Intelligent decision-making system and method based on big data analysis

Abstract

The invention discloses an intelligent decision-making system and method based on big data analysis, and the system comprises a data collection module which is used for collecting user behavior data; the data processing module is used for cleaning, denoising and preprocessing the collected user behavior data to form a user behavior data set; the behavior analysis module is used for analyzing user behaviors based on the user behavior data set, extracting feature data and constructing a user behavior model; the decision prediction module is used for constructing a decision prediction model based on the user behavior model and historical decision data; the real-time analysis module is used for acquiring data in real time and performing prediction analysis by adopting a decision prediction model to obtain a decision result; the decision output module is used for generating a decision report according to the decision result and visually presenting the decision report; according to the method, intelligent decision making is carried out by adopting the system. The method can effectively improve the scientificity, accuracy and efficiency of decision making, has a self-optimization capability, and can provide powerful decision making support.

Description

TECHNICAL FIELD

[0001] The present application relates to a decision-making system and method, in particular to an intelligent decision-making system and method based on big data analysis. BACKGROUND

[0002] With the rapid development of information technology, especially the wide application of big data technology and artificial intelligence technology, data analysis has become a key tool for decision support. However, the existing traditional decision-making method relies on the experience and knowledge of decision-makers, but with the rapid growth of data volume, the limitations of manual analysis are increasingly apparent, especially in dealing with massive and complex data, the defects of traditional methods are increasingly prominent, traditional decision-making methods usually rely on experience and intuition, which cannot effectively handle large data sets, resulting in low decision-making efficiency and insufficient accuracy, and it is difficult to respond to rapidly changing environments in real time, in addition, manual analysis is not only susceptible to personal subjective judgment, but also in the face of multi-source and multi-dimensional data, traditional methods are difficult to achieve comprehensive and accurate analysis, especially in the fields of finance, medicine, transportation, etc., a single data analysis method cannot meet the needs of rapid decision-making and dynamic adjustment, traditional methods have limited ability to discover potential associations and complex relationships between data, often ignoring the valuable information hidden behind a large amount of data, thereby affecting the quality and effectiveness of decision-making.

[0003] Therefore, the existing decision support system faces great challenges, and cannot timely and accurately process the growing and diversified data, and lacks flexible real-time response capability. Although big data technology and artificial intelligence technology provide potential solutions to improve decision-making efficiency, how to fully utilize the advantages of these technologies and break through the limitations of traditional decision-making methods remains a problem to be solved. SUMMARY

[0004] The technical problem to be solved by the present application is to provide an intelligent decision-making system and method based on big data analysis to overcome the shortcomings of the prior art.

[0005] To solve the above technical problems, the present application discloses an intelligent decision-making system and method based on big data analysis, wherein the system comprises: a data acquisition module, a data processing module, a behavior analysis module, a decision prediction module, a real-time analysis module and a decision output module; wherein, The data acquisition module is used for collecting user behavior data. The data processing module cleans, denoises and pre-processes the collected user behavior data to form a user behavior data set. The behavior analysis module analyzes user behavior based on the user behavior data set, extracts feature data and constructs a user behavior model. Decision prediction module, which builds a decision prediction model based on user behavior model and historical decision data; Real-time analysis module, which acquires data in real time and uses decision prediction models to perform prediction analysis and obtain decision results; The decision output module generates a decision report based on the decision results and visually presents the decision data.

[0006] Furthermore, the data processing module includes: Data cleaning unit, used to remove noise and process outliers from the collected data; Data normalization unit normalizes the cleaned data and standardizes the data format; The data integration unit integrates and stores the normalized data to obtain the user behavior data set.

[0007] Furthermore, the behavior analysis module performs classification analysis on the user behavior data in the user behavior data set and extracts key features.

[0008] Furthermore, the behavior analysis module includes: Behavior clustering unit and feature extraction unit, where The behavior clustering unit clusters user behaviors based on key behavior features. The clustering method is as follows: ; in, is the objective function, For the clusters, For the The mean of the clusters, User behavior data; The feature extraction unit extracts the key features of user behavior based on the clustering results to form a key feature set, namely the user behavior model.

