Data evaluation method and device, computer equipment and storage medium

Abstract

The invention belongs to the technical field of artificial intelligence, and relates to a data evaluation method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining traditional factor feature data and derivative feature data of a customer; preprocessing the traditional factor feature data and the derivative feature data to obtain target traditional factor feature data and target derivative feature data; calling a fusion prediction model; wherein the fusion prediction model comprises a first prediction model and a second prediction model; predicting the target traditional factor feature data based on a first prediction model to obtain a preliminary prediction result; predicting the preliminary prediction result and the derivative feature data based on a second prediction model to obtain a target prediction result; performing score conversion on the target prediction result to obtain target score data; and analyzing the target scoring data based on the intelligent agent to generate a customer group result. The method can be applied to a customer group analysis scene in the field of financial science and technology, and the processing efficiency and accuracy of customer group division are improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology and can be applied to the financial technology field, particularly to data evaluation methods, devices, computer equipment and storage media. Background Technology

[0002] In traditional corporate banking, customer segmentation relies heavily on information collected through offline channels and manual analysis, resulting in low efficiency and inaccuracy. Specifically, traditional methods typically segment customers based on basic operational data (such as company size and industry classification), lacking in-depth integration and analysis of multi-dimensional customer characteristics (such as transaction behavior, credit status, and potential needs). This crude segmentation approach struggles to accurately identify differentiated customer characteristics, leading to a poor match between customer segmentation results and actual business needs, thus impacting the effectiveness of risk assessment, product recommendations, and service strategy development.

[0003] For example, in SME lending within the financial sector, traditional methods may segment customers solely based on revenue size and collateral value, failing to adequately consider crucial factors such as cash flow stability, supply chain relationships, or industry cyclical risks. If a company operates in a seasonal industry with significant cash flow fluctuations, traditional methods might categorize it as low-risk and grant it excessively high credit lines, exposing the bank to potential default risks. This segmentation bias not only increases the operating costs of financial institutions but may also reduce customer satisfaction due to inaccurate service strategies.

[0004] Therefore, there is an urgent need to provide a customer segmentation method based on intelligent analysis to improve the efficiency and accuracy of customer segmentation, optimize the allocation of financial service resources, and enhance the market competitiveness of financial institutions. Summary of the Invention

[0005] The purpose of this application is to propose a data evaluation method, apparatus, computer equipment, and storage medium to solve the technical problem that the current customer group segmentation mainly relies on information collected through offline channels and manual analysis, resulting in low segmentation efficiency and insufficient accuracy.

[0006] Firstly, a data evaluation method is provided, including:

[0007] Obtain the customer's traditional factor characteristic data, and obtain the customer's derived characteristic data based on a preset information type;

[0008] The traditional factor feature data is preprocessed to obtain the corresponding target traditional factor feature data, and the derived feature data is preprocessed to obtain the corresponding target derived feature data;

[0009] Invoke a preset fusion prediction model; wherein the fusion prediction model includes a first prediction model and a second prediction model;

[0010] Based on the first prediction model, the target traditional factor feature data are predicted and processed to obtain the corresponding preliminary prediction results;

[0011] Based on the second prediction model, the preliminary prediction results and the derived feature data are processed to obtain the corresponding target prediction results;

[0012] The target prediction results are subjected to a scoring conversion process to obtain the corresponding target score data;

[0013] The target rating data is analyzed and processed based on a preset intelligent agent to generate the customer group results for the customer.

[0014] Secondly, a data evaluation device is provided, comprising:

[0015] The first acquisition module is used to acquire the customer's traditional factor characteristic data and acquire the customer's derived characteristic data based on a preset information type;

[0016] The preprocessing module is used to preprocess the traditional factor feature data to obtain the corresponding target traditional factor feature data, and to preprocess the derived feature data to obtain the corresponding target derived feature data.

[0017] The first calling module is used to call a preset fusion prediction model; wherein, the fusion prediction model includes a first prediction model and a second prediction model;

[0018] The first prediction module is used to perform prediction processing on the target traditional factor feature data based on the first prediction model to obtain the corresponding preliminary prediction results;

[0019] The second prediction module is used to perform prediction processing on the preliminary prediction result and the derived feature data based on the second prediction model to obtain the corresponding target prediction result.

[0020] The conversion module is used to perform a score conversion process on the target prediction result to obtain the corresponding target score data;

[0021] The analysis module is used to analyze and process the target rating data based on a preset intelligent agent to generate the customer group results for the customer.

[0022] Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described data evaluation method.

[0023] Fourthly, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the steps of the above-described data evaluation method.

[0024] In the above-mentioned data evaluation method, apparatus, computer equipment, and storage medium, the following steps are taken: First, traditional factor characteristic data of the customer is acquired, and derived characteristic data of the customer is acquired based on a preset information type. Then, the traditional factor characteristic data is preprocessed to obtain corresponding target traditional factor characteristic data, and the derived characteristic data is preprocessed to obtain corresponding target derived characteristic data. Next, a preset fusion prediction model is invoked, wherein the fusion prediction model includes a first prediction model and a second prediction model. Subsequently, the target traditional factor characteristic data is predicted based on the first prediction model to obtain a corresponding preliminary prediction result. The preliminary prediction result and the derived characteristic data are then predicted based on the second prediction model to obtain a corresponding target prediction result. Further, the target prediction result is processed through a scoring conversion to obtain corresponding target scoring data. Finally, the target scoring data is analyzed and processed based on a preset intelligent agent to generate the customer group results for the customer. Based on the above automated processing flow, this application obtains customers' traditional factor feature data and their derived feature data based on information type. Then, it preprocesses the traditional factor feature data and derived feature data to obtain corresponding target traditional factor feature data and target derived feature data. Next, it uses a first prediction model included in the fusion prediction model to predict the target traditional factor feature data to obtain preliminary prediction results. Then, it uses a second prediction model included in the fusion prediction model to predict the preliminary prediction results and derived feature data to obtain target prediction results. Subsequently, it performs a scoring conversion process on the target prediction results to obtain target scoring data. Finally, it analyzes and processes the target scoring data based on the use of an intelligent agent to automatically and accurately generate customer group results, improving the processing efficiency of customer group segmentation and ensuring the accuracy of the obtained customer group results. Attached Figure Description

