Information asset prediction method and intelligent prediction system
By extracting features and modeling risks in asset datasets, an optimal model is constructed, which solves the problems of insufficient prediction accuracy and flexibility in existing technologies and achieves efficient prediction of information assets in complex structural design scenarios.
Patent Information
- Application Number
- CN202511609212.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-05
- Publication Date
- 2026-03-06
AI Technical Summary
Existing information asset forecasting methods lack the ability to respond to changes in asset structure, market dynamics, and historical extreme scenarios in complex structural design scenarios, resulting in insufficient forecasting accuracy and flexibility.
By extracting features from standardized asset datasets, performing risk segmentation and modeling, and analyzing historical performance, an optimal model is constructed. Cash flow forecasting is then performed using LSTM neural networks, XGBoost ensemble models, and rule expert systems. Finally, through model validation, optimization, and iterative adjustments, an information asset report is generated.
It significantly improves the accuracy and flexibility of predictions in complex structural design scenarios, enabling it to better respond to changes in asset structure and market dynamics.
Smart Images
Figure CN121616403A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of information asset prediction, and more particularly to an information asset prediction method and intelligent prediction system. Background Technology
[0002] In the field of information asset forecasting, current asset-backed securities (ABS) cash flow forecasting primarily relies on modeling methods based on static pool data. Specifically, this method typically uses overall asset pool data (such as loan balance, default rate, and recovery rate) at a historical time point to construct a unified cash flow forecasting model. During the modeling process, parameters such as average lag period, default probability, and recovery period are statistically analyzed to estimate the future repayment cash flow trend of the asset pool. This method is simple to operate, has low implementation costs, and is suitable for the rapid evaluation of standardized products.
[0003] However, the aforementioned methods neglect the behavioral differences between different segments (risk segments) at the bottom of the asset pool, lacking the ability to respond to changes in asset structure, market dynamics, and historical extreme scenarios. This results in significantly insufficient prediction accuracy and flexibility in scenarios with complex structural designs. Therefore, the industry needs to design an information asset prediction method and intelligent prediction system to solve the above-mentioned technical problems. Summary of the Invention
[0004] In a first aspect, this application proposes an information asset prediction method, which includes: S1, extracting features from a standardized asset dataset to obtain an effective feature set, and then performing risk segmentation and segmented modeling and historical performance analysis to obtain a historical performance feature set; S2, based on the effective feature set and the historical performance feature set, performing cash flow prediction and model validation optimization to obtain a preferred model; S3, generating an information asset report based on the preferred model.
[0005] The further technical solution is that step S1 includes: S11, extracting features from the standardized asset dataset to obtain an effective feature set; S12, performing risk segmentation and modeling based on the effective feature set to obtain a risk segmentation set; S13, performing historical performance analysis based on the risk segmentation set to obtain a historical performance feature set.
[0006] The further technical solution is that step S2 includes: S21, performing cash flow forecasting based on the effective feature set and the historical performance feature set to obtain the forecast result corresponding to the cash flow forecast; S22, performing model verification and optimization on the forecast result to obtain the preferred model.
[0007] The further technical solution is that step S11 includes: S41, extracting basic features from the standardized asset dataset to obtain a basic feature matrix; S42, constructing derived features based on the basic feature matrix to obtain a derived feature matrix, merging the basic feature matrix and the derived feature matrix to obtain an extended feature set; S43, calculating the information gain ratio of the extended feature set to obtain an information gain ratio value, filtering features with an information gain ratio greater than a preset threshold to obtain an effective feature set.
