User commodity price recommendation and VR visualization method and system based on AI analysis

Abstract

The invention relates to the field of commodity price adjustment, in particular to a user commodity price recommendation and VR visualization method and system based on AI analysis. The method comprises the following steps: obtaining a real-time price data stream of a target commodity based on an AI model, carrying out time sequence price fluctuation trend analysis, and constructing a price fluctuation time-frequency characteristic diagram; collecting a user use behavior log of each platform, performing dynamic mining of a user consumption track, performing user price acceptance evaluation based on a price fluctuation time-frequency characteristic diagram, and constructing a user price elastic model; and carrying out multi-time-point sensitivity evolution identification based on the user price elastic model, then carrying out optimal price adjustment opportunity prediction, and generating a price adjustment time sequence optimization strategy. According to the invention, through dynamic directional price pushing, the ROI (input-output ratio) of an enterprise and the user experience are improved, through VR commodity visualization, more visual commodity information is displayed for clients, and the purchase rate of the users is improved.

Description

Technical Field

[0001] The present invention relates to the field of commodity price adjustment, and particularly to a method and system for user commodity price recommendation and VR visualization based on AI analysis. Background Art

[0002] With the rapid development of e-commerce and digital retail, consumers' demands for personalized and interactive shopping experiences are constantly increasing. When making purchase decisions among a vast amount of commodity information, users often face difficulties in information overload and price comparison. Traditional commodity recommendation systems mainly rely on historical purchase records and commodity attribute matching, and provide commodity recommendations for users through algorithms such as collaborative filtering and content recommendation. However, these methods usually ignore the differences in users' price sensitivity and personalized price preferences, resulting in the recommendation effect being difficult to accurately meet the budget and value demands of different users.

[0003] Meanwhile, the rapid development of virtual reality (VR) technology has provided consumers with a brand-new immersive shopping experience. Through the VR environment, users can achieve three-dimensional spatial display and interactive operations of commodities, enhancing the perceptual experience and intuitiveness of shopping. However, most current VR shopping platforms focus more on commodity display and lack personalized services combined with intelligent price recommendation, making it difficult to fully exploit the potential of VR technology in improving users' shopping decision-making efficiency and satisfaction.

[0004] Based on this, combining artificial intelligence (AI) technology and virtual reality technology to develop a method for user commodity price recommendation and VR visualization based on AI analysis has become an important research direction for enhancing the intelligent retail experience. This method accurately mines users' price-sensitive features and consumption preferences through deep learning and big data analysis, and dynamically generates personalized price recommendation strategies that meet users' budgets; at the same time, it uses VR technology to build a realistic three-dimensional commodity display environment, intuitively integrating price information with commodity attributes, enhancing users' perception and understanding of commodity value, and promoting more rational shopping decisions. Implementing this method also faces multiple technical challenges. On the one hand, the diversity and complexity of users' price preferences require the AI model to have efficient feature extraction and real-time prediction capabilities to handle massive dynamic data; on the other hand, the visualization of price and commodity information in the VR environment requires the design of a reasonable interaction interface and rendering mechanism to ensure the usability and immersion of the recommendation results. Summary of the Invention

[0005] To solve the above technical problems, the present invention proposes a method and system for user commodity price recommendation and VR visualization based on AI analysis to solve at least one of the above technical problems.

[0006] To achieve the above object, the present invention provides a method for user commodity price recommendation and VR visualization based on AI analysis, including the following steps: Step S1: Obtain the real-time price data stream of the target commodity based on the AI model, conduct a time-series price fluctuation trend analysis, and construct a time-frequency characteristic graph of price fluctuations; Step S2: Collect the user usage behavior logs of each platform, conduct dynamic mining of user consumption trajectories, and evaluate the user price acceptance based on the time-frequency characteristic graph of price fluctuations to construct a user price elasticity model; Step S3: Identify the evolution of multi-point sensitivity based on the user price elasticity model, then predict the optimal price adjustment timing, and generate an optimized time-series price adjustment strategy; Step S4: Predict the transfer of user interests based on the user usage behavior logs, calculate the time-series churn probability, and construct a multi-point user churn probability curve; Step S5: Dynamically constrain and optimize the multi-point user churn probability curve according to the optimized time-series price adjustment strategy and make a user personalized price push decision to construct a personalized price push strategy; Step S6: Execute the commodity price recommendation task based on the personalized price push strategy; and conduct a VR visualization display task of the three-dimensional form of the target commodity.

[0007] In this specification, a user commodity price recommendation and VR visualization system based on AI analysis is provided for executing the user commodity price recommendation and VR visualization method based on AI analysis as described above, including: A price fluctuation analysis module that obtains the real-time price data stream of the target commodity based on the AI model, conducts a time-series price fluctuation trend analysis, and constructs a time-frequency characteristic graph of price fluctuations; A user usage behavior module that collects the user usage behavior logs of each platform, conducts dynamic mining of user consumption trajectories, and evaluates the user price acceptance based on the time-frequency characteristic graph of price fluctuations to construct a user price elasticity model; An optimal price adjustment module that identifies the evolution of multi-point sensitivity based on the user price elasticity model, then predicts the optimal price adjustment timing, and generates an optimized time-series price adjustment strategy; A churn probability calculation module that predicts the transfer of user interests based on the user usage behavior logs, calculates the time-series churn probability, and constructs a multi-point user churn probability curve; A constraint optimization and solution module that dynamically constrains and optimizes the multi-point user churn probability curve according to the optimized time-series price adjustment strategy and makes a user personalized price push decision to construct a personalized price push strategy; A VR visualization module that executes the commodity price recommendation task based on the personalized price push strategy; and conducts a VR visualization display task of the three-dimensional form of the target commodity.