[0025] To more clearly illustrate the solutions in this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is an exemplary system architecture diagram to which this application can be applied;

[0027] Figure 2This is a flowchart of an embodiment of the data evaluation method according to this application;

[0028] Figure 3 This is a schematic diagram of a structure of an embodiment of the data evaluation apparatus according to this application;

[0029] Figure 4 This is a schematic diagram of the structure of one embodiment of the computer device according to this application. Detailed Implementation

[0030] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein in the specification of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "comprising" and "having," and any variations thereof, in the specification, claims, and foregoing drawings of this application, are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification, claims, or foregoing drawings of this application are used to distinguish different objects, not to describe a particular order.

[0031] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0032] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0033] like Figure 1 As shown, system architecture 100 may include terminal device 101, network 102, and server 103. Terminal device 101 may be a laptop 1011, tablet 1012, or mobile phone 1013. Network 102 is used as a medium to provide a communication link between terminal device 101 and server 103. Network 102 may include various connection types, such as wired, wireless communication links, or fiber optic cables, etc.

[0034] Users can use terminal device 101 to interact with server 103 via network 102 to receive or send messages, etc. Various communication client applications can be installed on terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social media platform software, etc.

[0035] Terminal device 101 can be various electronic devices with a display screen and support web browsing. In addition to laptops 1011, tablets 1012, or mobile phones 1013, terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 player (Moving Picture Experts Group Audio Layer IV), a laptop computer, and a desktop computer, etc.

[0036] Server 103 can be a server that provides various services, such as a backend server that provides support for the pages displayed on terminal device 101.

[0037] It should be noted that the data evaluation method provided in the embodiments of this application is generally executed by a server / terminal device, and correspondingly, the data evaluation device is generally set in the server / terminal device.

[0038] It should be understood that Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0039] Continue to refer to Figure 2 A flowchart illustrating an embodiment of the data evaluation method according to this application is shown. The order of steps in the flowchart can be changed, and some steps can be omitted, depending on different needs. The data evaluation method provided in this application embodiment can be applied to any scenario requiring data evaluation, and thus can be applied to products in these scenarios, such as data evaluation products in the financial insurance field. The data evaluation method includes the following steps:

[0040] Step S201: Obtain the customer's traditional factor characteristic data, and obtain the customer's derived characteristic data based on a preset information type.

[0041] In this embodiment, the data evaluation method operates on an electronic device (e.g., Figure 1The server / terminal device shown can acquire the customer's traditional factor characteristic data and derived characteristic data through wired or wireless connections. It should be noted that the aforementioned wireless connection methods may include, but are not limited to, 3G / 4G / 5G connections, WiFi connections, Bluetooth connections, WiMAX connections, Zigbee connections, UWB (ultra-Width Band) connections, and other currently known or future wireless connection methods. The implementing entity of this application is specifically a data evaluation system, which can be simply referred to as the system. The aforementioned customer's traditional factor characteristic data may include structured data such as the customer's financial data (e.g., total assets, debt ratio), behavioral data (e.g., transaction frequency, product holdings), and demographic information (e.g., age, occupation). The specific implementation process of acquiring the customer's derived characteristic data based on preset information types will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated upon here.

[0042] Step S202: Preprocess the traditional factor feature data to obtain the corresponding target traditional factor feature data, and preprocess the derived feature data to obtain the corresponding target derived feature data.

[0043] In this embodiment, the preprocessing includes: for traditional factor feature data, if certain financial indicators are missing, they can be reasonably estimated based on industry averages or historical customer data; for derived feature data, if missing indicators exist, they can be supplemented by re-invoking the large model or by interpolation based on other relevant factors. Simultaneously, outliers in the data are checked, such as abnormally high transaction amounts or abnormally low business scale scores, and their causes are analyzed. If the data entry is incorrect, it is corrected; if it is a true situation, it is retained and marked. Furthermore, since the dimensions and value ranges of traditional factor feature data and derived feature data may differ significantly, data standardization and normalization are required. For example, for numerical data, Z-score standardization can be used to make the mean 0 and the standard deviation 1; for categorical data, one-hot encoding can be performed to convert it into numerical form for subsequent model processing.

[0044] Step S203: Invoke a preset fusion prediction model; wherein the fusion prediction model includes a first prediction model and a second prediction model.

[0045] In this embodiment, the construction process of the above-mentioned fusion prediction model will be described in more detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0046] Step S204: Based on the first prediction model, perform prediction processing on the target traditional factor feature data to obtain the corresponding preliminary prediction results.

[0047] In this embodiment, by inputting the aforementioned target traditional factor feature data into the first prediction model, the first prediction model generates preliminary prediction results (such as median values ​​of customer potential) based on the traditional factor features. For example, the first prediction model outputs a continuous value (such as 0.2 to 0.8), representing a potential prediction based on the traditional factor feature data.

[0048] Step S205: Based on the second prediction model, perform prediction processing on the preliminary prediction result and the derived feature data to obtain the corresponding target prediction result.

[0049] In this embodiment, the specific implementation process of predicting the preliminary prediction result and the derived feature data based on the second prediction model to obtain the corresponding target prediction result will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0050] Step S206: Perform a scoring conversion process on the target prediction result to obtain the corresponding target scoring data.

[0051] In this embodiment, the specific implementation process of performing scoring conversion on the target prediction result to obtain the corresponding target score data will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0052] Step S207: Analyze and process the target rating data based on a preset intelligent agent to generate the customer group results for the customer.

