Data prediction method and device based on artificial intelligence, computer equipment and medium

Abstract

The invention belongs to the technical field of artificial intelligence, and relates to a data prediction method and device based on artificial intelligence, computer equipment and a storage medium, and the method comprises the steps: obtaining pre-collected user data of a target user; preprocessing the user data to obtain corresponding target processing data; performing data screening on the target processing data based on a preset business target to obtain screened behavior data; performing feature extraction on the behavior data based on a preset feature selection strategy to obtain corresponding feature data; calling a target behavior prediction model corresponding to the business target; performing prediction processing on the feature data based on the target behavior prediction model to obtain a corresponding behavior prediction result; and carrying out output processing on the behavior prediction result. In addition, the behavior prediction result can be stored in the block chain. The method can be applied to a behavior prediction scene in the field of financial science and technology, effectively improves the processing efficiency of behavior prediction, and guarantees the accuracy of an obtained behavior prediction result.

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 prediction methods, devices, computer equipment and storage media based on artificial intelligence. Background Technology

[0002] In traditional insurance service models, customer behavior prediction mainly relies on static rules or basic statistical analysis, resulting in insufficient prediction accuracy and timeliness. Specifically, traditional methods typically infer behavior based on customers' historical transaction records or simple demographic characteristics (such as age and region). This extensive prediction approach is ill-suited to the complexity of insurance customer behavior and the rapid changes in the market environment, causing insurance companies to be unable to complete customer behavior prediction processing in a timely and accurate manner, thereby affecting resource allocation efficiency and customer conversion rates.

[0003] For example, in the property insurance sector, traditional methods may only recommend renewal products based on a customer's historical insured amounts and types of insurance, without dynamically analyzing changes in the customer's business situation or market fluctuations. If the customer is a manufacturing company that has recently added a high-risk production line, traditional methods may fail to capture changes in risk characteristics and recommend property insurance products with insufficient coverage, resulting in a lack of protection for the customer in the event of a risk event. This predictive bias not only reduces customer satisfaction but may also cause insurance companies to lose their competitive edge in the market due to service lag.

[0004] Therefore, there is an urgent need to provide an intelligent and highly accurate method for predicting insurance customer behavior, so as to provide insurance companies with precise market analysis and strategy support. Summary of the Invention

[0005] The purpose of this application is to propose a data prediction method, apparatus, computer device, and storage medium based on artificial intelligence, in order to solve the technical problem that existing customer behavior prediction methods mainly rely on static rules or basic statistical analysis, which cannot complete the prediction and processing of customer behavior in a timely and accurate manner, resulting in low efficiency and low accuracy in customer behavior prediction.

[0006] Firstly, an artificial intelligence-based data prediction method is provided, including:

[0007] Acquire user data from pre-collected target users;

[0008] The user data is preprocessed to obtain the corresponding target processing data;

[0009] Based on preset business objectives, the target processing data is filtered to obtain filtered behavioral data.

[0010] Based on a preset feature selection strategy, feature extraction is performed on the behavioral data to obtain the corresponding feature data.

[0011] Invoke the target behavior prediction model corresponding to the stated business objective;

[0012] Based on the target behavior prediction model, the feature data is processed to obtain the corresponding behavior prediction result;

[0013] The predicted behavior results are then processed for output.

[0014] Secondly, an artificial intelligence-based data prediction device is provided, comprising:

[0015] The first acquisition module is used to acquire user data of the target users that have been collected in advance.

[0016] The preprocessing module is used to preprocess the user data to obtain the corresponding target processing data;

[0017] The first filtering module is used to filter the target processing data based on preset business objectives to obtain filtered behavioral data.

[0018] The extraction module is used to extract features from the behavioral data based on a preset feature selection strategy to obtain corresponding feature data;

[0019] The first calling module is used to call the target behavior prediction model corresponding to the business objective;

[0020] The prediction module is used to perform prediction processing on the feature data based on the target behavior prediction model to obtain the corresponding behavior prediction result;

[0021] The output module is used to process the output of the behavior prediction results.

[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 artificial intelligence-based data prediction 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 aforementioned artificial intelligence-based data prediction method.

[0024] In the aforementioned scheme implemented by the AI-based data prediction method, apparatus, computer equipment, and storage medium, the following steps are taken: First, user data of the target users is acquired in advance; then, the user data is preprocessed to obtain corresponding target processing data; then, the target processing data is filtered based on a preset business objective to obtain filtered behavioral data; next, features are extracted from the behavioral data based on a preset feature selection strategy to obtain corresponding feature data; subsequently, a target behavior prediction model corresponding to the business objective is invoked; and the feature data is predicted based on the target behavior prediction model to obtain a corresponding behavior prediction result; finally, the behavior prediction result is output. Based on the above automated processing flow, this application acquires user data of the target users in advance, preprocesses the user data to obtain target processing data, then filters the target processing data based on the business objective to obtain filtered behavioral data, then extracts features from the behavioral data based on a feature selection strategy to obtain feature data, then invokes a target behavior prediction model corresponding to the business objective, and predicts the feature data based on the target behavior prediction model to obtain a behavior prediction result, and finally outputs the behavior prediction result. Thus, unlike existing behavior prediction methods that rely on static rules or basic statistical analysis, this application uses a target behavior prediction model to predict and process relevant data of target users, which can efficiently and accurately complete the behavior prediction processing of target users, improve the processing efficiency of behavior prediction, and ensure the accuracy of the obtained behavior prediction 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 2 This is a flowchart of an embodiment of the artificial intelligence-based data prediction method according to this application;

[0028] Figure 3 This is a schematic diagram of a structure of an embodiment of the artificial intelligence-based data prediction device 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 prediction method based on artificial intelligence provided in the embodiments of this application is generally executed by a server / terminal device, and correspondingly, the data prediction device based on artificial intelligence 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 This document illustrates a flowchart of an embodiment of the AI-based data prediction method according to this application. The order of steps in the flowchart can be changed, and some steps can be omitted, depending on different needs. The AI-based data prediction method provided in this application can be applied to any scenario requiring behavior prediction, and thus can be applied to products in these scenarios, such as behavior prediction in the financial insurance field. The AI-based data prediction method includes the following steps:

[0040] Step S201: Obtain user data of the target users that have been collected in advance.

