to the regional index in combination with the parking lot location information, parking space usage, parking time and parking fee; and the parking space allocation strategy is dynamically processed according to the parking revenue.

Images