Big data risk assessment system based on AI

Abstract

The invention relates to the technical field of artificial intelligence, in particular to an AI-based big data risk assessment system, which comprises the following modules: a time offset correction module, a risk factor modeling module, a mutation factor monitoring module, a risk cross derivation module and a prediction optimization adjustment module. According to the method, the time density of multi-source data is aligned, a reference baseline of time series data is unified, the time consistency of subsequent factor modeling is improved, the identification capability of a linkage effect between factors is enhanced by combining the trend deviation rate and co-occurrence progressive relation of risk factors, and judgment is performed by utilizing the trend deviation degree and residual error continuity of an intersection point; risk wave and range identification under the multivariable effect is optimized, a prediction error deviation trend is combined with a weight self-adjusting mechanism, dynamic correction of model parameters is achieved, it is ensured that the model adapts to dynamic changes of the risk environment in the long-term operation process, a risk prediction path has the self-adaptive repairing capacity, and the risk prediction accuracy is improved. And the stability and accuracy of big data risk prediction are improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, in particular to a big data risk assessment system based on AI. Background Art

[0002] The field of artificial intelligence technology includes multiple sub-fields such as machine learning, deep learning, natural language processing, and computer vision. It aims to use computers to simulate human intelligence and complete tasks such as perception, reasoning, and decision-making. The core content of this technical field includes data collection and preprocessing, feature engineering, model training and optimization, reasoning calculation, etc. It relies on large-scale data sets to build intelligent algorithms to achieve pattern recognition, automatic reasoning, and adaptive learning. Artificial intelligence technology is widely applied in multiple fields such as speech recognition, image analysis, autonomous driving, medical diagnosis, and financial risk control. It relies on neural networks, statistical learning, and reinforcement learning to improve the decision-making ability of computers and improve the accuracy of prediction and analysis through continuous learning and optimization.

[0003] Among them, the big data risk assessment system based on AI refers to using artificial intelligence technology to analyze, identify, and evaluate risks for batch data, discover potential risk factors, and form an assessment result. The technical matters targeted by this system cover multiple links such as data acquisition, data preprocessing, risk modeling, risk classification, and risk prediction. Data acquisition usually collects multi-source heterogeneous data through web crawler technology, database connection, interface call, etc., and completes data preprocessing by combining steps such as data cleaning, deduplication, and format conversion. Risk modeling uses means such as statistical analysis, clustering analysis, and feature extraction to build a risk assessment index system and trains a classification model based on machine learning, such as decision trees, support vector machines, neural networks, etc., to achieve the classification of risk categories. Risk prediction builds a trend prediction model based on methods such as time series analysis and regression analysis, combines methods such as threshold setting and rule matching to identify potential risks, and combines a real-time monitoring mechanism to label and track abnormal data.

[0004] In the process of risk prediction using traditional big data risk assessment technologies, due to the asynchronous timestamps of multi-source data, there is a time-series misalignment after data integration, resulting in the risk assessment being based on an incorrect time series, which affects the prediction accuracy. The time-series changes of risk factors are not calculated for trend offsets under multiple time windows, so short-term fluctuations may be misjudged as long-term trends, reducing the assessment accuracy. The monitoring of abnormal data is usually based on static threshold settings and cannot adapt to the dynamic changes of mutation factors, resulting in false alarms or missed alarms and reducing the response ability to sudden risks. The identification of risk crossover points relies on the threshold comparison method of a single variable and it is difficult to effectively measure the complex interaction relationships between multiple variables, leading to the incompleteness of the risk crossover conduction path. The prediction error correction depends on a fixed weight adjustment method and cannot be adaptively optimized according to the error change trend, resulting in a decline in the accuracy of the prediction model during long-term operation and making it difficult to cope with the continuous changes in the risk environment. Summary of the Invention

[0005] To solve the technical problems existing in the prior art, an embodiment of the present invention provides a big data risk assessment system based on AI. The technical solution is as follows:

[0006] On the one hand, a big data risk assessment system based on AI is provided, and the system includes:

[0007] The time offset correction module collects the timestamp fields of multi-source data, identifies the positions of reference alignment data points by comparing the difference in the distribution density of adjacent segment data points, corrects the time stamp sequence of the data source, and establishes a data time-series offset value;

[0008] Based on the data time-series offset value, the risk factor modeling module extracts the fluctuation rate of the risk factor data for each time window, calculates the trend offset rate of the risk factor data in adjacent windows, sets risk factor association rules according to the co-occurrence and time-series relevance of risk factors, analyzes the progressive influence chain of risk factors, constructs a risk factor regression model, and establishes a risk factor influence gradient value;

[0009] Based on the risk factor influence gradient value, the mutation factor monitoring module detects the trajectory offset direction of abnormal data by monitoring the fluctuation range of abnormal data in the risk factor data, predicts the risk trigger time point, and generates a mutation factor trigger time point value;

[0010] The risk crossover derivation module calls the risk factor influence gradient value, calculates the difference in the parameter offset rate, identifies the candidate positions of risk crossover points, and analyzes the continuity of the crossover point trend through the deviation degree between the data trend fitting residual and the current crossover point trend, generating a risk crossover point impact range value.

[0011] As a further solution of the present invention, the data timing offset value includes timestamp difference distribution information, alignment reference point position, and segment density offset. The risk factor influence gradient value includes factor fluctuation rate, trend offset rate, and association rule group. The mutation factor trigger time point value specifically refers to the abnormal fluctuation interval, trajectory offset direction, and trigger prediction time. The risk intersection range value specifically refers to the rate difference, trend deviation degree, and trend continuity level.

[0012] As a further solution of the present invention, the time offset correction module includes:

[0013] The timestamp extraction sub-module collects the timestamp fields of multi-source data, extracts the data point timestamps within each data segment, records the start time, end time, and time interval of the timestamps in each data segment, and obtains the timestamp interval record value.

[0014] The data alignment determination sub-module, based on the timestamp interval record value, obtains the data point distribution density of each data segment, calculates the distribution density difference between adjacent segments, and sets the reference alignment point position according to the distribution density difference.