[0041] In this embodiment, the artificial intelligence-based data prediction method operates on an electronic device (e.g., Figure 1The server / terminal device shown can acquire user data of the target user 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-wideband) connections, and other currently known or future-developed wireless connection methods. The executing entity of this application is specifically a data prediction system, which can be simply referred to as the system. This application can be applied to behavioral prediction scenarios in the fintech field. User data matching the aforementioned target user can be acquired from multiple channels such as the insurance company's core business system, customer relationship management system (CRM), and customer service system. For example, the core business system stores customer purchase records, the CRM system contains basic customer information, and the customer service system records customer service information. Customer data includes basic customer information such as age, gender, occupation, marital status, and income level; purchase records, including the type of product purchased (such as home insurance, accident insurance, liability insurance, etc.), purchase time, purchase amount, and purchase channel; and service records, such as the number of complaints, reasons for complaints, number of inquiries, and content of inquiries.

[0042] Step S202: Preprocess the user data to obtain the corresponding target processing data.

[0043] In this embodiment, the specific implementation process of preprocessing the user data to obtain the corresponding target processing data will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0044] Step S203: Based on the preset business objectives, the target processing data is filtered to obtain the filtered behavioral data.

[0045] In this embodiment, the aforementioned business objectives may include the objectives of purchase intention prediction (i.e., determining the customer's potential interest in a product), purchase cycle prediction (i.e., predicting the time window for the customer's next purchase), or purchase amount prediction (i.e., predicting the range of amounts a customer will spend in a single purchase). Specifically, based on different prediction needs (purchase intention, purchase cycle, purchase amount), user data from different dimensions will be intelligently collected and integrated, and features will be dynamically adjusted in conjunction with real-time behavioral data to improve the accuracy and business applicability of subsequent behavioral predictions.

[0046] Specifically, if the aforementioned business objectives fall under the category of purchase intention prediction, the filtered behavioral data obtained from processing the target data may include: Behavioral signals: recent interaction frequency (e.g., number of times the app has been logged in or product pages have been viewed in the past 7 days); proactive behaviors (e.g., clicking "Contact Customer Service," "Favorite Product," "Add to Cart"); page dwell time (e.g., browsing a home insurance details page for more than 3 minutes). Historical preferences: types of products previously purchased (e.g., customers who have already purchased car insurance may be more interested in home insurance); preferred channels (e.g., preference for purchasing through the app or offline consultation). External data: credit scores (high-credit customers may be more sensitive to protection-related products); social media behavior (e.g., customers who follow insurance-related topics have a higher willingness to purchase).

[0047] If the above business objectives fall under the category of purchase cycle prediction, the filtered behavioral data obtained from processing the target data may include: Time series characteristics: historical purchase intervals (e.g., if the last car insurance purchase was 11 months ago, the predicted home insurance purchase cycle may refer to the car insurance cycle); behavioral time distribution (e.g., customers typically purchase insurance at the end of a quarter). Lifecycle stages: policy expiration reminders (e.g., the probability of purchasing home insurance may increase one month before car insurance expires); family event triggers (e.g., increased protection needs may occur after marriage or childbirth). External events: regional disaster warnings (e.g., the home insurance purchase cycle may shorten after a typhoon warning).

[0048] If the above business objectives fall under the category of purchase amount prediction, the filtered behavioral data obtained from processing the target data may include: Affordability: Income level (high-income customers may choose higher coverage amounts); Historical spending amounts (e.g., average annual premiums for past car insurance purchases). Product preferences: Selected supplementary insurance types (e.g., whether "water damage insurance" or "theft insurance" is selected); Coverage amount selection habits (e.g., preference for basic coverage or high-end protection). Promotion sensitivity: Past responses to discounts (e.g., whether purchase amounts increased due to "20% off the first year").

[0049] Step S204: Extract features from the behavioral data based on a preset feature selection strategy to obtain corresponding feature data.

[0050] In this embodiment, the aforementioned feature selection strategy refers to using the same method as the feature selection and extraction processing of historical behavior data during the model training process of the aforementioned target behavior prediction model to extract features from the aforementioned behavior data and obtain the corresponding feature data.

[0051] Step S205: Invoke the target behavior prediction model corresponding to the business objective.

[0052] In this embodiment, the specific construction process of the target behavior prediction model will be described in more detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0053] Step S206: Based on the target behavior prediction model, perform prediction processing on the feature data to obtain the corresponding behavior prediction result.

[0054] In this embodiment, by inputting the aforementioned feature data into the target behavior prediction model, the target behavior prediction model will predict the key behavioral indicators of the target user based on the feature data and output the corresponding behavior prediction results.

[0055] Specifically, if the aforementioned business objective falls under the category of purchase intention prediction, then the target behavior prediction model is a model with purchase intention prediction capabilities. This model outputs the probability value of a target user purchasing non-auto insurance products within a future period based on the learned data patterns. For example, the model outputs a 70% probability that a customer will purchase home insurance within the next month. Furthermore, based on the magnitude of the predicted probability, the target user is categorized into different purchase intention levels, such as high purchase intention (probability greater than 70%), medium purchase intention (probability between 30% and 70%), and low purchase intention (probability less than 30%), in order to subsequently develop targeted business processing strategies, such as sales strategies.