[0009] Furthermore, the classification analysis of the user behavior data in the user behavior dataset and the extraction of key features specifically include the following steps: Data preprocessing, defining a behavioral classification system based on business goals; Extract basic features related to classification; Use supervised learning methods to build and train classification models; Apply the trained classification model to classify and label user behaviors; Filter the feature subsets used for clustering in the classification results; Determine the appropriate number of clusters based on business needs or using the elbow method; K-means and DBSCAN methods were used to perform cluster analysis on similar behaviors; The clustering effect is evaluated by the profile coefficient index, and the clustering parameters are optimized; Calculate the feature statistics within each behavior category and clustering cluster; Identify key features using feature importance scores; Analyze the feature differences between different clusters and extract features with discriminative power, i.e. key features.

[0010] Further, the decision prediction module comprises: A decision modeling unit constructs a decision prediction model based on a user behavior model, and the decision prediction model is represented as follows: ; Wherein, is the target decision output, is the th key feature data of user behavior, is the corresponding th parameter, is the error term; A model optimization unit trains the decision prediction model based on collected historical decision data, and uses the method of minimizing the loss function to optimize the decision prediction model; An output unit is used to output the decision result of the decision prediction model.

[0011] Further, the loss function is represented as follows: ; Wherein, is the loss function, is the index of the data point, is the true value of the th data point, is the intercept term, is the regression coefficient of the th feature, is the th feature value of the th sample, is the total number of samples, is the total number of features.

[0012] Further, the optimization of the decision prediction model comprises the following in: Adjust the initial model parameters and adjustment parameters in the decision prediction model; Calculate the bias gradient to analyze the influence of each feature on the prediction bias under the current parameters, and determine the direction and amplitude of parameter adjustment; Combine the historical parameter adjustment situation to calculate the momentum and amplitude change of the current parameter adjustment; The calculation result of the momentum and amplitude variation of the current parameter adjustment is corrected; According to the corrected momentum and amplitude variation of the current parameter adjustment, the weight of each feature and the base intercept are adjusted; After each round of adjustment, the overall deviation of the decision prediction model on the training data and the comprehensive performance index on the validation data are calculated to judge the optimization effect; When the preset iteration round is completed, or the deviation of the decision prediction model reaches the threshold, the optimization process is ended.

[0013] Further, the real-time analysis module comprises: A data receiving unit for receiving user behavior data in real time; A real-time prediction unit for performing prediction analysis according to the real-time user behavior data and the decision prediction model; An adjustment suggestion unit for generating decision adjustment suggestions according to the prediction analysis result.

[0014] The present application also proposes an intelligent decision-making method based on big data analysis, which uses the aforementioned system to make intelligent decisions based on big data analysis, comprising the following steps: Step 1, collecting user behavior data; Step 2, cleaning, denoising and preprocessing the collected user behavior data to form a user behavior data set; Step 3, analyzing the user behavior mode according to the user behavior data in the user behavior data set, extracting feature data, and constructing a user behavior model; Step 4, constructing a decision prediction model based on the user behavior model and historical decision data; Step 5, collecting and analyzing current user behavior data in real time according to the decision prediction model, predicting and optimizing future decisions; Step 6, outputting the optimized decision result to complete the intelligent decision-making based on big data analysis.

[0015] Advantages: The intelligent decision-making system and method based on big data analysis can effectively improve the scientificity, accuracy and efficiency of decision-making, and has self-optimization ability, which adapts to the rapidly changing business needs, provides strong decision support for enterprises and organizations, and significantly improves the competitiveness and decision-making level of enterprises, especially in large-scale data environment. BRIEF DESCRIPTION OF DRAWINGS

[0016] The above and / or other aspects of the present application will become apparent from the following detailed description and the specific embodiments taken in conjunction with the accompanying drawings.

[0017] Figure 1 The system framework of the present application is shown in the figure.