[0008] The beneficial effects of the present invention are specifically as follows: By capturing price data in real time and conducting trend analysis, it helps to promptly grasp the market price dynamics and enhance the sensitivity of price monitoring. Constructing a time-frequency characteristic graph of price fluctuations can more comprehensively depict the short-term fluctuations and long-term trends of commodity prices, enabling high-precision price behavior modeling. It provides a solid data foundation for subsequent user price response analysis, elasticity modeling, and strategy formulation. Mining the consumption trajectories of users on various platforms helps to deeply understand the behavioral characteristics of users, such as price sensitivity, consumption cycle, and preferred categories. Combining user behavior analysis with price fluctuation characteristics can accurately evaluate the acceptance of users to price changes and construct an individualized price elasticity model. The elasticity model can be used to distinguish price-sensitive users from non-sensitive users, enabling more targeted strategy design. By analyzing the temporal evolution characteristics of user price sensitivity, the optimal time window for price adjustment can be identified, thereby enhancing the effect of price intervention. The timing and amplitude of price adjustment are more accurate, which helps to maximize the conversion rate while controlling the promotion cost. Constructing a dynamic time-series optimization strategy to achieve the forward-looking and intelligent price adjustment. Predicting the changing trend of user interests helps to detect potential lost users in advance and enhance customer relationship management capabilities. Constructing a time-series loss probability curve can dynamically monitor the user life cycle, assist in judging the changes in user value and intervention timing. It provides a quantitative basis for price intervention and retention strategies, improving the pertinence and effectiveness of the strategies. Executing commodity price recommendation operations based on personalized price push strategies to ensure that the recommended content meets the personalized needs of users and market rules. At the same time, through the VR visualization display of the three-dimensional form of the target commodity, it enhances the user's perception depth and interactive experience of the commodity. The VR display improves the intuitiveness and attractiveness of the commodity, helps users better understand the commodity value, promotes purchase decisions, and greatly enhances the immersion and satisfaction of the shopping experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 It is a schematic diagram of the step flow of a user commodity price recommendation and VR visualization method based on AI analysis according to the present invention; Figure 2 It is a schematic diagram of the detailed implementation steps of step S1; Figure 3 It is a schematic diagram of the detailed implementation steps of step S2; Figure 4 It is a schematic diagram of the detailed implementation steps of step S3. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0010] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0011] This application example provides a user commodity price recommendation and VR visualization method and system based on AI analysis. The execution subjects of the user commodity price recommendation and VR visualization method and system based on AI analysis include, but are not limited to, the following that carry this system: mechanical equipment, data processing platforms, cloud server nodes, network upload devices, etc., which can be regarded as general computing nodes of this application. The data processing platform includes, but is not limited to, at least one of an audio and image management system, an information management system, and a cloud data management system.

[0012] Please refer to Figures 1 to 4 , the present invention provides a user commodity price recommendation and VR visualization method based on AI analysis. The user commodity price recommendation and VR visualization method based on AI analysis includes the following steps: Step S1: Obtain the real-time price data stream of the target commodity based on the AI model, and conduct a time-series price fluctuation trend analysis to construct a time-frequency characteristic graph of price fluctuations; Step S2: Collect the user usage behavior logs of each platform, conduct dynamic mining of the user consumption trajectory, and evaluate the user price acceptance based on the time-frequency characteristic graph of price fluctuations to construct a user price elasticity model; Step S3: Identify the multi-time point sensitivity evolution based on the user price elasticity model, and then predict the optimal price adjustment timing to generate a time-series optimization strategy for price adjustment; Step S4: Predict the user interest transfer based on the user usage behavior logs, calculate the time-series churn probability, and construct a multi-time point user churn probability curve; Step S5: Dynamically constrain and optimize the solution of the multi-time point user churn probability curve according to the time-series optimization strategy for price adjustment and make a user personalized price push decision to construct a personalized price push strategy; Step S6: Execute the commodity price recommendation operation based on the personalized price push strategy; and conduct the VR visualization display operation of the three-dimensional form of the target commodity.

[0013] In the embodiment of the present invention, refer to Figure 1 , which is a schematic diagram of the step flow of a user commodity price recommendation and VR visualization method based on AI analysis of the present invention. In this example, the steps of the user commodity price recommendation and VR visualization method based on AI analysis include: Step S1: Obtain the real-time price data stream of the target commodity based on the AI model, and conduct a time-series price fluctuation trend analysis to construct a time-frequency characteristic graph of price fluctuations; In this embodiment, after obtaining the authorization of the user and the platform, the target products to be monitored, their identifiers (such as SKU, barcode, etc.) on major e-commerce platforms, and the relevant e-commerce platforms (such as Amazon, JD.com, Tmall, etc.) are determined. A crawler program is developed, and a scheduled task is set (for example, scraping once per hour) to automatically access the product pages of the specified e-commerce platforms and extract the current price data and relevant information (such as discounts, promotions, etc.). During the data collection process, the timestamp of each scrape is recorded to ensure the integrity of the time series of the data. Suppose the following price data is obtained within a certain period: Time 1: $20; Time 2: $21; Time 3: $19; The scraped price data is stored in a database to form a real-time price data stream, ensuring that the data can be conveniently analyzed and processed subsequently. Monitor the running status of the crawler to ensure the real-time and accuracy of the data, regularly check the data integrity, and avoid data loss caused by network problems or page structure changes. Extract the historical price data from the real-time price data stream to construct a price time series. Suppose the price data of each day is collected within the past week, forming the following time series: [Time 1: $20, Time 2: $21, Time 3: $19, Time 4: $22, Time 5: $21, Time 6: $20, Time 7: $23] The moving average method is used to smooth the price data, and the price changes in different time windows (such as 3-day moving average) are calculated to reduce the impact of short-term fluctuations. The calculation result may be: Moving Average 1 (from Time 3 to Time 5): ($19 + $22 + $21) / 3 = $20.67; By calculating the standard deviation and volatility of the price, the degree of price fluctuation is quantified. Suppose the price fluctuation range in the past week is $3 and the standard deviation is 0.75, indicating that the price fluctuation is relatively large. The time-frequency analysis technology (such as wavelet transform) is used to analyze the time-frequency characteristics of the price data, construct a time-frequency characteristic diagram of price fluctuation, and display the frequency components and time characteristics of the price fluctuation. Record the results of the fluctuation analysis, generate a price fluctuation trend analysis report, and describe in detail the change rules of the price and its potential influencing factors to provide data support for the subsequent optimization of the pricing strategy.