[0008] The further technical solution is as follows: Step S12 includes: S51, constructing a multi-dimensional segmentation space based on the effective feature set, and cross-combining the feature categories of each dimension to form an initial risk segmentation candidate set; S52, introducing a feature correlation matrix, calculating feature correlation using mutual information, and filtering highly correlated feature pairs according to a preset threshold to construct a simplified risk segmentation dimension set; S53, using an improved density peak clustering algorithm, segmenting asset samples based on the feature space weighted distance, automatically identifying cluster centers through local density and minimum distance, and outputting an initial risk segmentation set; S54, establishing a segmentation effectiveness evaluation function, quantitatively evaluating the initial risk segmentation set from the perspectives of intra-group similarity and inter-group difference, and obtaining an evaluation value; S55, if the evaluation value is lower than a preset optimization threshold, iteratively optimizing the initial risk segmentation set by merging similar segments and adjusting clustering parameters to obtain a risk segmentation set.
[0009] The further technical solution is as follows: Step S13 includes: S61, based on the risk segment set, selecting a subset of data from the historical asset pool database that meets the current segment conditions and is located within a preset historical time window; S62, calculating basic statistical indicators for the data subset of each risk segment to obtain the average repayment rate, default rate, and prepayment rate; S63, performing time series decomposition on the time series composed of the basic statistical indicators to obtain the trend term, seasonal term, and residual term, and using an anomaly detection algorithm to identify outlier fluctuations in the data subset to obtain anomaly quantitative scores; S64, integrating the average repayment rate, default rate, prepayment rate, trend term, seasonal term, residual term, and anomaly quantitative scores to form a segmented performance feature vector; S65, integrating the segmented performance feature vectors of all segments to obtain a historical performance feature set.
[0010] The further technical solution is as follows: Step S21 includes: S71, constructing a model input feature vector based on the effective feature set and the historical performance feature set; S72, for each segment, evaluating the characteristics of the segment based on the model input feature vector, and quantitatively calculating the evaluation value corresponding to the characteristic evaluation function from three dimensions: data richness, feature complexity, and historical volatility; S73, selecting a prediction model based on the evaluation value, wherein a high-quality segment uses an LSTM neural network, a medium-quality segment uses an XGBoost ensemble model, and a low-quality segment uses a rule expert system; S74, training the optimal parameters of the selected prediction model by minimizing the loss function, and then inputting the model input feature vector of the current segment to predict the cash flow sequence in the future time period, and obtaining the default amount and the repayment amount; S75, weighting and summing according to the proportion of the remaining principal of each segment, aggregating the predicted cash flow sequences of each segment, outputting the overall future cash flow sequence of the asset pool, and obtaining the prediction result.
[0011] The further technical solution is as follows: step S22 includes: S81, performing multi-period rolling backtesting based on historical data of a preset monthly period to generate a predicted cash flow sequence, comparing the predicted cash flow sequence with the actual cash flow period by period, and calculating the segmented prediction deviation; S82, calculating the mean square error and mean absolute error index based on the segmented prediction deviation to obtain the model performance evaluation result; S83, automatically capturing the objective trend indicators of the preset monthly period each month to detect whether the core parameters corresponding to the model performance evaluation result have shifted; S84, when a shift in the core parameters is detected, triggering the model reconstruction process, automatically adjusting the feature weights and model structure to obtain the optimal model.
[0012] In summary, existing information asset prediction methods lack the ability to respond to changes in asset structure, market dynamics, and historical extreme scenarios, resulting in insufficient prediction accuracy and flexibility in scenarios with complex structural designs. Therefore, the solution described in this application can significantly improve prediction accuracy and flexibility for information asset prediction in scenarios with complex structural designs. Attached Figure Description
[0013] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.
[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0015] Figure 1This is a flowchart of an information asset prediction method provided in an embodiment of the present invention.
[0016] Figure 2 Another flowchart of the information asset prediction method provided in this embodiment of the invention.
[0017] Figure 3 A simplified diagram of the electronic device provided in an embodiment of the present invention. Detailed Implementation
[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0019] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.
[0020] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.
[0021] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to one or any combination of the associated listed items and all possible combinations, and includes such combinations.