[0018] Figure 2 A schematic diagram of the step flow of the present application. DETAILED DESCRIPTION

[0019] As Figure 1 shown, the intelligent decision-making system based on big data analysis proposed by the present application comprises: a data acquisition module for acquiring user behavior data; a data processing module for cleaning, denoising and preprocessing the collected data to form a user behavior data set; a behavior analysis module for analyzing user behavior, extracting feature data and constructing a user behavior model; a decision prediction module for generating a decision prediction model based on the user behavior model and historical decision data; a real-time analysis module for real-time data acquisition and prediction analysis to output optimized decision results; a decision output module for generating a decision report and visualizing the decision data for decision support.

[0020] The data processing module internally comprises: a data cleaning unit for removing noise and handling outliers from the collected data; a data normalization unit for normalizing the data and standardizing the data format; a data integration unit for integrating and storing the cleaned data; This module uses advanced denoising and outlier processing algorithms to handle large amounts of user behavior data, ensuring data accuracy and quality. This innovation through data cleaning, normalization and integration makes the data more reliable and accurate in subsequent analysis.

[0021] The behavior analysis module internally comprises: a user behavior classification unit for classifying and analyzing user behavior data to extract key behavior features; a behavior clustering unit for clustering similar user behaviors to construct an effective user behavior model Wherein, the clustering formula is:

[0022] In the formula, is the objective function, is the th cluster, is the mean of the th cluster, is the user behavior data; a feature extraction unit configured to extract key features of user behavior and input into a decision prediction module; The module can analyze user behavior patterns in detail and extract key features by combining behavior classification and clustering analysis methods. Through clustering analysis of user behavior, different types of user groups are identified, and corresponding group behavior features are extracted to further refine user behavior features. This innovation optimizes user behavior modeling through clustering algorithms to generate more accurate user behavior models and improve prediction accuracy. The feature extraction step is specifically: Data preprocessing, based on business target definition behavior classification system; Extracting basic features related to classification (such as behavior frequency, duration, interval, etc.); Using supervised learning algorithms to build classification models; Applying the trained model to classify and label user behavior; Filtering the feature subset suitable for clustering in the classification results; Based on business needs or elbow method to determine a reasonable number of clusters; Using K-means and DBSCAN algorithms for clustering analysis of similar behaviors; Evaluate the clustering effect through indicators such as contour coefficients and optimize clustering parameters; Calculate the feature statistics (mean, variance) within each behavior category and clustering cluster; Using feature importance scores (such as information gain, feature contribution) to identify key features; Analyze the feature differences between different categories / clusters and extract features with discriminative power.

[0023] The decision prediction module includes: A decision modeling unit for training and building a decision prediction model based on historical data and user behavior models; The decision prediction model formula is: In this system, whether to purchase recommended goods on an e-commerce platform is taken as the output variable, the number of clicks in the last 24 hours (x1), the duration of stay (x2), and the number of historical purchases (x3) as input features. Linear regression coefficients are solved by least squares method to form the prediction model: y = 0.43x1 + 0.27x2 + 0.13x3 + 0.05

[0024] In the formula, is the target decision output, is the feature data is the model parameter, is an error term; a model optimization unit for tuning the prediction model to ensure its accuracy and timeliness, wherein the tuning formula is: a model optimization unit for tuning the prediction model to ensure its accuracy and timeliness, wherein the tuning formula is: The prediction model has an RMSE of 0.35 on the initial training set (30,000 samples), which decreases to 0.22 after 200 iterations of the Adam optimizer, and the F1 value on the validation set increases from 0.68 to 0.82.