[0014] Step S2: Collect the user usage behavior logs of each platform, conduct dynamic mining of the user consumption trajectory, and evaluate the user price acceptance based on the time-frequency characteristic diagram of price fluctuation to construct a user price elasticity model; In this embodiment, the types of user behavior data to be collected are clarified, including browsing records, click counts, adding to the shopping cart, favoriting, and purchasing behaviors, etc. These data are the direct manifestations of users' activities on the platform. Develop a log collection system, set the data collection frequency (such as real-time or end-of-day batch processing), to ensure that all relevant user behaviors can be captured. For example, the system may be set to collect the behavior data of all users once a day. In data collection, record the timestamp and user ID of each behavior to ensure data integrity and traceability. Suppose the following behavior data is collected within a day: User A: Browsed product X 5 times, clicked 2 times, added to the shopping cart 1 time, and purchased 1 time. User B: Browsed product Y 3 times, clicked 1 time, and did not add to the shopping cart or purchase. Store the collected user behavior logs in a database to form structured data for subsequent analysis. At the same time, ensure data accuracy, regularly check the running status of the log system to avoid data loss or errors. Extract the consumption trajectories of users from the user behavior logs. The trajectory of each user can be regarded as a sequence, recording the various steps experienced during the purchase process. For User A, the consumption trajectory may be: Browsed product X → Clicked product X → Added to the shopping cart → Purchased product X. Use sequence pattern mining algorithms (such as GSP or PrefixSpan) to identify frequent consumption trajectory patterns. The analysis result may find that "Browsing → Clicking → Adding to the shopping cart → Purchasing" is a common consumption path. Record the consumption trajectory of each user and its corresponding conversion rate, and find that the conversion rate of users on this trajectory is 30%. This indicates that 30% of users finally complete a purchase after browsing. Generate a consumption trajectory analysis report, detailing the consumption patterns of different user groups, providing data support for subsequent price acceptance evaluation and elasticity model construction. Combine the user consumption trajectories with the time-frequency characteristics graph of price fluctuations to analyze the acceptance of users at different price levels. Suppose the price fluctuation range is $20 - $30, and users' purchase decisions at different price points are affected by the fluctuations. Set price acceptance evaluation indicators and calculate users' purchase willingness at different prices. When the price is $25, User A's purchase willingness is 80%, while it drops to 50% at $30. Use a linear regression model to construct a user price elasticity model based on users' behavior data and price response data. According to the collected data, evaluate users' sensitivity to price changes and calculate the price elasticity coefficient. Suppose the price elasticity coefficient is -1.5, indicating that users are highly sensitive to price increases. Record the evaluation results and generate a user price acceptance analysis report, detailing users' responses to price fluctuations and their impacts on purchase decisions.

[0015] Step S3: Based on the user price elasticity model, identify the evolution of sensitivity at multiple time points, then predict the optimal price adjustment timing, and generate a price adjustment timing optimization strategy; In this embodiment, the price elasticity coefficient is extracted from the user price elasticity model, and this coefficient reflects the sensitivity of users to price changes. During a certain period, assuming the elasticity coefficient is -1.5, it means that for every 1% increase in price, the demand quantity decreases by 1.5%. Multiple time points are set for sensitivity analysis, and user behavior data and price response data at these time points are collected. The selected time points are T1 (before promotion), T2 (during promotion), and T3 (after promotion), and the purchase situations of users and price changes are recorded. At each time point, the price elasticity of users is calculated through regression analysis. At time point T1, the price is $25 and the purchase quantity is 100 pieces, and the price elasticity may be calculated as -1.8; at time point T2, the price rises to $30 and the purchase quantity drops to 70 pieces, and the elasticity is calculated as -2.0. The elasticity coefficients at different time points are recorded to generate a time-series sensitivity evolution graph to show the reaction trend of users to price changes. The analysis results may show that the sensitivity of users during the promotion period increases significantly. Based on the results identified from the sensitivity evolution, a prediction model is set, considering multiple influencing factors such as user purchase history, seasonal changes, and market competition status. For example, if it is found that the sensitivity of users during the promotion season is higher than usual, then this time period needs to be focused on. Historical data is collected to establish a multivariate prediction model, and the input parameters include historical prices, user sensitivity, market trends, etc. Assuming that the model output predicts that the sensitivity of users to price changes in the next week is -2.5, it means that the impact of price adjustment at this time point will be greater. By simulating the effects of different price adjustment timings, it is evaluated when to adjust the price to maximize the profit. The simulation results show that adjusting the price when the user sensitivity reaches the peak can improve the conversion rate, and finally it is determined to make the adjustment during the sensitivity peak period. The prediction results are recorded to generate a price adjustment timing prediction report, which details the reasons for choosing the prediction timing and its expected effects, providing a basis for subsequent price adjustment strategies. According to the predicted user sensitivity and the optimal adjustment timing, a price adjustment strategy is formulated. A specific plan for adjusting the price during the sensitivity peak period (such as before and after a certain promotion activity) is set. At the sensitivity peak, the price is adjusted from $30 to $28 to increase the purchase rate. The expected profits under different price adjustment strategies are calculated through optimization algorithms (such as linear programming or dynamic programming) to evaluate the advantages and disadvantages of different plans. Assuming that after the adjustment, the expected sales volume will increase from 70 pieces to 90 pieces, and the profit will increase from $2100 ($30 × 70) to $2520 ($28 × 90). The expected profits and related parameters of each price adjustment strategy are recorded to generate a price adjustment timing optimization strategy report, which details the implementation steps and expected effects of the strategy to ensure flexible adjustment according to market changes.