[0022] In this specification and the appended claims, there may be multiple ways of expressing the same technical feature or technical term, such as using a superordinate generalization, a subordinate limitation, or a synonym substitution. Those skilled in the art can clearly understand the substantially the same technical meaning referred to by different expressions based on their professional knowledge and in conjunction with the overall content of the specification and the drawings. The differences in different expressions are only reflected in the diversity of words and do not constitute a substantial modification or limitation to the technical solution, nor will they affect the certainty of the scope of protection of this patent claim or the full disclosure of the technical content of the specification.
[0023] Example 1 Please see Figures 1 to 2The image shows an information asset prediction method proposed by this invention. In a first aspect, the method includes: S1, extracting features from a standardized asset dataset to obtain an effective feature set, then performing risk segmentation and modeling, and historical performance analysis to obtain a historical performance feature set; S2, based on the effective feature set and the historical performance feature set, performing cash flow prediction and model validation optimization to obtain a preferred model; S3, generating an information asset report based on the preferred model.
[0024] The further technical solution is that step S1 includes: S11, extracting features from the standardized asset dataset to obtain an effective feature set; S12, performing risk segmentation and modeling based on the effective feature set to obtain a risk segmentation set; S13, performing historical performance analysis based on the risk segmentation set to obtain a historical performance feature set.
[0025] The further technical solution is that step S2 includes: S21, performing cash flow forecasting based on the effective feature set and the historical performance feature set to obtain the forecast result corresponding to the cash flow forecast; S22, performing model verification and optimization on the forecast result to obtain the preferred model.
[0026] A further technical solution is as follows: Step S11 includes: S41, extracting basic features from the standardized asset dataset to obtain a basic feature matrix; S42, constructing derived features based on the basic feature matrix to obtain a derived feature matrix, merging the basic feature matrix and the derived feature matrix to obtain an extended feature set; S43, calculating the information gain ratio of the extended feature set to obtain an information gain ratio value, filtering features with an information gain ratio greater than a preset threshold to obtain an effective feature set. In the above solution, the basic features include interest rate grouping features, principal grouping features, loan term features, initial aging features, initial status features, number of loan agreements features, and repayment method features.
[0027] The further technical solution is as follows: Step S12 includes: S51, constructing a multi-dimensional segmentation space based on the effective feature set, and cross-combining the feature categories of each dimension to form an initial risk segmentation candidate set; S52, introducing a feature correlation matrix, calculating feature correlation using mutual information, and filtering highly correlated feature pairs according to a preset threshold to construct a simplified risk segmentation dimension set; S53, using an improved density peak clustering algorithm, segmenting asset samples based on the feature space weighted distance, automatically identifying cluster centers through local density and minimum distance, and outputting an initial risk segmentation set; S54, establishing a segmentation effectiveness evaluation function, quantitatively evaluating the initial risk segmentation set from the perspectives of intra-group similarity and inter-group difference, and obtaining an evaluation value; S55, if the evaluation value is lower than a preset optimization threshold, iteratively optimizing the initial risk segmentation set by merging similar segments and adjusting clustering parameters to obtain a risk segmentation set. In the above scheme, the Segment corresponds to risk segmentation, that is, a multi-dimensional segmentation space is constructed based on the effective feature set, and the feature categories of each dimension are cross-combined to form an initial Segment candidate set.
[0028] The further technical solution is as follows: Step S13 includes: S61, based on the risk segment set, selecting a subset of data from the historical asset pool database that meets the current segment conditions and is located within a preset historical time window; S62, calculating basic statistical indicators for the data subset of each risk segment to obtain the average repayment rate, default rate, and prepayment rate; S63, performing time series decomposition on the time series composed of the basic statistical indicators to obtain the trend term, seasonal term, and residual term, and using an anomaly detection algorithm to identify outlier fluctuations in the data subset to obtain anomaly quantitative scores; S64, integrating the average repayment rate, default rate, prepayment rate, trend term, seasonal term, residual term, and anomaly quantitative scores to form a segmented performance feature vector; S65, integrating the segmented performance feature vectors of all segments to obtain a historical performance feature set.