[0053] In this embodiment, the specific implementation process of analyzing and processing the target scoring data based on a preset intelligent agent to generate the customer group result of the customer will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0054] This application first acquires the customer's traditional factor characteristic data and, based on a preset information type, acquires the customer's derived characteristic data. Then, it preprocesses the traditional factor characteristic data to obtain corresponding target traditional factor characteristic data, and preprocesses the derived characteristic data to obtain corresponding target derived characteristic data. Next, it calls a preset fusion prediction model, which includes a first prediction model and a second prediction model. Subsequently, it performs prediction processing on the target traditional factor characteristic data based on the first prediction model to obtain a corresponding preliminary prediction result. Then, it performs prediction processing on the preliminary prediction result and the derived characteristic data based on the second prediction model to obtain a corresponding target prediction result. Further, it performs a rating conversion process on the target prediction result to obtain corresponding target rating data. Finally, it analyzes and processes the target rating data based on a preset intelligent agent to generate the customer group results for the customer. Based on the above automated processing flow, this application obtains customers' traditional factor feature data and their derived feature data based on information type. Then, it preprocesses the traditional factor feature data and derived feature data to obtain corresponding target traditional factor feature data and target derived feature data. Next, it uses a first prediction model included in the fusion prediction model to predict the target traditional factor feature data to obtain preliminary prediction results. Then, it uses a second prediction model included in the fusion prediction model to predict the preliminary prediction results and derived feature data to obtain target prediction results. Subsequently, it performs a scoring conversion process on the target prediction results to obtain target scoring data. Finally, it analyzes and processes the target scoring data based on the use of an intelligent agent to automatically and accurately generate customer group results, improving the processing efficiency of customer group segmentation and ensuring the accuracy of the obtained customer group results.

[0055] In some optional implementations, the step S201 of obtaining the customer's derived feature data based on a preset information type includes the following steps:

[0056] Obtain the customer profile corresponding to the customer.

[0057] In this embodiment, a matching customer profile can be obtained by performing a customer profiling query. This customer profile may include basic dimensions such as: value customer (defined based on EVA and average daily deposit balance), credit customer (defined based on 38-day rolling credit average), registration period (company registration date), registered capital (company's registered capital according to business registration information), and industry (industry category). For example, "There is a corporate customer with a value customer attribute of 1.5 million, a credit customer attribute of 800,000, a registration period of 8 years, a registered capital of 12 million, and an industry of electronic information."

[0058] Call the preset target large language model and obtain the preset large model prompt words.

[0059] In this embodiment, the selection of the aforementioned target large language model is not specifically limited and can be determined according to actual business needs. For example, an LLM large model can be used. The prompt words for the large model are pre-constructed prompt words consisting of role settings, defined derived factors, and a scoring system. Specifically, regarding role settings: firstly, the role is explicitly set as a corporate banking client manager in the prompt words. This allows the large model to better think and reason from a professional perspective of corporate banking. For example, the prompt words could begin with "You are an experienced corporate banking client manager, responsible for deeply understanding the needs of corporate clients and providing precise services."

[0060] The defined derivative factors include: **Company Business Scale Score:** This score comprehensively considers factors such as the company's asset size, revenue, and number of employees, combined with industry averages. For example, a company with significantly larger assets than the industry average, stable revenue growth, and a reasonable number of employees will receive a higher score; conversely, a lower score will be given. **Settlement Transaction Volume Increase Potential:** This analyzes the company's past settlement transaction volume trends, business expansion, and market environment. If the company is in a period of rapid business expansion, establishing partnerships with more suppliers and customers, then the likelihood of an increase in settlement transactions is high, resulting in a higher potential score. **Settlement Amount Growth Potential:** This examines the company's expected business scale growth, product or service price trends, and market share changes. If the company anticipates launching new products or raising product prices, and its market share is expected to expand, then the settlement amount growth potential is significant, resulting in a higher score. **Company's Core Bank Value Enhancement Potential:** This considers the company's deposit balance, loan business, and fee income within the bank, as well as the company's future development plans and funding needs. If the company plans to increase its loan amount and expand its fee income within the bank, then its value enhancement potential within the bank is significant, resulting in a higher score. Diversified Settlement Methods Propensity Score: This score assesses a company's payment habits, liquidity needs, and receptiveness to new technologies and payment methods. Companies that frequently experiment with new payment methods and have high liquidity requirements will receive a higher score for diversified settlement methods. Counterparty Stability Score: This score analyzes the industry distribution, cooperation history, and creditworthiness of a company's counterparties. Companies with counterparties primarily concentrated in stable industries, with long-standing cooperation histories and good credit will receive a higher counterparty stability score.

[0061] Setting up a scoring system includes defining a score range of 1 to 10, ensuring that each score range has a clear definition. For example, 1-3 points indicate that the customer's characteristics or potential in that area are extremely low; 4-7 points indicate moderate; and 8-10 points indicate high. This ensures that the scores are differentiated and accurately reflect the characteristics and potential of the company's customers.

[0062] Based on the target large language model, the customer profile is inferred according to the prompt words of the large model to obtain the inference result corresponding to the information type.

[0063] In this embodiment, by inputting the aforementioned large model prompts and customer profiles into the target large language model, the target large language model will perform corresponding derivative factor mining processing on the customer profiles based on the large model prompts, and use the output inference results as the customer's derivative feature data. Here, the aforementioned information type refers to the information type that matches the derivative factors defined above.

[0064] The reasoning result is used as the customer's derived feature data.

[0065] In this embodiment, since the original customer profile information is relatively limited, the powerful reasoning capabilities of the large model, combined with public and private customer profile knowledge, generate a series of derived factor features, enabling a more comprehensive and in-depth characterization of the company's customers. This rich customer profile information provides comprehensive and detailed data support for accurately predicting customer value enhancement potential and uncovering business opportunities.