[0016] Step S4: Based on the user usage behavior log, predict the user interest transfer and calculate the time-series churn probability, and construct a multi-time-point user churn probability curve; In this embodiment, key information is extracted from the user usage behavior logs, including the user's browsing history, click behavior, purchase records, and timestamps. Suppose a user's behavior records in the past month are as follows: viewed product A 10 times, product B 5 times, product C 2 times, and finally purchased product A. These behavior data are converted into feature vectors, and the set features include: product browsing frequency; click-through rate; purchase conversion rate; relative interest change between products; and use historical data to train an interest transfer prediction model. The accuracy of the model is evaluated through cross-validation. Suppose the accuracy rate of the training set reaches 85%. The model may predict that the probability of the user's interest transfer to product B is 70%, which means that the user has a relatively high possibility of turning to focus on product B. Record the interest transfer prediction results of each user, generate a user interest transfer analysis report, and describe the interest changes of different user groups and their potential churn risks. Set the definition of churn, usually that the user does not make a purchase or log in within a certain period of time. Suppose the definition of churn is that the user does not have any interaction in the past 30 days. Extract historical churn data from the user behavior logs, construct a time series data set, and record the active status of each user. Record the active situation of user A in the past 60 days: Day 1: Active; Day 10: Active; Day 30: Inactive; Day 40: Inactive; Use a survival analysis model (such as Kaplan-Meier estimation or Cox proportional hazards model) to calculate the user's churn probability. Suppose it is found in the analysis that the churn probability of user A is 40% on the 30th day and 60% on the 60th day. Record the churn probability of each user at different time points, generate a multi-timepoint user churn probability curve, and show the changing trend of the user churn risk over time. The change in the churn probability curve may reveal that the user churn risk increases during certain periods, such as after a promotion ends. Generate a churn probability analysis report, describe in detail the changes in the churn probability and its potential influencing factors, and provide data support for subsequent user retention strategies.

[0017] Step S5: Dynamically constrain and optimize the multi-timepoint user churn probability curve according to the price adjustment time series optimization strategy and make a user personalized price push decision to construct a personalized price push strategy; In this embodiment, integrate the previously obtained multi-timepoint user churn probability curve and the price adjustment time series optimization strategy to establish an optimization model. The set objective functions include: Maximize the user's purchase conversion rate; Minimize the churn probability; On this basis, maximize the revenue; Set dynamic constraint conditions, such as: The probability of user churn cannot exceed a certain threshold (e.g., 30%); Limit on price adjustment range (e.g., each adjustment shall not exceed 10%); Use historical data for model training. Assume that the existing data shows that when the price is $30, the probability of user churn is 40%, and when the price is $28, the churn probability drops to 25%. Use an optimization algorithm to solve the model and obtain the optimal price adjustment plan at different time points. The optimization result shows that on the 5th day, adjusting the price from $30 to $28 is expected to reduce the churn probability to 25%. Record the results of each optimization, generate a dynamic constraint optimization analysis report, describe the optimization process and the final price adjustment strategy obtained, and provide a basis for subsequent personalized price push decisions. According to the user churn probability and the prediction results of interest transfer, divide users into different groups. Assume that users can be divided into high-churn-risk users and low-churn-risk users: High-churn-risk users: The churn probability exceeds 30%; Low-churn-risk users: The churn probability is lower than 30%; Develop personalized price strategies for different user groups. For high-churn-risk users, offer more attractive discounted prices (e.g., adjust from $30 to $27) to reduce the churn probability; for low-churn-risk users, maintain the original price or make a slight adjustment (e.g., $29). Send personalized price information to target users through a push system (such as emails, APP notifications, etc.) and set up a feedback mechanism to encourage users to provide feedback on their price perception. User A clicks to purchase after receiving the $27 discount and gives a feedback satisfaction of 90%. Record the user's reactions and purchase situations, evaluate the effectiveness of the personalized price push strategy, generate a personalized price push strategy report, and describe in detail the implementation effect and subsequent optimization suggestions.

[0018] Step S6: Execute the commodity price recommendation task based on the personalized price push strategy; and perform the 3D shape VR visualization display task of the target commodity.

[0019] In this embodiment, the price recommendation operation execution and the three-dimensional form VR visual display are linked with the real-time price index as the bridge. In terms of the execution of the commodity price recommendation operation, the system is based on the user portrait, integrating multi-dimensional data such as user interest tags (such as "smart home preference", "mid-to-high-end consumption"), behavior trajectories (clicking, staying, jumping paths, etc.), historical price responses (discount sensitivity, promotion participation), and the current market situation (such as holiday promotions, inventory pressure). The core of the system uses multi-objective optimization algorithms (such as MOEA / D or the reinforcement learning PPO model) to generate pricing strategies, and the objective function covers maximizing user satisfaction, enhancing platform revenue, and maintaining market competitiveness. The platform selects 200,000 user behavior data for training, extracts 300 key feature dimensions, controls the RMSE below 0.06 during model training, and finally outputs an "optimal price recommendation" bound to the user group and specific commodity. This price is dynamically corrected in combination with the current cost of the commodity, the market price index, and the user acceptance boundary, and is pushed to users for recommendation display. The recommendation module is iteratively trained once a day, and the response latency is controlled within 200ms to achieve near-real-time pricing feedback. In the part of the three-dimensional form VR visual display operation, the system introduces the three-dimensional point cloud model of the target commodity constructed before, and obtains a high-quality three-dimensional model through surface mesh reconstruction technology (such as Poisson Surface Reconstruction). The model has complete texture, color, and structural features, and the accuracy is controlled at the millimeter level. Then, the three-dimensional model is loaded into the VR environment through the Unity3D engine, and combined with the user recommendation result, the commodity is presented in the virtual display space in the form of "at the current recommended price". In the visual presentation, the price index is mapped to a color gradient (for example: when the index is higher than 0.8, bright red is used to represent the price peak, and when it is lower than 0.3, blue is used to represent a discount opportunity), and visual cues can be used on the model to allow users to intuitively feel the price floating range. If the user stays in the VR for a long time or interacts with the model (such as rotating, zooming, magnifying a part), the system will automatically display the historical trend of this price, comparison with similar commodities, and the price recommendation explanation box.