[0056] If the aforementioned business objective falls under the category of purchase cycle prediction, then the target behavior prediction model is one with purchase cycle prediction capabilities. This model predicts the timing or interval of a target user's next purchase based on their historical purchase times and characteristic data. For example, by analyzing a customer's purchase records over the past year, the model predicts that the customer's next purchase of accident insurance will be in 6 months. Furthermore, considering the uncertainty of customer purchasing behavior, a prediction range for the purchase cycle can be provided. For instance, it is highly probable that the customer's next purchase will occur between 5 and 7 months.

[0057] If the aforementioned business objective falls under the category of purchase amount prediction, then the target behavior prediction model is one with purchase amount prediction capabilities. This model predicts the range of amounts a target user will spend on their next product purchase. For example, it might predict that a customer will purchase liability insurance in the range of 3,000-5,000 yuan. Furthermore, in addition to the amount range, it can also predict the target user's average purchase amount. Through the analysis of a large amount of customer data and the model's calculations, the average amount a target user spends on non-motor insurance products each time can be determined.

[0058] Step S207: Output the behavior prediction results.

[0059] In this embodiment, the generated behavior prediction results can be sent to the business processing personnel related to the target user to complete the output processing of the behavior prediction results, so that the business processing personnel can provide corresponding business services to the target user based on the obtained behavior prediction results.

[0060] This application first acquires pre-collected user data of target users; then preprocesses the user data to obtain corresponding target processing data; next, it filters the target processing data based on preset business objectives to obtain filtered behavioral data; then, it extracts features from the behavioral data based on preset feature selection strategies to obtain corresponding feature data; subsequently, it calls a target behavior prediction model corresponding to the business objective; and then performs prediction processing on the feature data based on the target behavior prediction model to obtain corresponding behavior prediction results; finally, it outputs the behavior prediction results. Based on the above automated processing flow, this application acquires pre-collected user data of target users, preprocesses the user data to obtain target processing data, then filters the target processing data based on business objectives to obtain filtered behavioral data, then extracts features from the behavioral data based on feature selection strategies to obtain feature data, then calls a target behavior prediction model corresponding to the business objective, and performs prediction processing on the feature data based on the target behavior prediction model to obtain behavior prediction results, and finally outputs the behavior prediction results. Thus, unlike existing behavior prediction methods that rely on static rules or basic statistical analysis, this application uses a target behavior prediction model to predict and process relevant data of target users, which can efficiently and accurately complete the behavior prediction processing of target users, improve the processing efficiency of behavior prediction, and ensure the accuracy of the obtained behavior prediction results.

[0061] In some alternative implementations, prior to step S205, the electronic device may also perform the following steps:

[0062] Obtain pre-collected historical behavioral data corresponding to the stated business objectives.

[0063] In this embodiment, customer data for all insurance customers within a historical time period is obtained from multiple channels, including the insurance company's core business system, customer relationship management system (CRM), and customer service system. For example, the core business system stores customer purchase records, the CRM system contains basic customer information, and the customer service system records customer service details. The customer data includes basic customer information such as age, gender, occupation, marital status, and income level; purchase records, including the type of product purchased (e.g., home insurance, accident insurance, liability insurance), purchase time, purchase amount, and purchase channel; and service records, such as the number of complaints, reasons for complaints, number of inquiries, and content of inquiries. The specific value for the historical time period is not limited and can be set according to actual business needs; for example, it can be set to the past three years.

[0064] Furthermore, based on the aforementioned business objectives, the customer data is preprocessed and filtered, and the resulting processed data is used as the aforementioned historical behavior data. The process of generating this historical behavior data can be described in detail below, referring to the process of preprocessing the user data to obtain target processed data, and then filtering the target processed data based on preset business objectives to obtain the filtered behavior data.

[0065] The historical behavior data is processed by feature selection and extraction to obtain corresponding feature sample data.

[0066] In this embodiment, the specific implementation process of performing feature selection and extraction on the historical behavior data to obtain the corresponding feature sample data will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0067] Based on the feature sample data, multiple preset initial models of different types are trained to obtain multiple corresponding trained specified prediction models.

[0068] In this embodiment, the initial model mentioned above is a pre-built model including but not limited to random forest model, gradient boosting tree (GBDT) model and deep neural network (DNN) model.

[0069] Specifically, the process of training a random forest model using feature sample data to obtain a trained first-order prediction model includes: 1) Data preparation. Using processed feature sample data, separate features (such as purchase frequency, amount, product type, etc.) from labels (purchase intention / amount level). Perform one-hot encoding on categorical variables to ensure all features are numerical. Divide the training and validation sets proportionally (e.g., 80% / 20%) to maintain a balanced class distribution (stratified sampling). 2) Model initialization. Set random forest parameters: initialize the number of trees (n_estimators) to 100-500 and fine-tune through cross-validation. Set the maximum depth of a single tree (max_depth) to 5-15 to avoid overfitting. Enable out-of-bag (OOB) evaluation to monitor generalization performance. 3) Training and tuning. Fit the model to the training set. Generate a subset of data for each tree through bootstrap sampling and randomly select feature subsets for node splitting. Use grid search to tune hyperparameters (such as max_features, min_samples_split) with the F1 score (classification) or MSE (regression) on the validation set as the optimization objective. Record the out-of-bag error (OOB Error); if it does not decrease for several consecutive rounds, terminate training early (early stopping mechanism). 4) Validation and Output. Evaluate the model performance on the validation set, output the feature importance ranking, and identify key features (such as "historical purchase amount" having the highest weight). Save the model corresponding to the best parameter combination as the first designated prediction model for subsequent predictions.