[0066] This application obtains a customer profile corresponding to the customer; then calls a preset target large language model and obtains preset large model prompt words; subsequently, based on the target large language model, it performs inference processing on the customer profile according to the large model prompt words to obtain an inference result corresponding to the information type; and then uses the inference result as the customer's derived feature data. Based on the above processing flow, this application improves the richness and comprehensiveness of the generated derived feature data by obtaining a customer profile corresponding to the customer, and then using the target large language model to perform inference processing on the customer profile according to the obtained large model prompt words to obtain an inference result corresponding to the information type and using it as the required customer's derived feature data. This facilitates the depiction of customer profiles from more dimensions and in greater depth, providing comprehensive and detailed data support for accurate prediction of customer groups.

[0067] In some optional implementations of this embodiment, before step S203, the electronic device may further perform the following steps:

[0068] Obtain pre-collected historical customer factor characteristic data.

[0069] In this embodiment, relevant data from corporate clients is collected, including traditional factor feature data and large-scale model derived factor feature data. Traditional factor feature data can cover the client's historical financial data (such as various indicators in the balance sheet, income statement, and cash flow statement) and transaction records (such as transaction amount, transaction frequency, and counterparty). Large-scale model derived factor feature data is the feature data obtained through the derivative factor mining steps of the aforementioned LLM large-scale model cold start for historical clients, such as company business scale score, potential for increasing settlement number, potential for increasing settlement amount, potential for increasing the company's core value, preference for diversified settlement methods, and counterparty stability score.

[0070] Construct corresponding sample data based on the aforementioned factor feature data.

[0071] In this embodiment, the collected factor feature data is cleaned, and missing values ​​are handled. For traditional factor feature data, if certain financial indicators are missing, they can be reasonably estimated based on industry averages or historical customer data. For large-model derived factor feature data, if missing values ​​exist, they can be supplemented by re-invoking the large model or by interpolation based on other relevant factors. Simultaneously, outliers in the data are checked, such as abnormally high transaction amounts or abnormally low business scale scores. The causes are analyzed; if the data entry is incorrect, it is corrected; if it reflects the actual situation, it is retained and marked. Furthermore, since the dimensions and value ranges of traditional factors and large-model derived factors may differ significantly, the data needs to be standardized and normalized. For example, for numerical data, the Z-score standardization method can be used to make the data mean 0 and the standard deviation 1; for categorical data, one-hot encoding can be performed to convert it into numerical form for subsequent model processing. The cleaned, standardized, and normalized data are then used as the corresponding sample data.

[0072] Invoke the preset initial fusion model.

[0073] In this embodiment, the initial fusion model is a fusion model including an XGBoost model and a logistic regression model. The specific architecture design includes: XGBoost algorithm model initialization: determining initial values ​​for some key parameters of the XGBoost algorithm, such as the learning rate (usually set between 0.01 and 0.3), the maximum tree depth (usually set between 3 and 10), and the subsampling ratio (usually between 0.5 and 1). These initial values ​​can be set based on experience or by referring to relevant literature. Furthermore, the LLM large model is treated as an independent feature module and combined with the features of the XGBoost algorithm. Simultaneously, it can be considered whether further processing or filtering of the large model-derived features is needed, for example, removing features with less impact on the prediction target through feature importance analysis. Model fusion architecture: The XGBoost algorithm and the LLM large model inference-derived feature fusion algorithm are fused using a stacking approach. First, the XGBoost algorithm is used to train and predict traditional factor features, yielding preliminary prediction results. Then, these preliminary prediction results, along with factor features derived from the larger model, are used as new features and input into another model (i.e., a logistic regression model) for further training and prediction. This logistic regression model will comprehensively consider both the preliminary prediction results and the features derived from the larger model to obtain a more accurate final prediction result.

[0074] Obtain the preset optimization objective function and parameter adjustment strategy.

[0075] In this embodiment, the above-mentioned optimization objective function includes: Obj(θ) = L_0(θ) + L_llm(θ) + Ω(θ). Where, L_0(θ): A suitable loss function is selected based on the type of prediction task. For regression problems, such as predicting the specific numerical value of customer value enhancement potential, mean squared error (MSE) can be chosen as the loss function; for classification problems, such as dividing customers into high, medium, and low potential groups, cross-entropy loss function can be chosen. L_llm(θ): Similarly, a suitable loss function is selected based on the type of prediction task, ensuring consistency with the selection of L_0(θ) to reasonably measure the prediction error of large model-derived features. Ω(θ): Typically, L1 or L2 regularization functions are used to prevent model overfitting. L1 regularization can make the weights of some features zero, achieving feature selection; L2 regularization can make the feature weights more uniform, improving the stability of the model.

[0076] The parameter tuning strategies described above include: using cross-validation to adjust model parameters; dividing the dataset into training, validation, and test sets; training the model on the training set and evaluating its performance on the validation set; then adjusting model parameters, such as the learning rate and maximum tree depth, based on the performance on the validation set; and using grid search or random search to try different parameter combinations within a certain range, selecting the combination that minimizes the objective function Obj(θ) on the validation set. This iterative tuning continues until the model performance reaches a satisfactory level.

[0077] Based on the optimization objective function and the parameter adjustment strategy, the initial fusion model is trained and evaluated using the sample data until a generated model that meets the preset construction requirements is obtained.

[0078] In this embodiment, the training and evaluation of the initial fusion model includes: Model training and optimization algorithm selection: After adjusting the parameters, the fused model is trained using the entire training set. During training, optimization algorithms such as batch gradient descent or stochastic gradient descent can be used to minimize the objective function. Batch gradient descent updates parameters using all training data each time, resulting in slower convergence but better stability; stochastic gradient descent randomly selects a training sample for parameter update each time, resulting in faster convergence but potentially larger fluctuations. Model evaluation and evaluation metric selection: After training, the model is evaluated on the test set. Evaluation metrics are selected based on the prediction task. For regression problems, metrics such as mean squared error (MSE), mean absolute error (MAE), and R-squared can be selected; for classification problems, metrics such as accuracy, recall, and F1 score can be selected. Performance judgment: The performance of the model is judged based on the results of the evaluation metrics to determine whether it meets the requirements. If the model performance is not ideal, the reasons are analyzed, such as improper feature selection, unreasonable parameter adjustment, or problems with the model fusion strategy. Then, the parameters are readjusted or the model fusion strategy is improved, and training and evaluation are performed again until the model performance reaches the expected target.