[0020] In this embodiment, refer to Figure 2 , which is the schematic diagram of the detailed implementation steps of step S1. In this embodiment, the detailed implementation steps of the said step S1 include: Based on the AI model, perform real-time price tracking of the target commodity on multiple platforms to construct the real-time price data stream of the target commodity; Perform outlier adaptive identification on the real-time price data stream of the target commodity, and perform outlier elimination processing to obtain the abnormal optimized price data stream; Perform dynamic time window segmentation on the abnormal optimized price data stream to obtain a sequence of commodity price data streams with multiple time windows; Perform a time-series price fluctuation trend analysis on the commodity price data stream sequence to obtain price fluctuation trend characteristics; Extract the periodic characteristics, mutation characteristics, and trend characteristics of the price fluctuation trend characteristics, and perform multi-level decomposition to construct a price fluctuation time-frequency characteristic diagram.

[0021] In this embodiment, a real-time price data stream for a target product is constructed to obtain the latest price information for the target product on multiple platforms for subsequent analysis and decision-making. After obtaining authorization from the user and the relevant platform, the target product and its identification information on major e-commerce platforms (such as Amazon, JD.com, and Tmall) are determined, and a unique product identifier (such as a SKU or barcode) is established. A crawler program is developed to periodically (e.g., hourly) access each e-commerce platform to extract product price information and related attributes (such as discounts and promotions). A timestamp is set to record each captured data, ensuring that each data item is recorded with the acquisition time, forming a time series data stream. Suppose that within a certain time period, the captured price data is: [Time 1: $20, Time 2: $21, Time 3: $19]. The captured data is stored in a database to form a real-time price data stream for subsequent analysis and processing. The completeness and accuracy of the data are recorded to ensure that the data on each platform is updated in a timely manner. Descriptive statistical analysis is performed on the real-time price data stream, calculating the mean and standard deviation of the price within each time window. Assume that the mean price for a certain time period is $20 and the standard deviation is $1.5. Set an outlier threshold. Prices with a Z-score exceeding 3 are considered outliers. Calculate the Z-score for each price to identify outliers. For example, if a captured price is $25, the Z-score calculation may indicate it is an outlier. Identified outliers are removed to form an outlier-optimized price data stream. The resulting price data after removal is: [Time 1: $20, Time 2: $21, Time 3: $19]. Record the number of outliers removed and the reasons for removing them, and generate an outlier handling report for subsequent analysis and review. Determine the length of the time window, dividing it into hourly units, for example, each window is 1 hour. Assuming the total data stream contains 24 hours of data, divide it into 24 windows. Use the sliding window technique, moving forward one time unit at a time, to extract price data for the corresponding time period. Window 1 contains [Time 1, Time 2], Window 2 contains [Time 2, Time 3], and so on. Record the price data for each time window to form a multi-window commodity price data stream. The prices in window 1 are [$20, $21], and the prices in window 2 are [$21, $19]. Summarize each window and record basic statistical indicators such as the price mean and fluctuation range to provide basic data for subsequent price fluctuation trend analysis. Perform trend analysis on the data in each time window and calculate a moving average (such as a 3-hour moving average) to smooth price fluctuations. Calculate the average price of windows 1-3, assuming the resulting moving average is $20.5. Identify price fluctuation characteristics, such as sudden changes and cyclical fluctuations. Observe the price chart and record significant change points in price fluctuations. For example, suppose the price suddenly jumps from $20 to $25 in a window. Record this sudden change point. Analyze the trend characteristics of price fluctuations to identify upward, downward, and stable trends.If the price has been gradually rising in the past 6 hours, mark this period as an uptrend. Record the results of the trend analysis, generate a price fluctuation trend report, and describe in detail the fluctuation characteristics of the price, the mutation points, and their potential impact on the market. Decompose the price fluctuation trend, and use wavelet transform or Fourier transform to extract the periodic characteristics. Assume that an obvious period of 7 hours is found in the data, record and mark this period. Identify the mutation characteristics, analyze the sharp changes in the price during certain time periods, and calculate the change amplitude. If the price change exceeds 10% within a certain period of time, record it as a mutation characteristic. Extract the trend characteristics, analyze the long-term trend of the price, such as the uptrend or downtrend, and calculate the relevant slope and change rate. If the price has risen by 15% overall in the past 24 hours, mark it as an uptrend. Generate a price fluctuation characteristic report, and describe in detail the extracted periodic characteristics, mutation characteristics, and trend characteristics for subsequent decision-making and optimization. Utilize the extracted periodic characteristics, mutation characteristics, and trend characteristics to construct a time-frequency characteristic diagram. Perform time-frequency conversion on the price fluctuation data to generate a spectrogram. Display the time-frequency characteristics of the price fluctuation through a heat map or a three-dimensional graph. Use the shade of color to represent the intensity of the price fluctuation in different time periods to help users intuitively understand the price changes. Record the generation process and analysis results of the characteristic diagram, generate a time-frequency characteristic diagram report, and describe in detail the price fluctuation situation and its potential impact. Integrate the characteristic diagram into the user decision support system to provide users with a basis for real-time market dynamic analysis and trend prediction.