[0015] The timing correction sub-module calls the reference alignment point position, analyzes the offset between the timestamp of each data segment and the reference alignment point, corrects the data point timestamp of each segment according to the timestamp interval difference, and establishes the data timing offset value.

[0016] As a further solution of the present invention, the risk factor modeling module includes:

[0017] The fluctuation feature extraction sub-module, based on the data timing offset value, obtains the risk factor data within each time window, detects the numerical change amplitude of the risk factor in each time window and the change rate of adjacent data points, calculates the change gradient of the data points within each time window, and obtains the risk factor fluctuation rate value.

[0018] The trend offset calculation sub-module calls the risk factor fluctuation rate value, calculates the difference in the change rate of risk factor data between adjacent windows, analyzes the change trend of the risk factor in the time dimension, and obtains the risk factor trend offset value.

[0019] The regression model construction sub-module calls the risk factor trend offset value, sets the association rules between risk factors by analyzing the co-occurrence features and time series association features of multiple risk factors, combines the trend change amplitude of the risk factors, analyzes the progressive influence chain, constructs the risk factor regression model, and obtains the risk factor influence gradient value.

[0020] As a further solution of the present invention, the mutation factor monitoring module includes:

[0021] The abnormal point monitoring sub-module, based on the influence gradient value of the risk factor, detects the change amplitude of each data point by analyzing the numerical fluctuation in the risk factor data, calculates the deviation degree of each data point, detects abnormal data points, and obtains the abnormal data deviation value;

[0022] The trajectory analysis sub-module calls the abnormal data deviation value, obtains the time series information of the abnormal data points, calculates the change direction of the abnormal data between adjacent time points, analyzes the change trend of the abnormal data on the time axis, extracts the offset path with continuous data change trend, and obtains the abnormal data trajectory offset value;

[0023] The time point prediction sub-module calls the abnormal data trajectory offset value, calculates the trajectory extension rate by analyzing the trajectory change trend of the abnormal data, analyzes the amplitude of the trend change, predicts the risk trigger time point, and obtains the mutation factor trigger time point value.

[0024] As a further solution of the present invention, the specific formula for calculating the trajectory extension rate is:

[0025]

[0026] Calculate the trajectory extension rate;

[0027] wherein, V TRP represents the trajectory extension rate, X end represents the value of the trajectory end point, X start represents the value of the trajectory start point, X i′ represents the value of the i'-th data point, X i′+1 represents the value of the (i'+1)-th data point, n' represents the total number of abnormal data points in the trajectory, T end represents the time of the trajectory end point, T start represents the time of the trajectory start point, ΔT i′ represents the time interval between adjacent data points, n' represents the total number of abnormal data points in the time series, i' represents the index number of the abnormal data point in the time series, and i'+1 represents the index of the next data point of the current data point i' in the trajectory sequence.

[0028] As a further solution of the present invention, the risk cross-derivation module includes:

[0029] The parameter offset calculation sub-module, based on the influence gradient value of the risk factor, collects the change data of the risk factors in the continuous time series, calculates the parameter change rate of each risk factor between adjacent time nodes, obtains the change difference of the parameter rates between risk factors, and establishes the parameter rate offset difference;

[0030] The intersection point recognition sub-module calls the parameter rate offset difference value, predicts the intersection area position of the risk factor parameter change path according to the set rate offset threshold, and establishes a risk intersection candidate area;

[0031] The trend continuity analysis sub-module calls the risk intersection candidate area, obtains the parameter trend data within the intersection area, calculates the trend fitting residual value and the trend deviation amplitude, analyzes the continuity of the intersection point trend, and establishes a risk intersection point spread range value.

[0032] As a further solution of the present invention, the system further includes:

[0033] The prediction optimization and adjustment module calls the mutation factor trigger time point value and the risk intersection point spread range value, combines the deviation between the actual risk event backtracking data and the prediction result, calculates the prediction error deviation trend, adjusts the model parameters, including the weight coefficient of each risk factor, and generates a risk prediction correction value;

[0034] The risk prediction correction value includes an error deviation trend, a risk weight adjustment value, and a model parameter correction coefficient.

[0035] As a further solution of the present invention, the prediction optimization and adjustment module includes:

[0036] The error calculation sub-module calculates the time deviation between the predicted risk trigger time point and the actual risk event occurrence time based on the mutation factor trigger time point value and the risk intersection point spread range value, and establishes a prediction error time deviation value;

[0037] The change trend analysis sub-module calls the prediction error time deviation value, obtains the change data of the error value within a continuous time window, calculates the trend rate of the error change, and analyzes the error change direction, and establishes an error change trend value;

[0038] The parameter adjustment sub-module calls the error change trend value, and optimizes multiple parameters of the risk factor regression model through iterative adjustment, including the weight coefficient of each risk factor, and establishes a risk prediction correction value.

[0039] As a further solution of the present invention, the specific formula for optimizing multiple parameters of the risk factor regression model is:

[0040]

[0041] Calculate the adjusted risk factor weight;

[0042] Among them, W new represents the adjusted risk factor weight, W old represents the original risk factor weight, k represents a risk adjustment factor, V errRepresents the rate of change trend of the error, V base Represents the baseline change rate of the risk factor, ΔT err Represents the value of the error change trend.