[0079] The generative model is used as the fusion prediction model.

[0080] In this embodiment, after the fusion prediction model is trained and evaluated, it is used to predict new customer data, obtaining a value enhancement potential score for each customer. The score can be transformed based on the output value of the objective function, for example, mapping the output value to a score range of 1-10. Linear transformation or other suitable transformation methods can be used to map the model's output value to a specified score interval.

[0081] This application acquires pre-collected historical customer factor feature data; constructs corresponding sample data based on the factor feature data; then calls a preset initial fusion model; subsequently acquires a preset optimization objective function and parameter adjustment strategy; subsequently, based on the optimization objective function and parameter adjustment strategy, uses the sample data to train and evaluate the initial fusion model until a generated model that meets the preset construction requirements is obtained; finally, the generated model is used as the fusion prediction model. Based on the above processing flow, this application constructs sample data based on pre-collected historical customer factor feature data, and then uses the sample data to train and evaluate the initial fusion model based on the acquired optimization objective function and parameter adjustment strategy. This enables efficient and accurate construction of the required fusion prediction model, improves the construction efficiency of the fusion prediction model, and ensures the model performance of the obtained fusion prediction model.

[0082] In some alternative implementations, step S205 includes the following steps:

[0083] Obtain the preset fusion strategy.

[0084] In this embodiment, the above-mentioned fusion strategy specifically adopts a splicing processing strategy. The order of splicing processing is not specifically limited and can be set according to actual business needs. For example, the order of preliminary prediction results - target derived feature data or target derived feature data - preliminary prediction results can be adopted.

[0085] Based on the fusion strategy, the preliminary prediction results and the target derived feature data are fused to obtain the corresponding fused data.

[0086] In this embodiment, the preliminary prediction results and target derived feature data can be spliced ​​together in a prescribed order according to the selected fusion strategy to obtain the corresponding fused data.

[0087] Based on the second prediction model, the fused data is processed according to a preset specified processing method to obtain the corresponding specified prediction result.

[0088] In this embodiment, a second prediction model is loaded, including: weight coefficients (w0, w1, ..., wN): corresponding to the weights of each feature in the new fused data. Bias term (b): the intercept term of the model. The above-mentioned processing method refers to weighted summation calculation. The corresponding linear output can be obtained by weighting and summing the fused data according to the matched weight coefficients, and then the linear output is mapped to the [0,1] interval through the Sigmoid function to obtain the final prediction probability, which is used as the final target prediction result.

[0089] The specified prediction result is used as the target prediction result.

[0090] This application obtains a preset fusion strategy; then, based on the fusion strategy, it fuses the preliminary prediction result with the target derived feature data to obtain corresponding fused data; subsequently, based on the second prediction model, it processes the fused data according to a preset specified processing method to obtain a corresponding specified prediction result; and finally, it uses the specified prediction result as the target prediction result. Based on the above processing flow, this application obtains fused data by fusing the preliminary prediction result with the target derived feature data using a fusion strategy, and then processes the fused data according to a preset specified processing method using the second prediction model. This enables the efficient and accurate generation of corresponding target prediction results, improving the accuracy of the generated target prediction results and facilitating the subsequent use of the target prediction results to improve the accuracy of subsequently generated customer group results.

[0091] In some alternative implementations, step S206 includes the following steps:

[0092] Invoke the preset rating conversion rules.

[0093] In this embodiment, the above-mentioned scoring conversion rule refers to the linear mapping rule, which includes: linearly mapping the target prediction result (e.g., 0.15 to 0.95) output by the second prediction model to a scoring range of 1 to 10 points. The corresponding scoring conversion formula includes: score = 1 + (output value - minimum value) × (10 / (maximum value - minimum value)).

[0094] The target prediction result is transformed based on the scoring transformation rule to obtain the corresponding transformed data.

[0095] In this embodiment, based on the rules of the above-mentioned scoring conversion rules, the target prediction result can be input into the corresponding position in the above-mentioned scoring conversion formula for calculation, and the converted data generated by the calculation can be used as the corresponding target score data. For example, the target prediction result (output value) of 0.5 output by the second prediction model is mapped to 1 + (0.5 - 0.15) × (10 / 0.8) ≈ 4.5 points.

[0096] The transformed data is used as the target score data.

[0097] This application calls a preset scoring conversion rule; then, based on the scoring conversion rule, it converts the target prediction result to obtain corresponding converted data; subsequently, it uses the converted data as the target scoring data. Based on the above processing flow, this application calls a preset scoring conversion rule, then converts the target prediction result based on the use of the scoring conversion rule, and uses the generated converted data as the corresponding target scoring data. This enables efficient and accurate scoring conversion of the target prediction result, ensuring the accuracy of the obtained target scoring data.

[0098] In some optional implementations of this embodiment, step S207 includes the following steps:

[0099] The agent invokes a preset scoring mapping table.

[0100] In this embodiment, the aforementioned intelligent agent is a pre-built intelligent agent assistant. The aforementioned rating mapping table is a pre-built data table storing a one-to-one correspondence between rating intervals and customer groups. For example, 1-3 points: low-potential customers (require basic services). 4-7 points: medium-potential customers (require regular marketing). 8-10 points: high-potential customers (require personalized services).

[0101] The specified rating range that matches the target rating data is retrieved from the rating mapping table.

[0102] In this embodiment, the target score data is used to query the score mapping table to find the score range that matches the target score data.

[0103] Obtain customer group information corresponding to the specified rating range.

[0104] In this embodiment, customer group information corresponding to the specified rating interval can be extracted from the rating mapping table based on the one-to-one correspondence between the rating interval and the customer group, and used as the customer group result.

[0105] The customer group information is used as the customer group result for the customer.