[0022] In this embodiment, refer to Figure 3 , which is a schematic diagram of the detailed implementation steps of step S2. In this embodiment, the detailed implementation steps of step S2 include: Obtain the user usage behavior logs of multiple platforms; extract the behavior data of user browsing, clicking, favoriting, adding to the shopping cart, and purchasing according to the user usage behavior logs, and mark all the usage behavior particle sequences of the users; Based on the usage behavior particle sequences, conduct dynamic mining of the user consumption trajectory and extract the user consumption trajectory; Perform multi-behavior inertia and momentum estimation on the user consumption trajectory to obtain the consumption inertia and momentum under the influence of different behavior particles; According to the consumption inertia and momentum, conduct user value sensitivity analysis and construct a user price sensitivity curve; According to the sensitivity curve and the time-frequency characteristic diagram of the price fluctuation, conduct user price acceptance evaluation and construct a user price elasticity model.

[0023] In this embodiment, identify the e-commerce platforms to be collected (such as Amazon, JD.com, Tmall, etc.) and their related API interfaces to ensure that user behavior log data can be obtained. Develop a crawler program to regularly capture user behavior logs, including behaviors such as browsing, clicking, favoriting, adding to cart, and purchasing. Assume that the log data collected in a certain week is as follows: Browsing: 1000 times; Clicking: 300 times; Favoriting: 150 times; Adding to cart: 200 times; Purchasing: 50 times; Store the collected log data in a database to ensure data integrity and accuracy. At the same time, record the timestamp and user ID of each piece of data for subsequent analysis. Set the data update frequency, for example, update once a day to ensure that the latest user behavior data is used in the analysis process. Filter out relevant behavior data from the obtained user behavior logs to form a behavior dataset. Record the behavior sequence of a certain user as: User ID: 123; Behavior sequence: [browsing, clicking, favoriting, adding to cart, purchasing]; Mark the behavior sequence of each user to form a usage behavior particle sequence. Assume that the behavior particle sequence of user A is: [browsing, clicking, adding to cart, purchasing].

[0024] Integrate the particle sequences of all users to form a complete user behavior dataset for subsequent consumption trajectory mining. Record the behavior timestamp of each user to understand the time series characteristics of their behaviors in subsequent analysis. Use sequence pattern mining algorithms (such as GSP or PrefixSpan) to analyze the user behavior particle sequences to extract the consumption trajectories of users. The analysis results show that the common consumption trajectory of users is: [browsing → clicking → adding to cart → purchasing].

[0025] Construct a consumption trajectory graph based on the user's behavior sequence to show the user's behavior path and key nodes. Assume that in the dataset, the click-through rate is 30% and the collection rate is 15% after the user browses. Record the frequency and conversion rate of the consumption trajectory for subsequent analysis of the user's behavior patterns and preferences. Generate a consumption trajectory analysis report, describing in detail the user's behavior habits and their potential influencing factors. The goal of multi-behavior inertia and momentum estimation is to quantify the impact of different user behaviors on the consumption trajectory, thereby evaluating the user's consumption inertia. Use statistical analysis methods (such as regression analysis) for estimation. Conduct statistical analysis on the extracted user consumption trajectories to calculate the inertia and momentum of different behaviors. Assume that the inertia and momentum of the browsing behavior is 0.4 and the inertia and momentum of the click behavior is 0.6. Set the calculation formula for inertia and momentum, considering the frequency and conversion rate of the behavior occurrence to quantify the impact of different behaviors on the final purchase decision. If a user's click-through rate is 40%, then the impact momentum on the purchase decision can be estimated as: Impact momentum = click-through rate × conversion rate; Record the inertia and momentum of each behavior to form a user behavior momentum database for subsequent analysis and comparison. Generate an inertia and momentum estimation report, describing the impact degree of different behavior particles on the consumption trajectory to help understand the user's decision-making process. Based on the user's consumption trajectory and inertia and momentum, establish a user value sensitivity analysis model. Assume that the sensitivity threshold is set at 10%. If the price increases by more than this value, the user's purchase intention will decrease significantly. By analyzing the user's purchase history, calculate the user's sensitivity to price changes. If in past transactions, the user's reaction to a price increase was: When the price increases by 5%, the purchase rate decreases by 10%; When the price increases by 10%, the purchase rate decreases by 20%; Record the user's price sensitivity data, generate a user price sensitivity curve to show the change in the user's purchase intention at different price levels. Generate a sensitivity analysis report, describing in detail the user's reaction to price changes and providing data support for the pricing strategy. Combine the user price sensitivity curve with the price fluctuation time-frequency characteristic graph to analyze the user's acceptance under different price fluctuation conditions. When the price fluctuates greatly, the user's sensitivity to price may increase. Through scenario simulation, evaluate the user's acceptance under different price strategies. If the price of a certain commodity increases from $20 to $24, simulate the change in the user's purchase intention and record its acceptance. Generate a user price elasticity model to show the degree of the user's reaction to price changes, helping enterprises formulate more competitive pricing strategies. Record the evaluation results, generate a price acceptance analysis report, and provide suggestions on how to adjust the price strategy to increase sales.

[0026] In this embodiment, the specific steps for evaluating the user's price acceptance according to the sensitivity curve and the price fluctuation time-frequency characteristic graph and constructing the user price elasticity model are as follows: Identify price fluctuation mutations in the time-frequency characteristic graph of price fluctuations and mark the inflection points of price fluctuations; Calculate the fluctuation time length and fluctuation gradient of the inflection points of price fluctuations; Extract the initial occurrence timestamp of the inflection points of price fluctuations; Perform spatio-temporal matching mapping on the user price sensitivity curve based on the initial occurrence timestamp, and perform sensitivity correlation response evolution according to the fluctuation time length and fluctuation gradient to construct a price fluctuation-sensitivity response correlation network; Perform in-depth sensitive diffusion topology analysis based on the price fluctuation-sensitivity response correlation network to extract the user price sensitivity propagation path; Evaluate the user price acceptance based on the user price sensitivity propagation path and construct a user price acceptance elasticity model.