[0043] The beneficial effects brought by the technical solution provided by the embodiment of the present invention at least include:

[0044] By aligning the time density of multi-source data, unifying the reference baseline of time-series data, enhancing the time consistency of subsequent factor modeling, combining the trend deviation rate and co-occurrence progressive relationship of risk factors, enhancing the recognition ability of the linkage effect between factors, using the trend deviation degree and residual continuity judgment at the intersection point to optimize the recognition of the risk propagation range under the action of multiple variables, and adopting the prediction error deviation trend combined with the weight self-adjustment mechanism, the dynamic correction of model parameters is realized, ensuring that the model adapts to the dynamic changes of the risk environment during long-term operation, enabling the risk prediction path to have the ability of self-adaptive repair, and enhancing the stability and accuracy of big data risk prediction. Description of the Drawings

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0046] Figure 1 Is the system flow chart of the present invention;

[0047] Figure 2 Is the schematic diagram of the system framework of the present invention;

[0048] Figure 3 Is the flow chart of the time offset correction module of the present invention;

[0049] Figure 4 Is the flow chart of the risk factor modeling module of the present invention;

[0050] Figure 5 Is the flow chart of the mutation factor monitoring module of the present invention;

[0051] Figure 6 Is the flow chart of the risk cross-derivation module of the present invention;

[0052] Figure 7 Is the flow chart of the prediction optimization and adjustment module of the present invention. Detailed Embodiments

[0053] The following will describe the technical solutions in the present invention with reference to the drawings.

[0054] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to give examples, illustrations or explanations. Any embodiment or design solution described as an "example" in the present invention should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Rather, the use of the word "example" is intended to present concepts in a specific manner. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one of the two can be selected.

[0055] In the embodiments of the present invention, "image" and "picture" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, the meanings they express are the same. "(of)", "corresponding", and "corresponding" can sometimes be used interchangeably. It should be noted that when the difference is not emphasized, the meanings they express are the same.

[0056] In the embodiments of the present invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meanings they express are the same.

[0057] To make the technical problems to be solved, technical solutions and advantages of the present invention clearer, the following will be described in detail with reference to the drawings and specific embodiments.

[0058] The embodiments of the present invention provide a big data risk assessment system based on AI. Please refer to Figures 1 to 2 , the present invention provides a technical solution. The big data risk assessment system based on AI includes:

[0059] The time offset correction module collects the timestamp fields of multi-source data, identifies the positions of reference alignment data points by comparing the difference in the distribution density of adjacent segment data points, corrects the data source timestamp sequence, and establishes a data time series offset value;

[0060] The risk factor modeling module, based on the data time series offset value, extracts the fluctuation rate of the risk factor data for each time window, calculates the trend offset rate of the risk factor data for adjacent windows, sets risk factor association rules according to the co-occurrence and time series correlation of the risk factors, analyzes the progressive influence chain of the risk factors, constructs a risk factor regression model, and establishes a risk factor influence gradient value;

[0061] The mutation factor monitoring module, according to the risk factor influence gradient value, detects the trajectory offset direction of abnormal data by monitoring the fluctuation range of abnormal data in the risk factor data, predicts the risk trigger time point, and generates a mutation factor trigger time point value;

[0062] The risk cross-derivation module calls the risk factor influence gradient value, calculates the parameter offset rate difference, identifies the candidate positions of risk cross-points, analyzes the continuity of the cross-point trend through the deviation degree between the data trend fitting residual and the current cross-point trend, and generates the scope value affected by the risk cross-point;

[0063] The prediction optimization and adjustment module calls the triggering time point value of the mutation factor and the scope value affected by the risk cross-point, combines the deviation between the actual risk event backtracking data and the prediction result, calculates the prediction error deviation trend, adjusts the model parameters, including the weight coefficients of each risk factor, and generates the risk prediction correction value.

[0064] The data time series offset value includes the time stamp difference distribution information, the alignment reference point position, and the segment density offset. The risk factor influence gradient value includes the factor fluctuation rate, the trend offset rate, and the association rule group. The triggering time point value of the mutation factor is specifically the abnormal fluctuation interval, the trajectory offset direction, and the triggering prediction time. The scope value affected by the risk cross-point specifically refers to the rate difference, the trend deviation degree, and the trend continuity level. The risk prediction correction value includes the error deviation trend, the risk weight adjustment value, and the model parameter correction coefficient.

[0065] Please refer to Figure 2 and Figure 3 , the time offset correction module includes:

[0066] The time stamp extraction sub-module collects the time stamp fields of multi-source data, extracts the time stamps of data points within each data segment, records the start time, end time, and time interval of the time stamps in each data segment, and obtains the time stamp interval record value;

[0067] The time stamp extraction sub-module collects the time stamps of data points in each segment of multi-source data, obtains the time stamp fields of each data source, and extracts the time information of each data segment, and records the time stamp value of each data point one by one. For each data segment, record its start time and end time of the time stamp, and calculate the time interval. Taking bank transaction data, online credit platform data, and third-party payment platform data as examples, the data segments of each platform contain continuous data points. For bank transaction data, if a segment contains 200 transaction data points, its start time is the time stamp of the first data point, and the end time is the time stamp of the last data point. Call the start time and end time of the data segment to calculate the time interval and calculate the time span for subsequent data alignment and time series correction operations. Using the formula:

[0068] ΔT = T end - T start ;

[0069] Calculate the time interval, where ΔT is the time interval of the time segment, T end is the end time of the data segment, Tstart is the start time of the data segment.

[0070] Set T start = 12:00:00, T end = 12:03:19, substitute the set values for calculation:

[0071] ΔT = 12:03:19 - 12:00:00 = 199 seconds;

[0072] The calculation result shows that the duration of the data segment is 199 seconds, and this time span is used for subsequent distribution density calculation and data alignment processing. After obtaining the start time, end time, and interval time of each data segment, timestamp interval records are made to ensure the integrity of subsequent calculations.

[0073] Based on the timestamp interval record values, the data alignment determination sub-module obtains the data point distribution density of each data segment, calculates the distribution density difference between adjacent segments, and sets the position of the reference alignment point according to the distribution density difference;

[0074] The data alignment determination sub-module calls the above timestamp interval record values, calculates the data point distribution density of each data segment, and evaluates the data distribution characteristics between different data segments. Obtain the total number of data points within the data segment and calculate the distribution density of each segment. The calculation method of density is the ratio of the total number of data points to the time interval. According to the density characteristics of the data segment, calculate the density difference between adjacent segments, and determine the position of the reference alignment point according to the density difference for subsequent time series correction. Use the formula:

[0075]

[0076] Calculate the distribution density of the data points, where ρ is the data distribution density, N is the total number of data points within the data segment, and ΔT is the time interval of the data segment.