[0029] In the above embodiments, the definitions of terms in the scheme are as follows: First, Average Collection Rate: The ratio of the total cash actually collected to the total amount receivable in a specific period (usually monthly or quarterly). Second, Default Rate: The proportion of loan balances that have entered default status during the observation period to the total loan balance at the beginning of the period. Third, Prepayment Rate: The proportion of loans that borrowers prepay part or all of their loans before the agreed repayment deadline. Fourth, Trend Term: In time series decomposition, it represents the long-term, unidirectional upward or downward basic direction of data. Fifth, Seasonal Term: In time series, it refers to predictable, regular fluctuations caused by seasonal factors (such as holidays, climate, production cycles). Sixth, Residual Term: In time series decomposition, after removing the trend and seasonal terms, the remaining "random noise" that cannot be systematically explained. It represents unexpected fluctuations that the model failed to capture. Seventh, Anomaly Quantification Score: A comprehensive, standardized score used to quantify the degree of "anomaly" or "rarity" of a data point or event relative to historical patterns; the higher the score, the more anomaly it represents.
[0030] The further technical solution is as follows: Step S21 includes: S71, constructing a model input feature vector based on the effective feature set and the historical performance feature set; S72, for each segment, evaluating the characteristics of the segment based on the model input feature vector, and quantitatively calculating the evaluation value corresponding to the characteristic evaluation function from three dimensions: data richness, feature complexity, and historical volatility; S73, selecting a prediction model based on the evaluation value, wherein a high-quality segment uses an LSTM neural network, a medium-quality segment uses an XGBoost ensemble model, and a low-quality segment uses a rule expert system; S74, training the optimal parameters of the selected prediction model by minimizing the loss function, and then inputting the model input feature vector of the current segment to predict the cash flow sequence in the future time period, and obtaining the default amount and the repayment amount; S75, weighting and summing according to the proportion of the remaining principal of each segment, aggregating the predicted cash flow sequences of each segment, outputting the overall future cash flow sequence of the asset pool, and obtaining the prediction result.
[0031]
[0032] Referring to Formula E above, step S72 evaluates the characteristics of each segment based on the model input feature vector, quantifying the evaluation value corresponding to the characteristic evaluation function from three dimensions: first, data richness; second, feature complexity; and third, historical volatility. Furthermore, the evaluation is quantified from these three dimensions: data richness, feature complexity, and historical volatility.
[0033]
[0034] Then, according to the relationship between the evaluation value and θ1, and the relationship between the evaluation value and θ2, a prediction model is selected. Among them, the LSTM neural network is used for high-quality segments, the XGBoost integration model is used for medium-quality segments, and the rule-based expert system is used for low-quality segments.
[0035] (1) When E(Si(t)) > θ2 (high data quality threshold), the LSTM neural network model is selected.
[0036] (2) When θ1 < E(Si(t)) ≤ θ2, the XGBoost integrated learning model is selected.
[0037] (3) When E(Si(t)) ≤ θ1 (low data quality threshold), the rule-based expert system model is selected.
[0038] Furthermore, according to the above formula E, it is calculated whether it is greater than or less than the set thresholds θ1 or θ2. Generally, a large amount of data and a long time span (sample size > 5,000 records and time span > 24 months) are high-quality data; medium data volume, insignificant time series characteristics or certain volatility are medium-quality data; sparse data and newly emerging Segments or extreme cases are low-quality data.
[0039] Its further technical solution is that the step S22 includes: S81, performing multi-period rolling backtesting based on historical data of a preset monthly cycle to generate a predicted cash flow sequence, comparing the predicted cash flow sequence with the actual cash flow period by period, and statistically segmenting the prediction deviation; S82, calculating the mean square error and mean absolute error indicators based on the segmented prediction deviation to obtain the model performance evaluation result; S83, automatically capturing the objective trend indicators of the past preset monthly cycle every month to detect whether the core parameters corresponding to the model performance evaluation result deviate; S84, when it is detected that the core parameters deviate, triggering the model reconstruction process, automatically adjusting the feature weights and model structure to obtain an optimized model. In the above solution, multi-period rolling backtesting is performed based on historical data of a preset monthly cycle. The preset monthly cycle can be 6 months or 12 months or 24 months, or any integer multiple of months, and engineers in this field can understand and set it by themselves.