[0025]

[0026] is a loss function, is an index of a data point, is the true value of the th data point, is an intercept term, is the regression coefficient of the th feature, is the th feature value of the th sample, is the total number of samples, is the total number of features; Adjust the model parameters by repeating iterations until the loss function reaches a minimum value; an output unit for generating optimized decision results and providing them to decision makers; The decision prediction module combines machine learning and regression analysis techniques to accurately predict decisions based on historical data. The innovation of this module lies in using degree learning algorithms to comprehensively analyze multi-dimensional data, improving the accuracy and dynamic adaptability of the prediction model. First, determine the initial model parameters, which are based on the correlation between user clicks, dwell time, and historical purchase behavior in historical data. At the same time, set the adjustment parameters needed in the optimization process, including learning pace, momentum impact, and stability terms to avoid calculation problems, as well as related variables to record the iteration process; Calculate the bias gradient to analyze the influence of each feature on the prediction bias under the current parameters, and determine the direction and approximate amplitude of the parameter adjustment; Combine the historical parameter adjustment to calculate the momentum and amplitude changes of the current parameter adjustment, making the adjustment more stable and avoiding excessive fluctuations; Correct the above calculation results to eliminate the bias caused by insufficient data in the early iteration, making the adjustment direction more accurate; According to the results obtained, the weights of each feature and the base intercept are adjusted by an appropriate amplitude to make the model prediction closer to the actual situation; After each round of adjustment, the overall deviation of the model on the training data and the comprehensive performance index on the validation data are calculated to judge the optimization effect; When the preset number of iterations is completed or the model deviation is small enough, the optimization process is ended; During the optimization process, it is found that the overall deviation of the model gradually decreases, the prediction accuracy of the purchase behavior continuously improves, and finally reaches an ideal state; In order to make the model adapt to the changes of user behavior, it is necessary to periodically repeat the above optimization process with newly generated data, and when the model prediction accuracy is found to have decreased to a certain extent, timely emergency optimization is carried out, and at the same time, the parameters after each optimization are retained, so that when needed, it can be traced back to the previous version; Through such iterative optimization, the model can continuously adapt to new user behavior patterns, maintain prediction accuracy, and always have good timeliness. The optimization method adopted can find a better parameter combination faster than traditional methods by adaptively adjusting the amplitude of parameter update, which is especially suitable for processing large-scale user behavior data.

[0027] The real-time analysis module includes: A data receiving unit for receiving data streams in real time and performing preliminary processing; A real-time prediction unit for quickly predicting and analyzing based on real-time data and decision prediction model; An adjustment suggestion unit for generating decision adjustment suggestions and providing optimization suggestions to decision makers; The real-time analysis module can obtain the latest user behavior data in real time through fast data stream processing technology, combine with historical prediction model, and adjust the decision in real time. This innovation supports real-time decision and rapid response, and provides more timely decision basis for decision makers.

[0028] Referring to Figure 2 The intelligent decision-making method based on big data analysis is used to realize the intelligent decision-making system based on big data analysis, which includes: Collecting user behavior data; According to the user behavior data, the behavior patterns of users are analyzed, the feature data is extracted, and the user behavior model is constructed; Based on the user behavior model and historical decision data, a decision prediction model is generated; According to the decision prediction model, real-time analysis of current data is carried out to predict and optimize future decisions; Output the optimized decision result for decision makers to make decision support; The innovation of this module lies in its powerful data visualization function, which can present complex data analysis results in clear charts and reports to decision-makers, improving decision-making efficiency. Through interface interaction, decision-makers can further analyze and adjust optimization suggestions.

[0029] Based on user behavior data, analyze user behavior patterns, extract feature data, including: Collect user behavior data in different scenarios, obtain user operation records, clickstream data and purchase history; Preprocess behavior data, including denoising and normalization, to form a clear behavior dataset; Based on the dataset, classify and cluster user behavior through machine learning algorithms to extract effective behavior features; Input the extracted feature data into the decision prediction model to optimize the decision prediction result; This design combines data cleaning, normalization and integration processing, improving data quality and storage efficiency. Data normalization and standardization can effectively improve the effectiveness of subsequent analysis models and reduce prediction errors caused by data bias.

[0030] Generating a decision prediction model specifically includes: Based on historical decision data, use regression analysis and deep learning algorithms for model training; Combine user behavior model and historical decision data to generate a multi-dimensional decision prediction model; Optimize and verify the model to ensure prediction accuracy and continuously update the model according to actual business scenarios; The innovation of this module lies in the combination of regression analysis and deep learning algorithms to generate a multi-dimensional decision prediction model. The model optimization unit optimizes itself based on the loss function, continuously improving prediction accuracy. This multi-dimensional modeling and dynamic optimization greatly improves the accuracy and timeliness of decisions.