[0077] Set N = 200, ΔT = 199, substitute the set values for calculation:

[0078]

[0079] The calculation result shows that the distribution density of the data segment is 1.005 data points per second, and this value is used to calculate the density difference to determine the data alignment point. Call the calculated data distribution density, calculate the density difference between adjacent data segments, set the density difference threshold, and the data segments with density differences lower than this threshold will be selected as the reference alignment points to ensure the stability and consistency of data alignment.

[0080] The timing correction sub-module calls the position of the reference alignment point, analyzes the offset between the timestamp of each data segment and the reference alignment point, corrects the timestamps of the data points in each segment according to the difference in the timestamp intervals, and establishes the data timing offset value;

[0081] The timing correction sub-module calls the position of the reference alignment point, calculates the time offset of the data segment, and performs the timing correction operation. The time offset calculation is based on the difference between the end time of the data segment and the timestamp of the reference alignment point. For different data segments, their respective time offsets are calculated, and this offset is applied to correct the timestamps. By adjusting the timestamps, the times of all data segments are aligned, ensuring data synchronization and reducing data errors caused by time misalignment. Using the formula:

[0082] ΔT offset =T end -T align ;

[0083] Calculate the time offset, where ΔT offset is the time offset, T end is the end time of the data segment, and T align is the position of the reference alignment point.

[0084] Set T end =12:03:09, T align =12:03:19, substitute the set values for calculation:

[0085] ΔT offset =12:03:09 - 12:03:19= - 10 seconds;

[0086] The calculation result shows that the time offset is - 10 seconds, indicating that the timestamps of the data segment are advanced by 10 seconds as a whole. Call the calculated offset to correct all the timestamps of the data segment. Using the correction formula:

[0087] T corrected =T original +|ΔT offset |;

[0088] Among them, T corrected is the corrected timestamp, T original is the original data point timestamp, and |ΔT offset | is the absolute value of the time offset.

[0089] Set T original =12:00:05, ΔT offset = - 10, substitute the set values for calculation:

[0090] T corrected= 12:00:05 + 10 = 12:00:15;

[0091] The calculation results show that the timestamp is adjusted to 12:00:15 after correction to ensure the consistency of the time series. After correcting the timestamps of all data segments, the data time series offset value is finally established to provide accurate data input and ensure the stability of subsequent analysis.

[0092] Please refer to Figure 2 and Figure 4 , the risk factor modeling module includes:

[0093] Based on the data time series offset value, the fluctuation feature extraction sub-module obtains the risk factor data within each time window, detects the numerical change range of the risk factor and the change rate of adjacent data points in each time window, calculates the change gradient of the data points within each time window, and obtains the risk factor fluctuation rate value;

[0094] Based on the data time series offset value, the fluctuation feature extraction sub-module obtains the risk factor data within each time window and extracts the change characteristics of the risk factor data. For the financial risk control scenario, each time window can be set to one day, one hour or a shorter time period, which is divided according to business requirements. After obtaining the data within the time window, calculate the numerical change range of the risk factor in each time window, which can be calculated by the difference between the maximum value and the minimum value and is used to characterize the fluctuation within the time window. Further extract the change rate between adjacent data points, and the calculation method is the ratio of the numerical change amount between adjacent data points to the time interval, and this rate is used to evaluate the short-term fluctuation of the risk factor. Finally, calculate the change gradient of all data points within the time window, which can be used to identify the data change trend and establish the risk factor fluctuation rate value. Using the formula:

[0095]

[0096] where G BW is the change gradient, |X q+1 -X q | represents the numerical change amount between adjacent data points, N is the total number of data points within the time window, and ΔT is the total duration of the time window.

[0097] Suppose there are N = 5 data points within a certain time window, with values [10, 15, 13, 18, 20] respectively, and the total duration of the time window ΔT = 60 seconds. Calculate the change gradient:

[0098]

[0099] The calculation results show that the risk factor fluctuation rate value within this time window is 0.058, and the risk factor fluctuation rate value is obtained.

[0100] The trend deviation calculation sub-module calls the risk factor fluctuation rate value, calculates the difference in the change rate of risk factor data between adjacent windows, analyzes the change trend of risk factors in the time dimension, and obtains the risk factor trend deviation value;

[0101] The trend deviation calculation sub-module calls the risk factor fluctuation rate value, calculates the difference in the change rate of risk factor data between adjacent time windows, and analyzes the change trend in the time dimension. In financial risk control applications, if the financial health of an enterprise is used as a risk factor, the change rate of financial data within different time windows is calculated, and the rate change between adjacent time windows is compared. The calculation method of the change rate between adjacent windows is the difference between the risk factor change rates of the previous window and the next window, which is used to characterize the trend change direction of risk factors. For a certain risk factor, within the time window T m -1 and T m calculate the difference in the change rate and use this difference to construct the trend deviation value. Using the formula:

[0102] D TR =G BW,m -G BW,m-1 ;

[0103] where D TR is the trend deviation value, G BW,m is the change gradient of the current time window, and G BW,m-1 is the change gradient of the previous time window.

[0104] Set G BW,m =0.058, G BW,m-1 =0.040, calculate the trend deviation value:

[0105] D TR =0.058 - 0.040 = 0.018;

[0106] The calculation result shows that the trend deviation value of this risk factor is 0.018, indicating that the risk fluctuation in the current window has increased compared with the previous window. Obtain the risk factor trend deviation value.