[0027] In this embodiment, price data is extracted from the time-frequency characteristic graph of price fluctuations, and a threshold for mutation identification is set. Taking the standard deviation as the benchmark, if the price fluctuation exceeds ±2 standard deviations, it is considered that a mutation has occurred. By calculating the derivative of the price change, the inflection points of the fluctuation are identified. If the price changes in a certain time period are [20, 22, 25, 19, 30], then during the process of the price jumping from 19 to 30, the inflection point is marked. Record all identified inflection points of price fluctuations and generate an inflection point data set, including the timestamp, price value, and change amplitude. Assume the marked inflection point data is: Inflection point 1: Timestamp 1, price $25; Inflection point 2: Timestamp 2, price $30; Prepare the marked price fluctuation inflection point data for subsequent analysis to ensure data integrity and accuracy. Analyze the marked price fluctuation inflection points and calculate the fluctuation time length for each inflection point. If the time difference from inflection point 1 to inflection point 2 is 2 hours, then the fluctuation time length is 2 hours. Calculate the fluctuation gradient, defined as the ratio of the price change amplitude to the fluctuation time length. Record the fluctuation time length and fluctuation gradient for each inflection point to form an inflection point analysis report, including the characteristic information of the inflection points. Generate a comprehensive dataset containing the inflection point time length and gradient for subsequent sensitivity correlation analysis. Extract the initial occurrence timestamps from the marked price fluctuation inflection points to ensure that each inflection point has a corresponding time record. If the timestamp of inflection point 1 is 2023-05-01 12:00 and inflection point 2 is 2023-05-01 14:00, then record these timestamps. Generate an initial timestamp dataset containing the time information of all inflection points for subsequent spatio-temporal matching with the user price sensitivity curve. Ensure the accuracy of the timestamps, record the time zone information and collection method, and avoid data deviation caused by time differences. Integrate the extracted timestamps with the price fluctuation related data to provide basic data for subsequent analysis. Match the extracted initial occurrence timestamps with the user's price sensitivity curve. If the sensitivity of the user is high near the price fluctuation inflection point, then record the sensitivity value at that point. Estimate the price sensitivity at the time of the inflection point through interpolation to generate spatio-temporal matching data. If the timestamp of inflection point 1 is 12:00 and the sensitivity is 0.7, then record it as: Inflection point 1: Timestamp 2023-05-01 12:00, Sensitivity 0.7 to form a corresponding relationship dataset between price fluctuations and user sensitivity, providing a basis for subsequent analysis. Record the results of spatio-temporal matching, generate a matching report, and describe in detail the impact of price fluctuations on user behavior. Based on the fluctuation time length and fluctuation gradient, establish a sensitivity correlation response model. Assume that the fluctuation time length is 2 hours and the gradient is 2.5, and construct a model to evaluate the change in user sensitivity under these conditions. Through regression analysis, explore how the fluctuation characteristics affect the user's price sensitivity. If the longer the fluctuation time, the higher the user's sensitivity, then a relationship formula between sensitivity and fluctuation characteristics can be obtained. Record the parameters and results of the model, generate a sensitivity response evolution report, and describe the change in user sensitivity under different fluctuation characteristics. Form a response evolution dataset to provide a basis for subsequent sensitivity propagation analysis. Determine the network nodes and edges. The nodes represent the price fluctuation inflection points and user sensitivity, and the edges represent the relationship between the two. If inflection point 1 is connected to sensitivity 0.7, then record it as an edge. Use network analysis software (such as Gephi) to draw an association network diagram to show the relationship between price fluctuations and sensitivity. Distinguish different degrees of association intensity through colors and sizes. Record the results of network analysis, generate a price fluctuation-sensitivity association network report, and describe the network structure and key nodes. Through network analysis, identify the fluctuation characteristics that have the greatest impact on price sensitivity to provide data support for subsequent decision-making.Analyze the topological structure of the price volatility-sensitivity response correlation network, identify key nodes and propagation paths. Use the PageRank algorithm to evaluate the importance of nodes and determine the core nodes of sensitivity propagation. Record the sensitivity propagation paths, generate a topological analysis report, and describe the diffusion characteristics of sensitivity in the network and its influencing factors. Evaluate the sensitivity propagation effect under different volatility characteristics and analyze its potential impact on user behavior. Form a propagation path dataset to provide a basis for subsequent user acceptance evaluation. According to the user price sensitivity propagation path, evaluate the user's acceptance of different price changes. If the sensitivity path shows that the user's acceptance decreases when the price rises, record this result. Through model regression analysis, construct a user price acceptance elasticity model to evaluate the degree of the user's response to price changes. Assume that when the price increases by 5%, the user's purchase intention decreases by 20%. Record the evaluation results, generate a user price acceptance analysis report, and describe in detail the changes in acceptance and its impact on purchase decisions. Provide suggestions on how to optimize the pricing strategy to improve user acceptance and form a complete price elasticity model.

[0028] In this embodiment, refer to Figure 4 , which is a schematic diagram of the detailed implementation steps of step S3. In this embodiment, the detailed implementation steps of step S3 include: Perform multi-dimensional clustering on user groups based on the user usage behavior log to obtain multiple user group clustering clusters; Perform dynamic time-series evolution on the user price elasticity model and extract the user price sensitivity critical point; Based on the user price sensitivity critical point, perform adaptive matching on multiple user group clustering clusters and perform vectorization processing of price acceptance to obtain price preference characteristics of different user groups; Perform multi-timepoint sensitivity evolution identification according to the price preference characteristics of different user groups to obtain the sensitivity change characteristics under the user price preference at different time points; Predict the optimal price adjustment timing based on the sensitivity change characteristics under the user price preference and generate a time-series optimization strategy for price adjustment.