[0106] This application utilizes a pre-defined rating mapping table, initiated by an intelligent agent. It then retrieves a specified rating interval matching the target rating data from the mapping table, obtains customer group information corresponding to that interval, and subsequently uses this customer group information as the customer's customer group result. Based on this process, this application achieves efficient and accurate analysis of the target rating data, ensuring the accuracy of the generated customer group results.

[0107] In some optional implementations of this embodiment, after step S207, the electronic device may further perform the following steps:

[0108] Obtain the preset information parsing logic.

[0109] In this embodiment, by designing information parsing logic, the intelligent agent assistant can interpret business opportunity information (including customer rating data and derived feature data) based on information in the knowledge base. For example, when receiving business opportunity information from a customer, the intelligent agent assistant can extract relevant industry dynamic information from the knowledge base based on the customer's industry type, and analyze the opportunities and challenges the customer may face by combining the customer's business scale and value enhancement potential rating.

[0110] This involves pre-constructing a comprehensive knowledge base that includes industry knowledge, financial product knowledge, and market dynamics. This knowledge base will serve as the foundation for the intelligent agent to interpret business opportunities. For example, the knowledge base contains information such as business development trends in different industries, the characteristics of various financial products, and their applicable scenarios.

[0111] Based on the information parsing logic, the target score data and the target derived feature data are parsed to obtain the corresponding parsing results.

[0112] In this embodiment, based on the aforementioned intelligent agent, the aforementioned information parsing logic can be used to perform information parsing processing on the aforementioned target scoring data and target derived feature data, so as to generate corresponding parsing results.

[0113] Based on the analysis results, service suggestions corresponding to the customer are generated.

[0114] In this embodiment, the system dynamically generates personalized service recommendations based on the analysis results generated by the intelligent agent. For example, if the analysis results indicate that a high-potential customer plans to expand their business in the near future, a service recommendation to provide supply chain financial services to that customer will be generated.

[0115] The service recommendations are then processed for output.

[0116] In this embodiment, the output processing of service suggestions includes: First, associating the target score data and target derived feature data with the aforementioned service suggestions to ensure that the account manager team can clearly understand the situation of each customer and the corresponding service plan. Second, designing an information integration format so that the account manager team can easily view and use it. This can be done in the form of a report, integrating customer target score data, target derived feature data, service suggestions, and marketing strategies into a single report, which can include various forms of information display such as charts and text descriptions. Third, selecting an appropriate delivery channel to deliver the integrated information to the account manager team. This can be done through an internal enterprise information system, such as a Customer Relationship Management (CRM) system, to push the information to account managers; or via email, sending the integrated information as an attachment to the account manager. Furthermore, determining the frequency of information delivery, such as daily, weekly, or monthly. Simultaneously, establishing a notification mechanism to promptly notify the account manager team when new service suggestions are generated, ensuring they can obtain the latest information in a timely manner.

[0117] This application obtains a preset information parsing logic; then, based on the information parsing logic, it parses the target rating data and the target derived feature data to obtain corresponding parsing results; subsequently, it generates service suggestions corresponding to the customer based on the parsing results; and finally, it outputs the service suggestions. Based on the above processing flow, this application uses information parsing logic to parse the target rating data and target derived feature data to obtain parsing results, then automatically generates service suggestions corresponding to the customer based on the parsing results, and intelligently outputs the service suggestions. This achieves automatic and intelligent generation of customer-matched service suggestions based on the parsing results, improving the personalization and accuracy of the generated service suggestions, and the output of the service suggestions can help improve the work efficiency of the account manager team and customer satisfaction.

[0118] In some alternative implementations, the user information obtained is subject to user consent and complies with relevant laws and policies.

[0119] Furthermore, any software tools or components not belonging to our company that appear in the embodiments of this application are merely illustrative examples and do not represent actual use.

[0120] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0121] It should be emphasized that, in order to further ensure the privacy and security of the results for the aforementioned customer groups, these results can also be stored in a node of a blockchain.

[0122] The blockchain referred to in this application is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Essentially, a blockchain is a decentralized database, a chain of data blocks linked together using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. A blockchain can include an underlying blockchain platform, a platform product service layer, and an application service layer.

[0123] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0124] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0125] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing related hardware through computer-readable instructions. These computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the methods described above. The aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, optical disk, or read-only memory (ROM), or random access memory (RAM).

[0126] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0127] Further reference Figure 3 As a response to the above Figure 2 To implement the method shown, this application provides an embodiment of a data evaluation apparatus, which is similar to... Figure 2 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0128] like Figure 3 As shown, the data evaluation device 300 described in this embodiment includes: a first acquisition module 301, a preprocessing module 302, a first retrieval module 303, a first prediction module 304, a second prediction module 305, a conversion module 306, and an analysis module 307. Wherein:

[0129] The first acquisition module is used to acquire the customer's traditional factor characteristic data and acquire the customer's derived characteristic data based on a preset information type;

[0130] The preprocessing module is used to preprocess the traditional factor feature data to obtain the corresponding target traditional factor feature data, and to preprocess the derived feature data to obtain the corresponding target derived feature data.

[0131] The first calling module is used to call a preset fusion prediction model; wherein, the fusion prediction model includes a first prediction model and a second prediction model;

[0132] The first prediction module is used to perform prediction processing on the target traditional factor feature data based on the first prediction model to obtain the corresponding preliminary prediction results;

[0133] The second prediction module is used to perform prediction processing on the preliminary prediction result and the derived feature data based on the second prediction model to obtain the corresponding target prediction result.

[0134] The conversion module is used to perform a score conversion process on the target prediction result to obtain the corresponding target score data;

[0135] The analysis module is used to analyze and process the target rating data based on a preset intelligent agent to generate the customer group results for the customer.

[0136] In this embodiment, the operations performed by the above modules or units correspond one-to-one with the steps of the data evaluation method in the aforementioned embodiments, and will not be repeated here.