[0070] The process of training a gradient boosting tree model using feature sample data to obtain a trained first specified prediction model includes: 1) Data preparation. Continuous features in the feature sample data are binned (e.g., equal-frequency binning) and converted to categorical types to enhance model robustness. Numerical features are standardized (Z-Score standardization) to ensure a balanced contribution of features of different scales to the split. Training and validation sets are divided, and K-fold cross-validation (e.g., 5-fold) is used to reduce the impact of data fluctuations. 2) Model initialization. A GBDT framework (e.g., XGBoost / LightGBM) is selected, and basic parameters are set: the learning rate is initially 0.01-0.1 and gradually tuned. The maximum tree depth is set to 3-8 to limit the complexity of a single tree. Early stopping rounds (early_stopping_rounds = 50) are enabled to monitor validation set metrics. 3) Training and tuning. A gradient boosting strategy is used, fitting the residuals of the previous round in each iteration, and optimizing the loss function (e.g., logarithmic loss / mean squared error) through the negative gradient direction. Use Bayesian optimization to adjust key parameters (such as subsample and colsample_bytree) to balance bias and variance. Monitor validation set metrics (such as AUC / RMSE) during training; terminate training if there is no improvement after 10 consecutive epochs. 4) Validation and Output. Analyze feature gain and split count, and remove low-contribution features to simplify the model. Save the model corresponding to the early stopping epochs to ensure optimal generalization ability, and use it as the corresponding second specified prediction model.

[0071] The process of training a deep neural network using feature sample data to obtain a pre-trained, specified prediction model includes: 1) Data preparation. Encode high-dimensional sparse features (such as categorical variables) in the feature sample data using embedding layers to reduce dimensionality. Standardize continuous features to [0,1] or normalize them using Batch Normalization. Divide the data into training, validation, and test sets (e.g., 60% / 20% / 20%) to ensure consistent data distribution. 2) Model architecture design. Construct a multilayer perceptron (MLP). Example structure: Input layer: Number of nodes = feature dimension (e.g., 50 dimensions). Hidden layers: 2-3 layers, 128-256 nodes per layer, ReLU activation function. Output layer: Single node (regression task) or Softmax multi-node (classification task). Add Dropout layers (rate = 0.3-0.5) and L2 regularization (λ = 0.001) to prevent overfitting. 3) Training and tuning. Using the Adam optimizer with an initial learning rate of 0.001, and a learning rate scheduler (such as ReduceLROnPlateau), the batch size is set to 32-128, employing mini-batch gradient descent. The validation set loss is monitored; if it fails to decrease for five consecutive rounds, an early stopping mechanism is activated. The loss curve and weight distribution are visualized using Tensor Boards to diagnose the training process. 4) Validation and Output. The model performance is evaluated on the test set, comparing the improvement to the baseline model (such as logistic regression). SHAP values ​​are used to explain the contributions of key features, ensuring model interpretability. The trained weights and model structure are saved as the corresponding third-party prediction model and deployed as an online service.

[0072] Based on a preset model selection strategy, a target prediction model is selected from all the specified prediction models.

[0073] In this embodiment, the specific implementation process of selecting the target prediction model from all the specified prediction models based on the preset model screening strategy will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0074] The target prediction model is used as the target behavior prediction model corresponding to the business objective.

[0075] This application acquires pre-collected historical behavior data corresponding to the business objective; then performs feature selection and extraction processing on the historical behavior data to obtain corresponding feature sample data; subsequently, it trains multiple preset initial models of different types based on the feature sample data to obtain multiple trained specified prediction models; subsequently, based on a preset model selection strategy, it selects a target prediction model from all the specified prediction models; finally, it uses the target prediction model as the target behavior prediction model corresponding to the business objective. Based on the above processing flow, this application acquires pre-collected historical behavior data corresponding to the business objective, performs feature selection and extraction processing on the historical behavior data to obtain feature sample data, then trains multiple preset initial models of different types based on the feature sample data to obtain multiple trained specified prediction models, and then selects a target prediction model from all the specified prediction models based on a model selection strategy and uses it as the required target behavior prediction model corresponding to the business objective. Since different machine learning models have different characteristics and applicable scenarios, this application, by constructing multiple specified prediction models and comparing and selecting them based on a model selection strategy, can select the target behavior prediction model most suitable for the current data and business objective, effectively improving the selection accuracy of the target behavior prediction model.

[0076] In some optional implementations of this embodiment, the step of performing feature selection and extraction processing on the historical behavior data to obtain corresponding feature sample data includes the following steps:

[0077] Based on a preset statistical analysis strategy, feature selection processing is performed on the historical behavior data to obtain the corresponding first feature data.

[0078] In this embodiment, the aforementioned statistical analysis strategy refers to a strategy that combines statistical analysis methods and machine learning algorithms for selection. Specifically, the statistical analysis methods include: Correlation analysis: Calculating the correlation coefficient between each feature and the prediction target (such as purchase intention, purchase cycle, purchase amount), such as the Pearson correlation coefficient. Selecting features with high correlation to the prediction target. For example, correlation analysis may reveal a positive correlation between customer purchase frequency and purchase intention; that is, the higher the purchase frequency, the stronger the purchase intention. Therefore, purchase frequency is an important feature. Chi-square test: For categorical features and prediction targets, the chi-square test is used to determine whether there is a significant association between them. For example, testing whether there is a significant relationship between a customer's occupation (categorical feature) and the type of product purchased (prediction target). If the chi-square test result is significant, then the occupation feature is important for predicting the type of product a customer will purchase.