[0107] The regression model construction sub-module calls the risk factor trend deviation value, sets the association rules between risk factors by analyzing the co-occurrence characteristics and time-series correlation characteristics of multiple risk factors, combines the trend change amplitude of risk factors, analyzes the progressive influence chain, constructs a risk factor regression model, and obtains the risk factor influence gradient value;

[0108] The regression model construction sub-module calls the risk factor trend deviation value, sets the association rules between risk factors by analyzing the co-occurrence characteristics and time-series correlation characteristics of various risk factors. For the credit risk assessment in the financial industry, AI is used to extract the correlation between multiple factors, such as corporate cash flow, debt ratio, accounts receivable, etc. These factors may be correlated in time series. Therefore, when calculating the trend changes of each risk factor, it is necessary to combine its co-occurrence situation and analyze the progressive influence between different risk factors. The amplitude of the trend change of the risk factor can be used as a weight input to measure the contribution of each factor to the overall risk impact and is used to calculate the influence gradient of the risk factor. The formula is as follows:

[0109]

[0110] Where W FI is the weight of the i-th risk factor, |D TR,i | is the trend deviation value of this risk factor, is the sum of the trend deviation values of all risk factors, and M is the total number of risk factors.

[0111] Suppose the trend deviation values of three risk factors are D TR,1 = 0.018, D TR,2 = 0.025, D TR,3 = 0.015, calculate the weights of each factor:

[0112]

[0113] The calculation results show that the influence of risk factor 2 is relatively large, accounting for about 43.1%, and risk factors 1 and 3 account for 31.0% and 25.9% respectively. Based on the calculated weight values, a risk factor regression model is constructed to obtain the risk factor influence gradient value.

[0114] Please refer to Figure 2 and Figure 5 , the mutation factor monitoring module includes:

[0115] The outlier monitoring sub-module, based on the risk factor influence gradient value, detects the change amplitude of each data point by analyzing the numerical fluctuations in the risk factor data, calculates the deviation degree of each data point, detects outlier data points, and obtains the outlier data deviation value;

[0116] The abnormal point monitoring sub-module, based on the impact gradient value of risk factors, extracts the numerical value of each data point and calculates its fluctuation amplitude by analyzing the numerical fluctuation in the risk factor data. First, the numerical value of each data point is obtained and its time position is recorded. Subsequently, the difference between adjacent data points is calculated to form a data fluctuation amplitude matrix, which is used to depict the fluctuation state of risk factors at each time point. For each numerical value in this matrix, the deviation from the average value within the current time window is further calculated using the formula:

[0117]

[0118] where PD ip is the deviation degree of the ip-th data point, X ip is the numerical value of the ip-th data point, μ is the mean value within the current time window, and σ is the standard deviation within the current time window.

[0119] Suppose the numerical data within a certain time window is [20, 22, 18, 21, 23], with its mean μ = 20.8 and standard deviation σ ≈ 1.79. Calculate the deviation degree of the first data point:

[0120]

[0121] Calculate the deviation degree for each data point in turn and determine whether the deviation degree exceeds the preset deviation threshold. For example, set the deviation threshold to 1.5. If the deviation degree exceeds the threshold, mark this data point as an abnormal point and obtain the abnormal data deviation value.

[0122] The trajectory analysis sub-module calls the abnormal data deviation value, obtains the time series information of the abnormal data points, calculates the change direction of the abnormal data between adjacent time points, analyzes the change trend of the abnormal data on the time axis, extracts the continuous offset path of the data change trend, and obtains the abnormal data trajectory offset value;

[0123] The trajectory analysis sub-module calls the abnormal data deviation value, extracts the time series information of each abnormal data point, including the time point identifier, the numerical value of the data point, and its corresponding deviation degree. By analyzing the change of abnormal data between adjacent time points, calculate the numerical change direction of each abnormal data point compared to the previous data point, and use the change direction to track the offset trajectory of the abnormal data on the time axis. The calculation method of the change direction is the ratio of the numerical change amount between adjacent data points to the time interval, using the formula:

[0124]

[0125] where D ip is the change direction value of the ip-th data point, X ip+1 is the numerical value of the (ip + 1)-th data point, X ipis the value of the ip-th data point, and ΔT is the time interval between two points.

[0126] Assume that the values of the abnormal data points are [25, 28, 30, 33, 40], and the time interval between two points is 5 seconds. Calculate the first change direction value as:

[0127]

[0128] Calculate all the change direction values in sequence, and select the data points with continuous change directions and relatively large deviation amplitudes. Further analyze the fluctuation trend between the data points, mark the trend change path as the data offset path, and obtain the abnormal data trajectory offset value.

[0129] The time point prediction sub-module calls the abnormal data trajectory offset value, calculates the trajectory extension rate by analyzing the trajectory change trend of the abnormal data, analyzes the trend change amplitude, predicts the risk trigger time point, and obtains the mutation factor trigger time point value;

[0130] The specific formula for calculating the trajectory extension rate is:

[0131]

[0132] Calculate the trajectory extension rate;

[0133] Among them, V TRP represents the trajectory extension rate, X end represents the value of the trajectory end point, X start represents the value of the trajectory start point, X i′ represents the value of the i'-th data point, X i′+1 represents the value of the (i'+1)-th data point, n' represents the total number of abnormal data points in the trajectory, T end represents the time of the trajectory end point, T start represents the time of the trajectory start point, ΔT i′ represents the time interval between adjacent data points, n' represents the total number of abnormal data points in the time series, i' represents the index number of the abnormal data point in the time series, and i'+1 represents the index of the next data point of the current data point i' in the trajectory sequence.

[0134] Formula:

[0135]

[0136] Detailed explanation of the formula and the formula calculation derivation process:

[0137] The formula is used to calculate the trajectory weighted extension rate, and the obtained result is used to predict the trajectory development trend of the abnormal data point and further deduce the trigger time point value of the mutation factor.