[0137] In some optional implementations of this embodiment, the first acquisition module 301 includes:

[0138] The first acquisition submodule is used to acquire the customer profile corresponding to the customer;

[0139] The second acquisition submodule is used to call the preset target large language model and obtain the preset large model prompt words;

[0140] The reasoning submodule is used to perform reasoning processing on the customer profile based on the target large language model and the prompt words of the large model, so as to obtain the reasoning result corresponding to the information type.

[0141] The first determining submodule is used to use the reasoning result as the customer's derived feature data.

[0142] In this embodiment, the operations performed by the above modules or units correspond one-to-one with the steps of the data evaluation method in the aforementioned embodiments, and will not be repeated here.

[0143] In some optional implementations of this embodiment, the data evaluation apparatus further includes:

[0144] The second acquisition module is used to acquire pre-collected historical customer factor feature data;

[0145] The construction module is used to construct corresponding sample data based on the factor feature data;

[0146] The second calling module is used to call the preset initial fusion model;

[0147] The third acquisition module is used to acquire the preset optimization objective function and parameter adjustment strategy;

[0148] The processing module is used to train and evaluate the initial fusion model using the sample data based on the optimization objective function and the parameter adjustment strategy, until a generated model that meets the preset construction requirements is obtained.

[0149] A determination module is used to use the generated model as the fusion prediction model.

[0150] In this embodiment, the operations performed by the above modules or units correspond one-to-one with the steps of the data evaluation method in the aforementioned embodiments, and will not be repeated here.

[0151] In some optional implementations of this embodiment, the second prediction module 305 includes:

[0152] The third acquisition submodule is used to acquire the preset fusion strategy;

[0153] The fusion submodule is used to fuse the preliminary prediction result and the target derived feature data based on the fusion strategy to obtain the corresponding fused data.

[0154] The processing submodule is used to process the fused data according to a preset specified processing method based on the second prediction model to obtain the corresponding specified prediction result;

[0155] The second determining submodule is used to use the specified prediction result as the target prediction result.

[0156] In this embodiment, the operations performed by the above modules or units correspond one-to-one with the steps of the data evaluation method in the aforementioned embodiments, and will not be repeated here.

[0157] In some optional implementations of this embodiment, the conversion module 306 includes:

[0158] The first submodule is used to invoke the preset scoring conversion rules;

[0159] The conversion submodule is used to convert the target prediction result based on the scoring conversion rules to obtain the corresponding converted data;

[0160] The third determining submodule is used to use the transformed data as the target scoring data.

[0161] In this embodiment, the operations performed by the above modules or units correspond one-to-one with the steps of the data evaluation method in the aforementioned embodiments, and will not be repeated here.

[0162] In some optional implementations of this embodiment, the analysis module 307 includes:

[0163] The second calling submodule is used to call a preset scoring mapping table based on the agent;

[0164] The query submodule is used to query the specified rating range that matches the target rating data from the rating mapping table;

[0165] The fourth acquisition submodule is used to acquire customer group information corresponding to the specified rating range;

[0166] The fourth determining submodule is used to use the customer group information as the customer group result of the customer.

[0167] In this embodiment, the operations performed by the above modules or units correspond one-to-one with the steps of the data evaluation method in the aforementioned embodiments, and will not be repeated here.

[0168] In some optional implementations of this embodiment, the data evaluation apparatus further includes:

[0169] The fourth acquisition module is used to acquire preset information parsing logic;

[0170] The parsing module is used to perform information parsing on the target score data and the target derived feature data based on the information parsing logic to obtain the corresponding parsing results;

[0171] The generation module is used to generate service suggestions corresponding to the customer based on the parsing results;

[0172] The output module is used to process the service recommendations.

[0173] In this embodiment, the operations performed by the above modules or units correspond one-to-one with the steps of the data evaluation method in the aforementioned embodiments, and will not be repeated here.

[0174] To address the aforementioned technical problems, embodiments of this application also provide a computer device. Please refer to [link / reference needed]. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0175] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected via a system bus. It should be noted that only the computer device 4 with components 41-43 is shown in the figure; however, it should be understood that it is not required to implement all the shown components, and more or fewer components can be implemented alternatively. Those skilled in the art will understand that the computer device described here is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0176] The computer device can be a desktop computer, laptop, handheld computer, or cloud server, etc. The computer device can interact with the user via a keyboard, mouse, remote control, touchpad, or voice control.

[0177] The memory 41 includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 41 may be an internal storage unit of the computer device 4, such as the hard disk or memory of the computer device 4. In other embodiments, the memory 41 may also be an external storage device of the computer device 4, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the computer device 4. Of course, the memory 41 may also include both the internal storage unit and its external storage device of the computer device 4. In this embodiment, the memory 41 is typically used to store the operating system and various application software installed on the computer device 4, such as computer-readable instructions for data evaluation methods. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or will be output.

[0178] In some embodiments, the processor 42 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip. The processor 42 is typically used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to execute computer-readable instructions stored in the memory 41 or to process data, such as executing computer-readable instructions for the data evaluation method.

[0179] The network interface 43 may include a wireless network interface or a wired network interface, which is typically used to establish communication connections between the computer device 4 and other electronic devices.

[0180] Compared with the prior art, the embodiments of this application have the following beneficial effects:

[0181] In this embodiment, traditional factor feature data of customers is obtained, and derived feature data of customers is obtained based on information type. Then, the traditional factor feature data and derived feature data are preprocessed to obtain corresponding target traditional factor feature data and target derived feature data. Then, the target traditional factor feature data is predicted based on the first prediction model included in the fusion prediction model to obtain a preliminary prediction result. Then, the preliminary prediction result and the derived feature data are predicted based on the second prediction model included in the fusion prediction model to obtain a target prediction result. Subsequently, the target prediction result is processed by score conversion to obtain target score data. Finally, the target score data is analyzed and processed based on the use of the intelligent agent to achieve automatic and accurate generation of customer group results, which improves the processing efficiency of customer group segmentation and ensures the accuracy of the obtained customer group results.