[0079] Machine learning algorithms assist in feature selection, including: Using decision tree algorithms: A decision tree model is built, and features are selected based on their importance scores within the decision tree. A higher importance score indicates a greater contribution of that feature to the prediction objective. For example, in a decision tree model predicting customer purchase amounts, if the feature "product type" has a high importance score, it is retained as an important feature. Using random forest algorithms: Random forest algorithms can also output feature importance rankings. By building multiple decision trees and integrating their predictions, the average importance score of each feature across all decision trees is calculated, thus selecting the more important features.

[0080] The first feature data is subjected to feature dimensionality reduction processing to obtain the corresponding second feature data.

[0081] In this embodiment, the aforementioned feature dimensionality reduction processing refers to feature dimensionality reduction based on Principal Component Analysis (PCA). PCA projects the original data onto a new coordinate system through linear transformation; the new coordinate axes are called principal components. The principal components are sorted according to their variance; the larger the variance, the more information the principal component contains. Specifically, the number of principal components to retain is determined based on the cumulative variance contribution rate. Typically, principal components with a cumulative variance contribution rate of 80%-90% or higher are selected, thus reducing the dimensionality of the data while retaining most of the information. For example, if calculations show that the cumulative variance contribution rate of the top 5 principal components reaches 85%, then these 5 principal components are selected to represent the original data.

[0082] The second feature data is subjected to recursive feature elimination processing to obtain the corresponding third feature data.

[0083] In this embodiment, the recursive feature elimination process includes: Initial model construction: Constructing an initial machine learning model, such as a random forest model, using all selected features. Feature importance evaluation: Evaluating the importance of each feature based on the model's output. Feature removal: Removing the features with the lowest importance, and then reconstructing the model using the remaining features. Iterative process: Repeating the above steps until a preset number of features is reached or the model's performance no longer significantly improves. The final optimal feature subset is the feature set used for subsequent model training, i.e., the aforementioned feature sample data.

[0084] The third feature data is used as the feature sample data.

[0085] In this embodiment, by selecting and extracting features that significantly impact the prediction target (business objective) from massive amounts of data, the complexity and computational load can be reduced, improving the training efficiency and predictive performance of the model. By combining statistical analysis methods with machine learning algorithms, important features can be screened more comprehensively and accurately. Feature dimensionality reduction and optimal feature subset selection based on recursive feature elimination further optimize the feature set, enabling the model to focus more on key factors, thereby improving the accuracy and reliability of predictions.

[0086] This application uses a preset statistical analysis strategy to perform feature selection processing on the historical behavior data to obtain corresponding first feature data; then, it performs feature dimensionality reduction processing on the first feature data to obtain corresponding second feature data; subsequently, it performs recursive feature elimination processing on the second feature data to obtain corresponding third feature data; and finally, it uses the third feature data as the feature sample data. Based on the above processing flow, this application uses a statistical analysis strategy to perform feature selection processing on historical behavior data to obtain first feature data, then performs feature dimensionality reduction processing on the first feature data to obtain second feature data, then performs recursive feature elimination processing on the second feature data, and uses the obtained third feature data as the required feature sample data. This enables comprehensive and accurate screening of important features, improving the accuracy and intelligence of the generated feature sample data.

[0087] In some optional implementations, the step of selecting the target prediction model from all the specified prediction models based on a preset model selection strategy includes the following steps:

[0088] Obtain the evaluation index data for each of the specified prediction models.

[0089] In this embodiment, evaluation metrics corresponding to the aforementioned business objectives are predetermined. Then, based on these evaluation metrics, evaluation metric data for each specified prediction model are calculated using a pre-defined validation set. Specifically, if the task corresponding to the aforementioned business objective is a classification task, accuracy, recall, and F1 score are used as the corresponding evaluation metrics. If the task corresponding to the aforementioned business objective is a regression task, mean squared error (MSE) and mean absolute error (MAE) are used as the corresponding evaluation metrics.

[0090] The evaluation index data of each specified prediction model are calculated and processed based on a preset performance calculation strategy to obtain the performance score of each specified prediction model.

[0091] In this embodiment, the performance calculation strategy described above can specifically adopt a weighted summation strategy. According to the actual processing requirements, the weight coefficients corresponding to each evaluation index are pre-set. Then, the evaluation index data of each specified prediction model are weighted and summed with the weight coefficients, and the calculated results are used as the performance scores of each specified prediction model.

[0092] All the performance scores are numerically compared to select the target performance score with the highest value.

[0093] In this embodiment, all the obtained performance scores are compared numerically, and the target performance score with the highest value is selected from all the performance scores based on the numerical comparison results.

[0094] Obtain the target model corresponding to the target performance score from all the specified prediction models, and use the target model as the target prediction model.

[0095] In this embodiment, the target model is the prediction model that corresponds to the target performance score among all the specified prediction models.

[0096] This application obtains evaluation index data for each specified prediction model; then, based on a preset performance calculation strategy, it calculates and processes the evaluation index data of each specified prediction model to obtain a performance score for each specified prediction model; subsequently, it compares all the performance scores to select the target performance score with the highest value; and then, it obtains the target model corresponding to the target performance score from all the specified prediction models and uses this target model as the target prediction model. Based on the above processing flow, this application obtains evaluation index data for each specified prediction model, calculates and processes the evaluation index data of each specified prediction model based on a performance calculation strategy to obtain a performance score for each specified prediction model, then compares all the performance scores to select the target performance score with the highest value, and then uses the target model corresponding to the target performance score obtained from all the specified prediction models as the required target prediction model. This enables intelligent and accurate model comparison and selection of multiple specified prediction models, ensuring that the final selected target prediction model achieves a balance between performance and practicality, thus improving the accuracy and intelligence of target prediction model selection.

[0097] In some alternative implementations, step S202 includes the following steps:

[0098] The user data is cleaned to obtain the corresponding first processed data.