[0031] Real-time analysis of current data, prediction and optimization of future decisions specifically includes: Obtain real-time data stream, perform data cleaning and preprocessing; Input real-time data into the decision prediction model and conduct decision analysis based on current business needs; Based on the analysis results, optimize decisions and generate real-time adjustment suggestions to support rapid response and adjustment of decision strategies; This method innovatively combines user behavior data with machine learning algorithms, improving the accuracy of behavior pattern recognition through data classification, clustering and feature extraction. This enables the system to respond to different user needs in a timely manner, providing accurate decision support for decision-makers.

[0032] Output the optimized decision result for decision maker to make decision support, specifically including: Based on the optimization result, generate a decision report and visually present the decision data; Support decision makers to interact with optimization suggestions through intelligent system interface, and provide strong operability decision tools; The system displays'real-time hot-selling list' 'predicted transaction user pool' and other modules in the form of decision console, and allows operation personnel to adjust the recommendation strategy. The system refreshes feedback data every 5 minutes and iteratively optimizes the result.

[0033] According to user feedback, further analyze and adjust the optimization suggestion to improve the prediction accuracy and decision efficiency of the system; The method processes multi-dimensional data through deep learning technology, models combined with historical decision data, can generate more accurate and self-adaptive decision prediction model. This innovation can optimize decision accuracy and enhance the flexibility of the model to adapt to changing business needs.

[0034] The use process of the application is: Step one: user behavior data is collected from multiple sources; Step two: remove noise from the collected data, handle outliers, and ensure data quality; Step three: unify the data format, standardize and normalize the data, and ensure data consistency; Step four: integrate and store the cleaned data to prepare the data set for subsequent analysis; Step five: use classification algorithm to classify user behavior data and extract key behavior features; Step six: cluster similar user behavior data to build an effective user behavior model; Step seven: extract key features from clustered data as input for decision prediction model; Step eight: build a user behavior model based on the extracted features to reveal user behavior patterns; Step nine: combine historical decision data and user behavior model to generate decision prediction model through regression analysis and deep learning algorithm; Step ten: use historical data to train the decision prediction model to ensure the model can accurately predict future decisions; Step eleven: tune the prediction model, adjust model parameters, and improve prediction accuracy and timeliness; Step twelve: continuously receive new user behavior data streams through real-time data receiving unit; Step thirteen: perform preliminary processing on the real-time received data, including cleaning, denoising and other operations; Step fourteen: combine real-time data with decision prediction model for rapid prediction analysis; Step fifteen: based on real-time prediction results, generate adjustment suggestions for optimized decisions; Step sixteen: visualize the optimized decision results through the decision output module, and generate a decision report; Step seventeen: through the intelligent system interface, decision makers can interact with the optimization suggestions and adjust the decisions; Step eighteen: according to the feedback of decision makers, further adjust the prediction model and decision optimization suggestions, improve the prediction accuracy and decision efficiency of the system.