[0138] Parameter meanings and setting values:

[0139] X end The value representing the trajectory end point, set to 85, reflecting the value of the abnormal data point at the trajectory end point;

[0140] X start The value representing the trajectory start point, set to 30, reflecting the value of the abnormal data point at the trajectory start point;

[0141] X i′ The value representing the i'-th data point, set to [35, 50, 65, 75], reflecting the value of the abnormal data point in the trajectory;

[0142] n' represents the total number of abnormal data points in the trajectory, set to 4, reflecting the number of abnormal points contained in the data trajectory in the time series;

[0143] T end The time representing the trajectory end point, set to 25 seconds, reflecting the time stamp when the abnormal data point reaches the end point;

[0144] T start The time representing the trajectory start point, set to 5 seconds, reflecting the initial time stamp of the abnormal data point;

[0145] ΔT i′ Represents the time interval between adjacent data points, set to [4, 6, 5, 5], reflecting the time change interval of the data points on the time axis;

[0146] Represents the cumulative change of all abnormal data points in the trajectory,

[0147] Represents the square root of the sum of squares of all time intervals in the trajectory,

[0148] Substitute the parameters into the formula for calculation:

[0149]

[0150] Analysis of the calculation results:

[0151] The calculation result of 2.74 represents the weighted extension rate of abnormal data points in the trajectory. This value reflects the growth trend of the abnormal data trajectory and is used to speculate on the future development direction of the data trajectory. A higher value indicates a faster rate of change in the trajectory of abnormal data points, which may trigger mutation factors in a shorter period of time, while a lower value indicates a slower rate of change in the trajectory, and mutation factors may be triggered after a longer period of time. The result will be further used to calculate the triggering time point value of the mutation factor and identify the time nodes at which risks occur.

[0152] Please refer to Figure 2 and Figure 6 , the risk cross-derivation module includes:

[0153] The parameter offset calculation sub-module, based on the influence gradient value of risk factors, collects the change data of risk factors in a continuous time series, calculates the parameter change rate of each risk factor between adjacent time nodes, obtains the change difference of the parameter rates between risk factors, and establishes the parameter rate offset difference.

[0154] The parameter offset calculation sub-module, based on the influence gradient value of risk factors, collects the change data of risk factors in a continuous time series, extracts the values of risk factors at different time nodes, and calculates the parameter change rate between adjacent time points. By calculating the change rate of risk factor values within each time window, a risk factor rate sequence is constructed. For the change rates of risk factors in adjacent time windows, calculate the difference in change rates within adjacent time windows, using the formula:

[0155] ΔS pq =S pq+1 -S pq ;

[0156] where, ΔS pq is the parameter change rate difference within the pq-th time window, S pq+1 is the change rate of the pq + 1-th time window, and S pq is the change rate of the pq-th time window.

[0157] Suppose the change rates of a certain risk factor in three consecutive time windows are 2.5, 3.0, and 4.2 respectively. Calculate the rate difference between the first time window and the second time window:

[0158] ΔS1=3.0 - 2.5=0.5;

[0159] Calculate the rate change values of all adjacent time windows in sequence, extract the time windows with larger change rates, screen out the key change nodes of risk factors, and establish the parameter rate offset difference.

[0160] The intersection point recognition sub-module calls the parameter rate offset difference, predicts the intersection area position of the risk factor parameter change path according to the set rate offset threshold, and establishes a risk intersection candidate area;

[0161] The intersection point recognition sub-module calls the parameter rate offset difference, analyzes the change patterns of the risk factor change rates within all time windows, and based on the set rate offset threshold, selects the time windows where the rate changes exceed the threshold. For all selected time windows, further analyze the risk factor parameter change paths and find the intersection area positions of different risk factor parameter change paths. The determination of the intersection area is based on the intersection points of multiple risk factor change paths, using the formula:

[0162]

[0163] where, P int is the intersection point position of the risk factor change path, S A and S B are the change rates of two risk factors within the intersection area respectively, T A and T B are the corresponding time intervals.

[0164] Suppose the change rates of two certain risk factors within the intersection area are 2.8 and 3.4 respectively, and the corresponding time intervals are 6 seconds and 4 seconds respectively, calculate the intersection point position:

[0165]

[0166] Select the time windows with relatively stable intersection point position values and establish a risk intersection candidate area.

[0167] The trend continuity analysis sub-module calls the risk intersection candidate area, obtains the parameter trend data within the intersection area, calculates the trend fitting residual value and the trend deviation amplitude, analyzes the continuity of the intersection point trend, and establishes the risk intersection point influence range value;

[0168] The trend continuity analysis sub-module calls the risk intersection candidate area, analyzes the parameter trend data of all risk factors within the intersection area, and calculates the trend fitting residual value. The trend fitting residual value is calculated based on the historical trend data, using the formula:

[0169]

[0170] where, R fit is the trend fitting residual, X ia is the actual value of the ia-th data point, X fit is the fitting value of this point, and na is the total number of data points.

[0171] Assume that the risk factor data points in the intersection area are [5.2, 4.8, 5.1, 4.9, 5.0] respectively, and the fitted data point values are [5.0, 5.0, 5.0, 5.0, 5.0]. Calculate the trend fitting residuals:

[0172]

[0173] Further calculate the trend deviation amplitude and analyze the continuity of the intersection point trend, and finally establish the value of the risk intersection point affected range.

[0174] Please refer to Figure 2 and Figure 7 , the prediction optimization and adjustment module includes:

[0175] The error calculation sub-module calculates the time deviation between the predicted risk trigger time point and the actual risk event occurrence time based on the mutation factor trigger time point value and the risk intersection point affected range value, and establishes a predicted error time deviation value;

[0176] The error calculation sub-module collects the predicted risk trigger time point and the actual risk event occurrence time based on the mutation factor trigger time point value and the risk intersection point affected range value, and calculates the time deviation. Extract the mutation factor trigger time point value, determine the time position corresponding to the mutation factor, obtain the risk intersection point affected range value, screen the key time points in the intersection area, arrange all the time points in chronological order, and calculate the time difference between the predicted time point and the actual event occurrence time, using the formula:

[0177] ΔT = T pred - T actual ;

[0178] where, ΔT is the predicted error time deviation value, T pred is the predicted risk trigger time point, and T actual is the actual risk event occurrence time.

[0179] Assume that the mutation factor trigger time point value of a certain mutation factor is 12.5 seconds, and the actual event occurrence time within the risk intersection point affected range is 10.8 seconds. Then the predicted error is calculated as follows:

[0180] ΔT = 12.5 - 10.8 = 1.7 seconds;

[0181] Calculate the error values of all predicted time points and establish a predicted error time deviation value.