[0099] In this embodiment, the data cleaning process includes: removing duplicate data: identifying and deleting duplicate records by comparing their unique identifiers (such as customer IDs). For example, in purchase records, if the same customer purchases the same product at the same time, through the same channel, and for the same amount, it is determined to be a duplicate record and deleted. Correcting erroneous data: correcting data that is clearly illogical. For example, data with negative ages or ages exceeding a reasonable range (such as over 150 years old) are reasonably estimated and corrected based on other relevant information (such as occupation, type of product purchased, etc.).

[0100] Missing values ​​are filled into the first processed data to obtain the corresponding second processed data.

[0101] In this embodiment, the missing value imputation process includes: analyzing the cause of the missing data: understanding whether the missing data is due to system failure, customer non-compliance, or other reasons. For example, the missing income level in a customer's basic information may be because the customer is unwilling to disclose it. Selecting an imputation method: For numerical data, such as age or purchase amount, the mean, median, or mode can be used for imputation. For example, for a customer with a missing age, the average age of all customers is calculated, and this average is used to imput the missing age. For categorical data, such as occupation or marital status, the category with the highest frequency of occurrence can be selected for imputation based on the data distribution.

[0102] The second processed data is subjected to outlier processing to obtain the corresponding third processed data.

[0103] In this embodiment, the outlier handling includes: identifying outliers: outliers are identified using statistical methods (such as box plots, standard deviation, etc.) or business rules. For example, records where the purchase amount far exceeds the average purchase amount of similar products may be considered outliers. Handling outliers: for suspected outliers, the authenticity of the data is further verified. If it is confirmed to be erroneous data, it is corrected or deleted; if it is a real extreme case, the data can be retained, but special processing methods are adopted in subsequent analysis, such as grouping it separately for analysis.

[0104] The third processed data is used as the target processed data.

[0105] This application performs data cleaning on the user data to obtain corresponding first processed data; then, it performs missing value imputation on the first processed data to obtain corresponding second processed data; subsequently, it performs outlier handling on the second processed data to obtain corresponding third processed data; and finally, it uses the third processed data as the target processed data. Based on the above processing flow, this application, by performing data cleaning, missing value imputation, and outlier handling on user data, can efficiently and accurately complete the preprocessing of user data, which is beneficial for providing clean, complete, and accurate target processed data for the target behavior prediction model, thereby effectively improving the reliability and effectiveness of data prediction.

[0106] In some optional implementations of this embodiment, after step S207, the electronic device may further perform the following steps:

[0107] Obtain the preset clustering strategy.

[0108] In this embodiment, the above-mentioned grouping strategy includes classifying customers with high purchase intention / short purchase cycle / large purchase amount as high-level customers, and customers with low purchase intention / long purchase cycle / small purchase amount as low-level customers. High purchase intention refers to a purchase probability greater than a preset probability threshold; short purchase cycle refers to a purchase interval less than a preset interval threshold; and large purchase amount refers to an average purchase amount greater than a preset amount threshold. Furthermore, the values ​​of the above-mentioned probability threshold, interval threshold, and amount threshold are not specifically limited and can be set according to actual business needs.

[0109] The behavior prediction results are analyzed based on the segmentation strategy to generate target customer groups corresponding to the target users.

[0110] In this embodiment, the behavior prediction results can be analyzed based on the strategy content of the above-mentioned grouping strategy, and then the target users can be grouped according to the analysis results, and the target customer groups (high-end customers or low-end customers) corresponding to the target users can be determined.

[0111] Obtain the target product push strategy corresponding to the target customer group.

[0112] In this embodiment, the target product push strategy includes the following: for high-end customers, pushing high-end, comprehensive non-motor insurance product packages, such as integrated insurance packages including home insurance, accident insurance, and liability insurance; for low-end customers, pushing some entry-level products with lower prices and simpler coverage to attract their attention. In addition, promotional activities are promptly launched based on customer purchase cycle predictions. For example, for customers with short purchase cycles, promotional information such as discounts and gifts is sent via SMS and email before their expected next purchase date to stimulate purchases.

[0113] Based on the product push strategy, the corresponding product push process is performed on the target user.

[0114] In this embodiment, product push processing corresponding to the target user can be executed based on the strategy content of the above-mentioned product push strategy.

[0115] This application obtains a preset segmentation strategy; then analyzes the behavior prediction results based on the segmentation strategy to generate target customer groups corresponding to the target users; subsequently, it obtains a target product push strategy corresponding to the target customer groups; and then executes corresponding product push processing on the target users based on the product push strategy. Based on the above processing flow, this application analyzes the behavior prediction results based on the use of segmentation strategies to generate target customer groups corresponding to the target users, then obtains target product push strategies corresponding to the target customer groups, and then executes corresponding product push processing on the target users based on the use of the target product push strategies. This allows for the automatic and accurate implementation of personalized product pushes for target users, which is beneficial for improving target user satisfaction and purchase conversion rates, enabling insurance companies to better adapt to market changes, meet customer needs, and enhance market competitiveness.

[0116] In some optional implementations of this embodiment, after step S207, the electronic device may further perform the following steps:

[0117] The behavior prediction results are processed to generate a chart, resulting in the corresponding target chart.

[0118] In this embodiment, various matching charts, such as bar charts, line charts, and pie charts, are generated based on the behavior prediction results output by the aforementioned target behavior prediction model. For example, a bar chart can be used to display the probability distribution of purchase intentions among different customer groups, and the height of the bars can be used to intuitively compare the strength of purchase intentions among different groups; a line chart can be used to display the trend of sales revenue over time, clearly showing the fluctuations in sales performance.