[0035] In summary, the advantages of the present application are: The system can establish an accurate user behavior model through comprehensive analysis of user behavior data, thereby predicting the future behavior trend of users. This data-based prediction can avoid the common biases and errors in manual decision-making, making the decision more accurate and rational; The system can collect and analyze data in real time, and combine the latest decision prediction model for rapid response. Whether it is the change of market trend, the fluctuation of user behavior, or the adjustment of other business needs, the system can provide timely optimization suggestions to provide timely decision support for decision makers; Through automatic data analysis and prediction, the system greatly reduces the need for manual intervention and improves decision efficiency. Decision makers can directly see the decision report and optimization suggestions through the visual interface, thereby quickly responding to avoid missing market opportunities due to delayed decisions; The system relies entirely on data analysis, avoiding the limitations of subjective experience or human judgment. In the era of big data, the system can fully exploit and utilize the information in large-scale data to provide more scientific and reasonable decision-making basis; Through machine learning algorithms, the system can optimize itself according to new data, continuously improving the prediction accuracy of the decision model. This allows the system to automatically adapt to changing environments and optimize the decision-making process, avoiding outdated prediction results affecting decisions; The system can present complex analysis results through a graphical interface, allowing decision makers to clearly understand the decision basis and analysis process. This not only increases the transparency of decision-making, but also improves the decision makers' trust in the decision-making process; Through in-depth analysis of user behavior, the system can predict changes in user demand, helping enterprises optimize resource allocation. For example, in e-commerce platforms, the system can predict which goods have more market demand based on user purchase behavior and browsing habits, thereby achieving precise inventory management and personalized marketing; The system can be customized according to different business scenarios, whether it is financial risk control, medical diagnosis, advertising, or supply chain management, it can provide suitable intelligent decision support, which makes the system have a wide range of application scenarios and strong adaptability.

[0036] Embodiments: Suppose an e-commerce platform wants to optimize its product recommendation system by analyzing user behavior data to improve user conversion rates, the platform wants to build an intelligent decision-making system that can accurately predict user purchase intentions based on historical behavior and real-time data, and provide personalized recommendations to adjust decision-making strategies and optimize sales strategies in real time.

[0037] This embodiment is based on a real data set: user behavior logs collected by an e-commerce platform from April to June 2025, a total of 412,586 data, involving four types of behavior: clicks, views, adds, and purchases, data fields include user ID, behavior time, page type, behavior type, etc.

[0038] Data collection module: collect user operation records, clickstream data, browsing history, search terms, purchase history, and user dwell time behavior data on the e-commerce platform, through the log collection script set on each page of the platform, real-time record and upload each item of user operation data; Data processing module: clean, denoise, normalize and integrate the collected data to form a standardized user behavior dataset, use denoising algorithms to remove noise caused by system failures or abnormal inputs, convert different formats of data into a unified standard format to ensure that subsequent analysis is not affected by different data scales, combine data from different sources into a database to form a dataset for subsequent analysis; Behavior analysis module: use classification algorithms to classify user behavior into different types, cluster users based on behavior patterns using K-means algorithm, extract user purchase tendencies, browsing interests and other features from the clustering results, and pass these features to the decision prediction module for further analysis; Decision prediction module: use regression analysis and deep learning to train decision prediction models, analyze user behavior data and historical purchase records to predict the likelihood of a user purchasing a certain type of product within the next 24 hours, use gradient descent to optimize the prediction model to minimize prediction error, ensure the accuracy of the decision prediction model, and generate optimized recommendations based on the current model, such as recommending specific products or promotions to specific users; Among them, the model is built by TensorFlow platform, and a three-layer network is built by DNNRegression. The input layer contains 6 features (clicks, add-to-cart times, average dwell time, gender, age group, and promotion click preference). The training accuracy rate reaches 89.2%.

[0039] Real-time analysis module: Real-time receive user's click stream data and behavior data, and carry out preliminary cleaning and preprocessing on the data. Combine real-time data with historical decision-making model to quickly predict user's purchase intention. According to the real-time analysis result, provide real-time adjustment suggestions for decision makers, such as adjusting recommended goods and adjusting advertising strategy. For example, during the 618 activity, the system identifies that the click rate of users on the full-reduction type goods has increased by 37%. The model immediately increases the weight of related goods and suggests that the 'limited discount' label should be displayed preferentially to improve the click conversion rate.