[0182] The change trend analysis sub-module calls the predicted error time deviation value, obtains the change data of the error value within the continuous time window, calculates the trend rate of the error change, and analyzes the error change direction, and establishes an error change trend value;

[0183] The change trend analysis sub-module calls the predicted error time deviation value, obtains the change data of the error value within a continuous time window, calculates the trend rate of the error change, and analyzes the direction of the error change. For the error values of multiple time windows, calculate the change rate of the error values within adjacent time windows, and construct an error change trend sequence. The error change trend rate adopts the formula:

[0184]

[0185] Among them, V err is the error change trend rate, ΔT ib+1 and ΔT ib are the error values of two adjacent time windows respectively, and T w is the time window size.

[0186] Suppose the error values within the time windows T1 and T2 are 1.7 seconds and 2.4 seconds respectively, and the window size is 5 seconds, then the error change trend rate is calculated as follows:

[0187]

[0188] Calculate the error change trend rate of all time windows and establish the error change trend value.

[0189] The parameter adjustment sub-module calls the error change trend value, and through iterative adjustment, optimizes multiple parameters of the risk factor regression model, including the weight coefficient of each risk factor, and establishes a risk prediction correction value;

[0190] The specific formula for optimizing multiple parameters of the risk factor regression model is:

[0191]

[0192] Calculate the adjusted risk factor weight;

[0193] Among them, W new represents the adjusted risk factor weight, W old represents the original risk factor weight, k represents the risk adjustment factor, V err represents the error change trend rate, V base represents the benchmark change rate of the risk factor, and ΔT err represents the error change trend value.

[0194] Formula:

[0195]

[0196] Formula details and formula calculation derivation process:

[0197] The formula is used to calculate the adjusted risk factor weight, and the obtained result is used to optimize the parameter adjustment in the risk factor regression model;

[0198] Parameter meanings and set values:

[0199] W new is the adjusted risk factor weight, used to correct the risk assessment ability of the model;

[0200] W old is the original risk factor weight, indicating the relative importance of the risk factor in the model calculation before error correction, and the set value is 0.75;

[0201] k is the risk adjustment factor, and the set value is 0.3;

[0202] V err is the error change trend rate, indicating the change rate of the prediction error. Set V err = 0.5;

[0203] V base is the benchmark change rate of the risk factor, indicating the standard error change rate under the condition of no additional influence, and the set value is 0.25;

[0204] ΔT err is the error change trend value, that is, the change trend of the error value within a continuous time window. Set ΔT err = 0.0625;

[0205] Substitute the parameters into the formula for calculation:

[0206]

[0207] W new = 0.75×(1 - 0.3×0.2427);

[0208] W new = 0.75×(1 - 0.0728);

[0209] W new = 0.75×0.9272 = 0.6954;

[0210] The calculation result W new = 0.6954 indicates the adjustment degree of the weight factor of the risk factor after error correction, which is reduced from the original 0.75 to 0.6954. This value is used to correct the weight parameter of the risk factor in the risk assessment model to optimize the overall assessment accuracy and adjust the importance distribution of each risk factor.

[0211] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions according to the embodiments of the present invention are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (such as infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more collections of available media. The available media can be magnetic media (such as floppy disks, hard disks, magnetic tapes), optical media (such as DVDs), or semiconductor media. The semiconductor media can be a solid-state drive.

[0212] It should be understood that the term "and / or" in this document is merely a description of the association relationship between associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. Here, A and B can be singular or plural. In addition, the character " / " in this document generally represents an "or" relationship between the associated objects before and after, but it may also represent an "and / or" relationship, which can be specifically understood with reference to the context.

[0213] In the present invention, "at least one" means one or more, and "a plurality" means two or more. "At least one of the following" or its similar expressions refer to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, or c can represent: a, b, c, a - b, a - c, b - c, or a - b - c, where a, b, and c can be single or multiple.

[0214] It should be understood that in various embodiments of the present invention, the magnitudes of the sequence numbers of the above processes do not mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0215] Those of ordinary skill in the art will realize that the units and algorithm steps of each example described in connection with the embodiments disclosed herein can be implemented in electronic hardware, or in a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Skilled professionals can use different methods for each specific application to implement the described functions, but such implementation should not be considered to exceed the scope of the present invention.

[0216] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the devices, apparatuses, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0217] In several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0218] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place, or can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0219] In addition, the functional units in each embodiment of the present invention can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0220] When the above-described function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or a part of this technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0221] As described above, the above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. An AI-based big data risk assessment system, characterized in that, The system includes: The time offset correction module collects the timestamp fields of multi-source data, identifies the positions of reference alignment data points by comparing the difference in the distribution density of adjacent segment data points, corrects the data source timestamp sequence, and establishes a data timing offset value; The risk factor modeling module, based on the data timing offset value, extracts the fluctuation rate of the risk factor data for each time window, calculates the trend offset rate of the risk factor data between adjacent windows, sets risk factor association rules according to the co-occurrence and timing correlation of risk factors, analyzes the progressive influence chain of risk factors, constructs a risk factor regression model, and establishes a risk factor influence gradient value; The mutation factor monitoring module, according to the risk factor influence gradient value, detects the fluctuation range of abnormal data in the risk factor data, detects the trajectory offset direction of the abnormal data, predicts the risk trigger time point, and generates a mutation factor trigger time point value; The risk cross-derivation module calls the risk factor influence gradient value, calculates the difference in the parameter offset rate, identifies the candidate positions of risk cross-points, and analyzes the continuity of the cross-point trend through the deviation degree between the data trend fitting residual and the current cross-point trend, and generates a risk cross-point spread range value.