[0119] The behavior prediction results are processed to generate a report, resulting in a corresponding target report.

[0120] In this embodiment, a detailed target report is generated based on the behavior prediction results output by the aforementioned target behavior prediction model. This report presents the behavior prediction results, analysis conclusions, and recommendations in text form. The report may include customer behavior analysis, sales trend forecasting, market opportunity assessment, and other related content, providing comprehensive information for insurance company staff.

[0121] Calls the preset display page.

[0122] In this embodiment, the aforementioned display page is a pre-built visualization page for data display. This display page may include pre-defined chart display areas and report display areas.

[0123] The target chart and target report are visualized based on the display page.

[0124] In this embodiment, the target chart and the target report can be visualized by displaying the target chart in the chart display area of ​​the display page and the target report in the report display area of ​​the display page.

[0125] The system also provides decision support functions, enabling insurance company staff to intuitively understand business dynamics and trends and make more accurate decisions. Specifically, based on behavioral prediction results and visualized information (target charts and target reports), the system automatically recommends appropriate sales strategies. For example, when a decline in sales of a certain product type is predicted, the system recommends adjusting product promotion strategies, such as increasing advertising channels, conducting online marketing activities, or optimizing product combinations to launch more attractive packages. In addition, the system can monitor customer behavior data and credit status in real time. When it detects a decline in a customer's credit rating and reduced purchase intention, it promptly issues a risk warning, reminding staff to pay attention to the customer's risk situation. Staff can then take corresponding measures based on the warning information, such as communicating with the customer and adjusting sales strategies, to mitigate risk.

[0126] This application generates a target chart from the behavior prediction results; then generates a target report from the behavior prediction results; finally, it calls a preset display page; and subsequently, it visualizes the target chart and target report based on the display page. Based on this process, this application improves the intelligence of the behavior prediction result display by using a display page to visualize the generated target chart and target report, enabling relevant business personnel to quickly obtain key information, identify problems and potential opportunities in the business, and improve the work efficiency and scientific decision-making of business personnel.

[0127] In some alternative implementations, the user information obtained is subject to user consent and complies with relevant laws and policies.

[0128] 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.

[0129] 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.

[0130] It should be emphasized that, to further ensure the privacy and security of the above behavioral prediction results, these results can also be stored in a node of a blockchain.

[0131] 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.

[0132] 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.

[0133] 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.

[0134] 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 with computer-readable instructions. These computer-readable instructions can be stored in a computer-readable storage medium. When executed, the program can include the processes of the embodiments of the above methods. 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).

[0135] 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.

[0136] Further reference Figure 3 As a response to the above Figure 2 To implement the method shown, this application provides an embodiment of an artificial intelligence-based data prediction device, which is similar to... Figure 2 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0137] like Figure 3 As shown, the data prediction device 300 based on artificial intelligence described in this embodiment includes: a first acquisition module 301, a preprocessing module 302, a first filtering module 303, an extraction module 304, a first calling module 305, a prediction module 306, and an output module 307.

[0138] in:

[0139] The first acquisition module 301 is used to acquire user data of the target user that has been collected in advance.

[0140] Preprocessing module 302 is used to preprocess the user data to obtain the corresponding target processing data;

[0141] The first filtering module 303 is used to filter the target processing data based on a preset business objective to obtain filtered behavioral data.

[0142] Extraction module 304 is used to extract features from the behavioral data based on a preset feature selection strategy to obtain corresponding feature data;

[0143] The first calling module 305 is used to call the target behavior prediction model corresponding to the business objective;

[0144] Prediction module 306 is used to perform prediction processing on the feature data based on the target behavior prediction model to obtain the corresponding behavior prediction result;

[0145] The output module 307 is used to output the behavior prediction results.

[0146] In this embodiment, the operations performed by the above modules or units correspond one-to-one with the steps of the artificial intelligence-based data prediction method in the aforementioned implementation method, and will not be repeated here.

[0147] In some optional implementations of this embodiment, the artificial intelligence-based data prediction device further includes:

[0148] The second acquisition module is used to acquire pre-collected historical behavior data corresponding to the business objective;

[0149] The processing module is used to perform feature selection and extraction processing on the historical behavior data to obtain corresponding feature sample data;

[0150] The training module is used to train multiple preset initial models of different types based on the feature sample data to obtain multiple corresponding trained specified prediction models.

[0151] The second filtering module is used to filter out the target prediction model from all the specified prediction models based on a preset model filtering strategy.

[0152] The determination module is used to use the target prediction model as the target behavior prediction model corresponding to the business objective.

[0153] In some optional implementations of this embodiment, the processing module includes:

[0154] The first processing submodule is used to perform feature selection processing on the historical behavior data based on a preset statistical analysis strategy to obtain the corresponding first feature data.

[0155] The second processing submodule is used to perform feature dimensionality reduction processing on the first feature data to obtain the corresponding second feature data.

[0156] The third processing submodule is used to perform recursive feature elimination processing on the second feature data to obtain the corresponding third feature data.

[0157] The first determining submodule is used to use the third feature data as the feature sample data.

[0158] In some optional implementations of this embodiment, the second filtering module includes:

[0159] The acquisition submodule is used to acquire the evaluation index data of each of the specified prediction models;

[0160] The calculation submodule is used to calculate and process the evaluation index data of each specified prediction model based on a preset performance calculation strategy to obtain the performance score of each specified prediction model.

[0161] The filtering submodule is used to compare the values ​​of all the performance scores to filter out the target performance score with the highest value.

[0162] The second determining submodule is used to obtain the target model corresponding to the target performance score from all the specified prediction models, and use the target model as the target prediction model.