[0040] Decision output module: Based on the optimized prediction results, generate decision report, which includes user's purchase tendency, recommended goods, potential conversion rate and other key data. Through data visualization technology, present the decision data to the decision maker, so that the decision maker can quickly understand and make adjustments. The decision maker can interact with the optimization suggestions through the system interface, such as adjusting the recommended goods algorithm and modifying the promotion activity strategy. Specific analysis steps: In this embodiment, the behavior data of five users is selected to analyze whether they will purchase the recommended goods. The main focus is on the number of clicks, dwell time and historical purchase times of each user. At the same time, it is also known whether they actually purchased the goods for subsequent verification. Firstly, cluster analysis is carried out, the purpose is to classify users with similar behavior into one class. According to business requirements, this implementation plans to divide users into two categories, one is high purchase intention, and the other is low purchase intention. At the beginning, this implementation roughly classifies three users as high intention group and the other two as low intention group. Then, the average value of each group in the three features is calculated. Next, the difference between the feature value of each user and the average value of the group is calculated to measure the rationality of the grouping. The smaller the difference, the more similar the behavior of the users in the same group. After calculation and optimization, the final grouping is determined: three users belong to high intention group and two belong to low intention group. From this clustering result, the key features are extracted. By comparing the average values of the two groups, it is found that the high intention group has significantly higher click times, dwell time and historical purchase times than the low intention group. These three features can effectively distinguish the purchase intention of users, among which the difference in click times and dwell time is more prominent, becoming the core features. After that, the purchase probability of each user is calculated by using the prediction model established before, which is established by analyzing the historical data and determining the influence degree of the number of clicks, the length of stay, and the number of historical purchases on the purchase behavior. The three feature data of each user are substituted into the model to obtain the corresponding purchase probability prediction value, and then a judgment standard is set. When the prediction value reaches a certain level, it is considered that the user will purchase. According to this standard, the five users are predicted, and it is found that the prediction result is completely consistent with their actual purchase situation, which also shows that the model is effective. In general, by clustering analysis, users are divided into different intention groups, key features are extracted, and purchase probability is calculated and judged by using the prediction model, which can accurately predict whether the user will purchase the recommended goods.

[0041] In a specific implementation, the present application provides a computer storage medium and a corresponding data processing unit, wherein the computer storage medium can store a computer program, and the computer program can run the invention content and part or all steps in each embodiment of the intelligent decision-making system and method based on big data analysis provided by the present application when executed by the data processing unit. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.

[0042] Those skilled in the art can clearly understand that the technical solutions in the embodiments of the present application can be realized by means of a computer program and its corresponding general hardware platform. Based on such understanding, the technical solutions in the embodiments of the present application can be embodied in the form of a computer program, i.e. a software product, which can be stored in a storage medium, including a plurality of instructions for causing a device (which can be a personal computer, a server, a single-chip microcomputer, an MCU, or a network device, etc.) containing a data processing unit to execute the method described in each embodiment or some parts of the embodiments of the present application.

[0043] The present application provides a thought and method of an intelligent decision-making system and method based on big data analysis. There are many methods and ways to realize the technical solutions, and the above description is only the preferred embodiment of the present application. It should be pointed out that for ordinary skilled persons in the technical field, without departing from the principles of the present application, some improvements and refinements can be made, which should also be regarded as the protection scope of the present application. The components not explicitly described in the embodiments can be realized by using existing technology.

Claims

1. An intelligent decision-making system based on big data analysis, characterized in that: include: Data acquisition module, data processing module, behavior analysis module, decision prediction module, real-time analysis module and decision output module; among them, Data collection module, used to collect user behavior data; The data processing module cleans, denoises, and preprocesses the collected user behavior data to form a user behavior data set; Behavior analysis module, which analyzes user behavior based on user behavior datasets, extracts feature data and builds user behavior models; Decision prediction module, which builds a decision prediction model based on user behavior model and historical decision data; Real-time analysis module, which acquires data in real time and uses decision prediction models to perform prediction analysis and obtain decision results; The decision output module generates a decision report based on the decision results and visually presents the decision data.

2. The intelligent decision-making system based on big data analysis according to claim 1, characterized in that: The data processing module includes: Data cleaning unit, used to remove noise and process outliers from the collected data; Data normalization unit normalizes the cleaned data and standardizes the data format; The data integration unit integrates and stores the normalized data to obtain the user behavior data set.

3. The intelligent decision-making system based on big data analysis according to claim 2, characterized in that: The behavior analysis module performs classification analysis on the user behavior data in the user behavior data set and extracts key features.