[0182] This application also provides another embodiment, namely, providing a computer-readable storage medium storing computer-readable instructions that can be executed by at least one processor to cause the at least one processor to perform the steps of the data evaluation method described above.

[0183] Compared with the prior art, the embodiments of this application have the following main advantages:

[0184] In this embodiment, traditional factor feature data of customers is obtained, and derived feature data of customers is obtained based on information type. Then, the traditional factor feature data and derived feature data are preprocessed to obtain corresponding target traditional factor feature data and target derived feature data. Then, the target traditional factor feature data is predicted based on the first prediction model included in the fusion prediction model to obtain a preliminary prediction result. Then, the preliminary prediction result and the derived feature data are predicted based on the second prediction model included in the fusion prediction model to obtain a target prediction result. Subsequently, the target prediction result is processed by score conversion to obtain target score data. Finally, the target score data is analyzed and processed based on the use of the intelligent agent to achieve automatic and accurate generation of customer group results, which improves the processing efficiency of customer group segmentation and ensures the accuracy of the obtained customer group results.

[0185] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0186] Obviously, the embodiments described above are only some embodiments of this application, not all embodiments. The accompanying drawings show preferred embodiments of this application, but do not limit the patent scope of this application. This application can be implemented in many different forms; rather, the purpose of providing these embodiments is to provide a more thorough and comprehensive understanding of the disclosure of this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this application's specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the scope of patent protection of this application.

Claims

1. A data evaluation method, characterized in that, Includes the following steps: Obtain the customer's traditional factor characteristic data, and obtain the customer's derived characteristic data based on a preset information type; The traditional factor feature data is preprocessed to obtain the corresponding target traditional factor feature data, and the derived feature data is preprocessed to obtain the corresponding target derived feature data; Invoke a preset fusion prediction model; wherein the fusion prediction model includes a first prediction model and a second prediction model; Based on the first prediction model, the target traditional factor feature data are predicted and processed to obtain the corresponding preliminary prediction results; Based on the second prediction model, the preliminary prediction results and the derived feature data are processed to obtain the corresponding target prediction results; The target prediction results are subjected to a scoring conversion process to obtain the corresponding target score data; The target rating data is analyzed and processed based on a preset intelligent agent to generate the customer group results for the customer.

2. The data evaluation method according to claim 1, characterized in that, The step of obtaining the customer's derived feature data based on a preset information type specifically includes: Obtain the customer profile corresponding to the customer; Call the preset target large language model and obtain the preset large model prompt words; Based on the target large language model, the customer profile is inferred according to the prompt words of the large model to obtain the inference result corresponding to the information type; The reasoning result is used as the customer's derived feature data.

3. The data evaluation method according to claim 1, characterized in that, Before the step of invoking the preset fusion prediction model, the following is also included: Obtain pre-collected historical customer factor characteristic data; Construct corresponding sample data based on the aforementioned factor feature data; Invoke the preset initial fusion model; Obtain the preset optimization objective function and parameter adjustment strategy; Based on the optimization objective function and the parameter adjustment strategy, the initial fusion model is trained and evaluated using the sample data until a generated model that meets the preset construction requirements is obtained. The generative model is used as the fusion prediction model.

4. The data evaluation method according to claim 1, characterized in that, The step of performing prediction processing on the preliminary prediction result and the derived feature data based on the second prediction model to obtain the corresponding target prediction result specifically includes: Obtain the preset fusion strategy; Based on the fusion strategy, the preliminary prediction results and the target derived feature data are fused to obtain the corresponding fused data; Based on the second prediction model, the fused data is processed according to a preset specified processing method to obtain the corresponding specified prediction result; The specified prediction result is used as the target prediction result.

5. The data evaluation method according to claim 1, characterized in that, The step of performing a scoring conversion process on the target prediction result to obtain the corresponding target scoring data specifically includes: Invoke the preset rating conversion rules; Based on the scoring conversion rules, the target prediction results are converted to obtain the corresponding converted data. The transformed data is used as the target score data.

6. The data evaluation method according to claim 1, characterized in that, The step of analyzing and processing the target rating data based on a preset intelligent agent to generate the customer group results for the customer specifically includes: The agent invokes a preset scoring mapping table; Retrieve the specified rating range that matches the target rating data from the rating mapping table; Obtain customer group information corresponding to the specified rating range; The customer group information is used as the customer group result for the customer.

7. The data evaluation method according to claim 1, characterized in that, After the step of performing a scoring conversion process on the target prediction result to obtain the corresponding target score data, the method further includes: Obtain the preset information parsing logic; Based on the information parsing logic, the target score data and the target derived feature data are parsed to obtain the corresponding parsing results; Based on the analysis results, service suggestions corresponding to the customer are generated. The service recommendations are then processed for output.

8. A data evaluation device, characterized in that, include: The first acquisition module is used to acquire the customer's traditional factor characteristic data and acquire the customer's derived characteristic data based on a preset information type; The preprocessing module is used to preprocess the traditional factor feature data to obtain the corresponding target traditional factor feature data, and to preprocess the derived feature data to obtain the corresponding target derived feature data. The first calling module is used to call a preset fusion prediction model; wherein, the fusion prediction model includes a first prediction model and a second prediction model; The first prediction module is used to perform prediction processing on the target traditional factor feature data based on the first prediction model to obtain the corresponding preliminary prediction results; The second prediction module is used to perform prediction processing on the preliminary prediction result and the derived feature data based on the second prediction model to obtain the corresponding target prediction result. The conversion module is used to perform a score conversion process on the target prediction result to obtain the corresponding target score data; The analysis module is used to analyze and process the target rating data based on a preset intelligent agent to generate the customer group results for the customer.

9. A computer device, characterized in that, The system includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the data evaluation method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-readable instructions that, when executed by a processor, implement the steps of the data evaluation method as described in any one of claims 1 to 7.

Images