2. The AI-based big data risk assessment system according to claim 1, wherein, The data timing offset value includes timestamp difference distribution information, alignment reference point position, and segment density offset amount. The risk factor influence gradient value includes factor fluctuation rate, trend offset rate, and association rule group. The mutation factor trigger time point value specifically refers to the abnormal fluctuation interval, trajectory offset direction, and trigger prediction time. The risk cross-point spread range value specifically refers to the rate difference, trend deviation degree, and trend continuity level.

3. The AI-based big data risk assessment system according to claim 1, characterized in that, The time offset correction module includes: The timestamp extraction sub-module collects the timestamp fields of multi-source data, extracts the data point timestamps within each data segment, records the start time, end time, and time interval of the timestamps in each data segment, and obtains the timestamp interval record value; The data alignment determination sub-module, based on the timestamp interval record value, obtains the data point distribution density of each data segment, calculates the difference in the distribution density between adjacent segments, and sets the position of the reference alignment point according to the distribution density difference; The timing correction sub-module calls the position of the reference alignment point, analyzes the offset of the timestamp of each data segment from the reference alignment point, and corrects the data point timestamp of each segment according to the timestamp interval difference to establish a data timing offset value.

4. The AI-based big data risk assessment system according to claim 1, characterized in that, The risk factor modeling module includes: The fluctuation feature extraction sub-module, based on the data timing offset value, obtains the risk factor data within each time window, detects the numerical change amplitude of the risk factor and the change rate of adjacent data points in each time window, calculates the change gradient of the data points within each time window, and obtains the risk factor fluctuation rate value; The trend offset calculation sub-module calls the risk factor fluctuation rate value, calculates the difference in the change rate of the risk factor data between adjacent windows, analyzes the change trend of the risk factor in the time dimension, and obtains the risk factor trend offset value; The regression model construction sub-module calls the risk factor trend deviation value, sets the association rules between risk factors by analyzing the co-occurrence characteristics and time-series correlation characteristics of multiple risk factors, combines the trend change range of risk factors, analyzes the progressive influence chain, constructs a risk factor regression model, and obtains the influence gradient value of risk factors.

5. The AI-based big data risk assessment system according to claim 1, characterized in that The mutation factor monitoring module includes: The outlier monitoring sub-module, based on the influence gradient value of the risk factor, detects the change range of each data point by analyzing the numerical fluctuation in the risk factor data, calculates the deviation degree of each data point, detects outlier data points, and obtains the outlier data deviation value; The trajectory analysis sub-module calls the outlier data deviation value, obtains the time-series information of outlier data points, calculates the change direction of outlier data between adjacent time points, analyzes the change trend of outlier data on the time axis, extracts the offset path with continuous data change trend, and obtains the outlier data trajectory deviation value; The time point prediction sub-module calls the outlier data trajectory deviation value, calculates the trajectory extension rate by analyzing the trajectory change trend of outlier data, analyzes the trend change range, predicts the risk trigger time point, and obtains the mutation factor trigger time point value.

6. The AI-based big data risk assessment system according to claim 5, wherein, The specific formula for calculating the trajectory extension rate is: Calculate the trajectory extension rate; Among them, V TRP represents the trajectory extension rate, X end represents the value of the trajectory termination point, X start represents the value of the trajectory starting point, X i′ represents the value of the i'-th data point, X i′+1 represents the value of the (i'+1)-th data point, n' represents the total number of abnormal data points in the trajectory, T end represents the time of the trajectory termination point, T start represents the time of the trajectory starting point, ΔT i′ represents the time interval between adjacent data points, n' represents the total number of abnormal data points in the time series, i' represents the index number of the abnormal data point in the time series, and i'+1 represents the index of the next data point of the current data point i' in the trajectory sequence.

7. The AI-based big data risk assessment system according to claim 1, wherein, The risk intersection derivation module includes: The parameter deviation calculation sub-module, based on the influence gradient value of the risk factor, collects the change data of risk factors in the continuous time series, calculates the parameter change rate of each risk factor between adjacent time nodes, obtains the change difference of the parameter rates between risk factors, and establishes the parameter rate deviation difference; The intersection point identification sub-module calls the parameter rate deviation difference, predicts the intersection area position of the risk factor parameter change path according to the set rate deviation threshold, and establishes a risk intersection candidate area; The trend continuity analysis sub-module calls the risk intersection candidate area, obtains the parameter trend data in the intersection area, calculates the trend fitting residual value and the trend deviation amplitude, analyzes the continuity of the intersection point trend, and establishes the risk intersection point impact range value.

8. The AI-based big data risk assessment system according to claim 1, characterized in that, The system further includes: The prediction optimization and adjustment module calls the mutation factor trigger time point value and the risk intersection point impact range value, combines the deviation between the actual risk event backtracking data and the prediction result, calculates the prediction error deviation trend, and adjusts the model parameters, including the weight coefficient of each risk factor, to generate a risk prediction correction value; The risk prediction correction value includes the error deviation trend, the risk weight adjustment value, and the model parameter correction coefficient.

9. The AI-based big data risk assessment system according to claim 1, wherein, The prediction optimization and adjustment module includes: The error calculation sub-module, based on the mutation factor trigger time point value and the risk intersection point impact range value, calculates the time deviation between the predicted risk trigger time point and the actual risk event occurrence time, and establishes the prediction error time deviation value; The change trend analysis sub-module calls the prediction error time deviation value, obtains the change data of the error value in the continuous time window, calculates the trend rate of the error change, and analyzes the error change direction, and establishes the error change trend value; The parameter adjustment sub-module calls the error change trend value, and through iterative adjustment, optimizes multiple parameters of the risk factor regression model, including the weight coefficient of each risk factor, and establishes a risk prediction correction value.

10. The AI-based big data risk assessment system according to claim 9, wherein, The specific formula for optimizing multiple parameters of the risk factor regression model is as follows: Calculate the adjusted risk factor weights; Among them, W new represents the adjusted risk factor weight, W old represents the original risk factor weight, k represents the risk adjustment factor, V err represents the error change trend rate, V base represents the benchmark change rate of the risk factor, ΔT err represents the error change trend value.

Images