[0040] In summary, current asset securitization (ABS) cash flow forecasting primarily relies on modeling methods based on static pool data in the field of information asset forecasting. Specifically, this method typically uses overall asset pool data (such as loan balance, default rate, and recovery rate) at a historical time point to construct a unified cash flow forecasting model. During modeling, parameters such as average lag period, default probability, and recovery period are used to estimate the future repayment cash flow trend of the asset pool. This method is simple to operate, has low implementation costs, and is suitable for the rapid evaluation of standardized products. However, this method ignores the behavioral differences between different segments (risk segments) at the bottom of the asset pool and lacks responsiveness to changes in asset structure, market dynamics, and historical extreme scenarios. This results in significantly insufficient forecasting accuracy and flexibility in scenarios with complex structures. Therefore, the solution described in this application, used for information asset forecasting in scenarios with complex structures, can significantly improve forecasting accuracy and flexibility.
[0041] Example 2 Please see Figure 3 , Figure 3 This invention provides a block diagram of an electronic device. The electronic device can be a terminal or a server. The terminal can be a smartphone, tablet computer, laptop computer, desktop computer, personal digital assistant, wearable device, or other electronic device with communication capabilities. It includes a processor 111, a communication interface 112, a memory 113, and a communication bus 114. The processor 111, communication interface 112, and memory 113 communicate with each other via the communication bus 114.
[0042] Memory 113 is used to store computer programs.
[0043] In one embodiment of the present invention, the processor 111, when executing the program stored in the memory 113, implements the method provided in any of the foregoing method embodiments.
[0044] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program may be stored in a storage medium, which is a computer-readable storage medium. The computer program is executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.
[0045] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions, but such implementations should not be considered beyond the scope of this invention.
[0046] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of each unit is only a logical functional division, and there may be other division methods in actual implementation. For example, units or components may be combined or integrated into another system, or some features may be ignored or not executed.
[0047] The steps in the method of this invention can be adjusted, merged, or reduced in order according to actual needs. The units in the device of this invention can be merged, divided, or reduced according to actual needs. Furthermore, the functional units in the various embodiments of this invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.
[0048] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a 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 all or part of the 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 to cause a computer device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.
[0049] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.
[0050] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Since these modifications and variations fall within the scope of the claims and their equivalents, this invention also intends to include these modifications and variations.
[0051] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.
Claims
1. An information asset prediction method characterized by, The information asset prediction method comprises the following steps: S1, feature extraction is performed on the standardized asset dataset to obtain an effective feature set, and then risk segmentation modeling and historical performance analysis are performed to obtain a historical performance feature set; S2, based on the effective feature set and the historical performance feature set, cash flow prediction and model verification optimization are performed to obtain an optimized model; S3, an information asset report is generated according to the optimized model.
2. The information asset prediction method of claim 1, wherein, The step S1 comprises: S11, feature extraction is performed on the standardized asset dataset to obtain an effective feature set; S12, risk segmentation modeling is performed based on the effective feature set to obtain a risk segmentation set; S13, historical performance analysis is performed based on the risk segmentation set to obtain a historical performance feature set.
3. The information asset forecasting method of claim 2, wherein, The step S2 comprises: S21, cash flow prediction is performed based on the effective feature set and the historical performance feature set to obtain a prediction result corresponding to the cash flow prediction; S22, the prediction result is subjected to model verification optimization to obtain an optimized model.