4. The intelligent decision-making system based on big data analysis according to claim 3, characterized in that: The behavior analysis module includes: Behavior clustering unit and feature extraction unit, where The behavior clustering unit clusters user behaviors based on key behavior features. The clustering method is as follows: ； in, is the objective function, For the clusters, For the The mean of the clusters, User behavior data; The feature extraction unit extracts the key features of user behavior based on the clustering results to form a key feature set, namely the user behavior model.

5. The intelligent decision-making system based on big data analysis according to claim 4 is characterized in that: The classification analysis of the user behavior data in the user behavior dataset and the extraction of key features specifically include the following steps: Data preprocessing, defining a behavioral classification system based on business goals; Extract basic features related to classification; Use supervised learning methods to build and train classification models; Apply the trained classification model to classify and label user behaviors; Filter the feature subsets used for clustering in the classification results; Determine the appropriate number of clusters based on business needs or using the elbow method; K-means and DBSCAN methods were used to perform cluster analysis on similar behaviors; The clustering effect is evaluated by the silhouette coefficient indicator and the clustering parameters are optimized; Calculate feature statistics within each behavior category and cluster; Identify key features using feature importance scores; Analyze the feature differences between different clusters and extract the distinguishing features, which are the key features.

6. The intelligent decision-making system based on big data analysis according to claim 5, characterized in that: The decision prediction module includes: The decision modeling unit builds a decision prediction model based on the user behavior model. The decision prediction model is expressed as follows: ； in, is the target decision output, For user behavior Key feature data, For the corresponding parameters, is the error term; The model optimization unit trains the decision prediction model based on the collected historical decision data and optimizes the decision prediction model by minimizing the loss function; The output unit is used to output the decision results of the decision prediction model.

7. The intelligent decision-making system based on big data analysis according to claim 6, characterized in that: The loss function is expressed as follows: ； in, is the loss function, is the index of the data point, For the The true value of the data point, is the intercept term, For the The regression coefficient of the feature, For the The first sample eigenvalues, is the total number of samples, Total number of features.

8. The intelligent decision-making system based on big data analysis according to claim 7, characterized in that: The optimization of the decision prediction model includes the following: The initial model parameters and adjustment parameters in the decision prediction model are missing; Calculate the deviation gradient, analyze the impact of each feature on the prediction deviation under the current parameters, and determine the direction and magnitude of parameter adjustment; Combined with historical parameter adjustments, calculate the momentum and amplitude changes of the current parameter adjustments; Correct the calculation results of momentum and amplitude changes of the current parameter adjustment; Adjust the weights of each feature and the basic intercept according to the momentum and amplitude changes of the corrected current parameter adjustment; After each round of adjustment, the overall deviation of the decision prediction model on the training data and the comprehensive performance indicators on the validation data are calculated to determine the optimization effect; The optimization process ends when the preset iteration rounds are completed or the deviation of the decision prediction model reaches a threshold.

9. The intelligent decision-making system based on big data analysis according to claim 8, characterized in that: The real-time analysis module includes: A data receiving unit, used to receive user behavior data in real time; Real-time prediction unit, which performs prediction analysis based on real-time user behavior data and decision prediction models; The adjustment suggestion unit generates decision adjustment suggestions based on the prediction analysis results.

10. An intelligent decision-making method based on big data analysis, characterized in that: Using any system as described in claims 1 to 9 to make intelligent decisions based on big data analysis includes the following steps: Step 1: Collect user behavior data; Step 2: Clean, denoise, and preprocess the collected user behavior data to form a user behavior dataset; Step 3: Analyze the user behavior patterns based on the user behavior data in the user behavior dataset, extract feature data, and build a user behavior model; Step 4: Build a decision prediction model based on the user behavior model and historical decision data; Step 5: Based on the decision prediction model, collect and analyze current user behavior data in real time to predict and optimize future decisions; Step 6: Output the optimized decision results and complete intelligent decision-making based on big data analysis.

Images