[0163] In some optional implementations of this embodiment, the first filtering module 303 includes:

[0164] The fourth processing submodule is used to perform data cleaning processing on the user data to obtain the corresponding first processed data;

[0165] The fifth processing submodule is used to fill in missing values ​​in the first processed data to obtain the corresponding second processed data;

[0166] The sixth processing submodule is used to perform outlier processing on the second processed data to obtain the corresponding third processed data;

[0167] The third determining submodule is used to use the third processed data as the target processed data.

[0168] In some optional implementations of this embodiment, the artificial intelligence-based data prediction device further includes:

[0169] The third acquisition module is used to acquire the preset clustering strategy;

[0170] The analysis module is used to analyze the behavior prediction results based on the segmentation strategy and generate target customer groups corresponding to the target users.

[0171] The fourth acquisition module is used to acquire the target product push strategy corresponding to the target customer group;

[0172] The push module is used to perform corresponding product push processing on the target user based on the product push strategy.

[0173] In some optional implementations of this embodiment, the artificial intelligence-based data prediction device further includes:

[0174] The first generation module is used to perform chart generation processing on the behavior prediction results to obtain the corresponding target chart.

[0175] The second generation module is used to perform report generation processing on the behavior prediction results to obtain the corresponding target report;

[0176] The second calling module is used to call the preset display page;

[0177] The display module is used to visualize the target chart and the target report based on the display page.

[0178] 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.

[0179] 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.

[0180] 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.

[0181] 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 prediction methods based on artificial intelligence. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or will be output.

[0182] 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, for example, to execute computer-readable instructions for the artificial intelligence-based data prediction method.

[0183] 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.

[0184] 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 artificial intelligence-based data prediction method described above.

[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 prediction method based on artificial intelligence, characterized in that, Includes the following steps: Acquire user data from pre-collected target users; The user data is preprocessed to obtain the corresponding target processing data; Based on preset business objectives, the target processing data is filtered to obtain filtered behavioral data. Based on a preset feature selection strategy, feature extraction is performed on the behavioral data to obtain the corresponding feature data. Invoke the target behavior prediction model corresponding to the stated business objective; Based on the target behavior prediction model, the feature data is processed to obtain the corresponding behavior prediction result; The predicted behavior results are then processed for output.

2. The data prediction method based on artificial intelligence according to claim 1, characterized in that, Before the step of invoking the target behavior prediction model corresponding to the business objective, the method further includes: Acquire pre-collected historical behavioral data corresponding to the stated business objectives; The historical behavior data is processed by feature selection and extraction to obtain corresponding feature sample data; Based on the feature sample data, multiple preset initial models of different types are trained to obtain multiple corresponding trained specified prediction models. Based on a preset model selection strategy, a target prediction model is selected from all the specified prediction models. The target prediction model is used as the target behavior prediction model corresponding to the business objective.

3. The data prediction method based on artificial intelligence according to claim 2, characterized in that, The step of performing feature selection and extraction processing on the historical behavior data to obtain corresponding feature sample data specifically includes: Based on a preset statistical analysis strategy, feature selection processing is performed on the historical behavior data to obtain the corresponding first feature data; The first feature data is subjected to feature dimensionality reduction processing to obtain the corresponding second feature data; The second feature data is subjected to recursive feature elimination processing to obtain the corresponding third feature data; The third feature data is used as the feature sample data.

4. The data prediction method based on artificial intelligence according to claim 2, characterized in that, The step of selecting the target prediction model from all the specified prediction models based on a preset model selection strategy specifically includes: Obtain the evaluation index data for each of the specified prediction models; The evaluation index data of each specified prediction model are calculated and processed based on a preset performance calculation strategy to obtain the performance score of each specified prediction model. All the performance scores are numerically compared to select the target performance score with the highest value. Obtain the target model corresponding to the target performance score from all the specified prediction models, and use the target model as the target prediction model.

5. The data prediction method based on artificial intelligence according to claim 1, characterized in that, The step of preprocessing the user data to obtain the corresponding target processing data specifically includes: The user data is cleaned to obtain the corresponding first processed data; The first processed data is filled with missing values ​​to obtain the corresponding second processed data; The second processed data is subjected to outlier handling to obtain the corresponding third processed data; The third processed data is used as the target processed data.

6. The data prediction method based on artificial intelligence according to claim 1, characterized in that, Following the step of outputting the behavior prediction result, the method further includes: Obtain the preset grouping strategy; Based on the grouping strategy, the behavior prediction results are analyzed to generate target customer groups corresponding to the target users; Obtain the target product push strategy corresponding to the target customer group; Based on the product push strategy, the corresponding product push process is performed on the target user.

7. The data prediction method based on artificial intelligence according to claim 1, characterized in that, Following the step of outputting the behavior prediction result, the method further includes: The behavior prediction results are processed to generate a chart, resulting in the corresponding target chart. The behavior prediction results are processed to generate a report, resulting in a corresponding target report; Call the preset display page; The target chart and target report are visualized based on the display page.

8. A data prediction device based on artificial intelligence, characterized in that, include: The first acquisition module is used to acquire user data of the target users that have been collected in advance. The preprocessing module is used to preprocess the user data to obtain the corresponding target processing data; The first filtering module is used to filter the target processing data based on preset business objectives to obtain filtered behavioral data. The extraction module is used to extract features from the behavioral data based on a preset feature selection strategy to obtain corresponding feature data; The first calling module is used to call the target behavior prediction model corresponding to the business objective; The prediction module is used to perform prediction processing on the feature data based on the target behavior prediction model to obtain the corresponding behavior prediction result; The output module is used to process the output of the behavior prediction results.

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 prediction method based on artificial intelligence 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, which, when executed by a processor, implement the steps of the data prediction method based on artificial intelligence as described in any one of claims 1 to 7.

Images