4. The information asset prediction method of claim 3, wherein, The step S11 comprises: S41, basic features are extracted from the standardized asset dataset to obtain a basic feature matrix; S42, derived features are constructed based on the basic feature matrix to obtain a derived feature matrix, and the basic feature matrix and the derived feature matrix are combined to obtain an expanded feature set; S43, information gain ratios of the expanded feature set are calculated to obtain information gain ratio values, and features with information gain ratios greater than a preset threshold value are screened to obtain an effective feature set.
5. The information asset prediction method of claim 4, wherein, The step S12 comprises: S51, a multi-dimensional segmentation space is constructed based on the effective feature set, and cross combinations are performed on feature categories of each dimension to form an initial risk segmentation candidate set; S52, a feature correlation matrix is introduced, feature correlation is calculated using mutual information, and high-correlation feature pairs are screened according to a preset threshold value to construct a simplified risk segmentation dimension set; S53, an improved density peak clustering algorithm is adopted to segment asset samples based on feature space weighted distances, clustering centers are automatically identified from local density and minimum distance, and an initial risk segmentation set is output; S54, a segmentation effectiveness evaluation function is established, and the initial risk segmentation set is quantitatively evaluated from two aspects of intragroup similarity and intergroup difference to obtain an evaluation value; S55, if the evaluation value is lower than a preset optimization threshold value, the initial risk segmentation set is iteratively optimized by merging similar segments and adjusting clustering parameters to obtain a risk segmentation set.
6. The information asset prediction method of claim 5, wherein, The step S13 comprises: S61, based on the risk segmentation set, a data subset meeting the current segmentation condition and located within a preset historical time window is screened from a historical asset pool database; S62, basic statistical indicators are calculated for the data subset of each risk segmentation to obtain average repayment rates, default rates and early repayment rates; S63, time series decomposition is performed on time series composed of the basic statistical indicators to obtain trend items, seasonal items and residual items, and an abnormal detection algorithm is adopted to identify outliers in the data subset to obtain abnormal quantitative scores; S64, the average repayment rate, the default rate, the early repayment rate, the trend item, the seasonal item, the residual item and the abnormal quantitative score are fused to form a segmentation performance feature vector; S65, integrate the segment performance feature vectors of all segments to obtain a historical performance feature set.
7. The information asset prediction method of claim 6, wherein, The step S21 comprises: S71, constructing a model input feature vector based on the effective feature set and the historical performance feature set; S72, for each segment, performing characteristic evaluation on the segment based on the model input feature vector, and quantitatively calculating an evaluation value corresponding to a characteristic evaluation function from three dimensions of data richness, feature complexity and historical volatility; S73, selecting a prediction model according to the evaluation value, wherein an LSTM neural network is used for a high-quality segment, an XGBoost integrated model is used for a medium-quality segment, and a rule-based expert system is used for a low-quality segment; S74, training optimal parameters of the selected prediction model by minimizing a loss function, then inputting a model input feature vector of a current segment, predicting a cash flow sequence in a future time period, and obtaining a default amount and a repayment amount; S75, aggregating the predicted cash flow sequences of the segments by weighted summation according to the remaining principal ratio of the segments, outputting a whole future cash flow sequence of the asset pool, and obtaining a prediction result.
8. The information asset prediction method of claim 7, wherein, The step S22 comprises: S81, performing multi-period rolling backtest based on historical data of a preset month period to generate a predicted cash flow sequence, comparing the predicted cash flow sequence with an actual cash flow period by period, and calculating a segment prediction deviation; S82, calculating mean square error and mean absolute error indicators based on the segment prediction deviation to obtain a model performance evaluation result; S83, automatically grabbing objective trend indicators of a past preset month period every month, and detecting whether a core parameter corresponding to the model performance evaluation result deviates; S84, when it is detected that the core parameter deviates, triggering a model reconstruction process, automatically adjusting feature weights and model structures, and obtaining an optimal model.
9. An intelligent prediction system characterized by, The intelligent prediction system is used to implement the information asset prediction method according to any one of claims 1 to 8.