[0029] In this embodiment, extract key features from the user behavior log, such as the number of views, click-through rate, number of times added to the shopping cart, purchase frequency, and price sensitivity, etc., to form a user feature dataset. Assume the extracted features are: User A: Number of views 100, click-through rate 0.3, purchase frequency 0.05; User B: Number of views 150, click-through rate 0.4, purchase frequency 0.1; Standardize the data to ensure that different features are within the same dimension and avoid the excessive influence of a certain feature. Use the K-means clustering algorithm, set the number of clusters K (for example, 3), cluster the users, and generate multiple user groups. The clustering results may be: Group 1: High-frequency purchasing users; Group 2: Low-frequency browsing users; Group 3: Potential customers; Record the features and the number of users in each cluster, generate a user group clustering report for subsequent analysis and decision-making. Combine the purchase history and price change data of users to build a user price elasticity model and set the elasticity calculation formula. Price elasticity can be defined as: Price elasticity = Percentage change in demand / Percentage change in price; For different user groups, conduct dynamic time series analysis to identify the changes in price elasticity at different time periods. During the promotion period, the price elasticity of a certain user group may increase to -2.0. Extract the critical points of user price sensitivity, that is, the points where significant changes in demand are caused by price changes, and record these critical points and their corresponding times. It is found that when the price is $20, the demand drops significantly, and the critical point is $20. Generate a dynamic evolution report to describe the trend of user price elasticity changing over time and its influencing factors. Match the extracted user price sensitivity critical points with the characteristics of each user group. For user groups with higher price sensitivity, match price strategies with critical points between $15 - $20. Generate a price acceptance vector, convert the critical point information into vector form for subsequent analysis. Assume the price acceptance vector of user group 1 is: [15, 18, 20]; Generate corresponding price acceptance vectors for each user group to form a user price preference feature dataset. Record the matching results and generate an adaptive matching report to detail the price acceptance characteristics of each user group. Analyze the changes in their price sensitivity for different user groups with time as the dimension. During a specific promotion period, record the sensitivity of user groups to price changes.

[0030] Assume that at different time points, the sensitivity of users to price changes is: Time point 1: -1.5; Time point 2: -2.0; Time point 3: -0.8; Identify the evolving characteristics of sensitivity, record the sensitivity changes at different time points, and analyze the reasons. It is found that the sensitivity increases significantly during promotions. Generate a sensitivity evolution identification report to describe the user price preferences and their changing trends at different time points. Based on the user price preferences and sensitivity change characteristics, construct a price adjustment timing prediction model. Use machine learning algorithms (such as random forest or support vector machine) to predict the best timing for price adjustment. Set the prediction parameters, including the price change range, user sensitivity, historical purchase data, etc. Assume that during the data training process, the prediction accuracy of the model for price adjustment is 85%. Through model prediction, identify the best price adjustment timing. For example, at a certain time point, when the user's price sensitivity reaches the peak, adjusting the price may bring the greatest benefit. Record the prediction results and generate a price adjustment timing prediction report to provide specific suggestions on price adjustment.

[0031] In this embodiment, step S4 includes the following steps: Extract the platform historical marketing intervention records and short-term hot spot propagation trend data based on the user usage behavior logs; Perform commodity interest offset perception on the platform historical marketing intervention records and short-term hot spot propagation trend data to obtain a potential interest offset vector; Calculate the user expected price gap for the price fluctuation time-frequency characteristic diagram to obtain a user expected price gap tensor; Perform expected difference gradient analysis on the user expected price gap tensor to obtain the positive and negative gradient curves of the user expected difference; Perform user interest transfer prediction based on the potential interest offset vector and the positive and negative gradient curves of the user expected difference to obtain an interest transfer risk prediction value; Calculate the time series churn probability based on the interest transfer risk prediction value and construct a multi-time point user churn probability curve.

[0032] In this embodiment, extract the marketing activity records participated by users from the user behavior logs, including information such as promotions, discounts, and advertisement displays. Assume that in the past three months, the platform has carried out 5 large-scale promotion activities, and the records are as follows: Activity 1: 20% discount; Activity 2: Reduce 50 yuan for purchases over a certain amount; Activity 3: Buy one get one free; Meanwhile, collect short-term hot topic spread data related to these activities, such as user discussions on social media, search trends, etc. Suppose during Activity 1, the discussion volume of relevant topics increased by 150%. Integrate this data to form a comprehensive dataset containing user reactions, activity types, and spread trends for subsequent analysis. Record the time period and source of data extraction to ensure the accuracy and integrity of the data, providing a reliable basis for subsequent analysis. The goal of perceived commodity interest deviation is to identify changes in users' interest in commodities after participating in different marketing activities, forming a potential interest deviation vector. Use user interest models and sentiment analysis techniques for perception. Analyze the changes in users' interest in commodities based on the extracted historical marketing records. The click-through rate and purchase rate of a certain commodity before and after the activity may show obvious changes. Design an interest deviation metric to calculate the interest change of users before and after the activity. If the click-through rate of a certain commodity for a user before Activity 1 was 0.2, and it rose to 0.4 after the activity, then the calculated deviation is: Deviation = Click-through rate after activity - Click-through rate before activity = 0.4 - 0.2 = 0.2; Integrate the interest deviation data of different users to form a potential interest deviation vector. The deviation vector of User A may be [0.2, -0.1, 0.3], indicating the interest changes in three commodities. Record the deviation vectors of users and generate an interest deviation analysis report to describe the impact of different marketing activities on users' interest. Extract the expected price and actual price data of users from the time-frequency characteristics graph of price fluctuations. Suppose the expected price of a user during a certain time period was $50, while the actual price was $60. Calculate the user's expected price gap, defined as the difference between the actual price and the expected price: Gap = Actual price - Expected price = 60 - 50 = 10; Integrate the expected price gaps of all users to form a user expected price gap tensor, recording the expected gap of each user and the corresponding timestamp. Generate an expected price gap analysis report to describe the gap between users' expectations and the actual situation regarding price changes and its potential impact. Conduct a gradient analysis on the user expected price gap tensor to calculate the rate of change of the price gap at each time point. If the gaps at consecutive time points are [10, 8, 12], then calculate the gradient between adjacent time points: Gradient = Gap(t + 1) - Gap(t) = 12 - 8 = 4; Record the gradient at each time point and generate positive and negative gradient curves to identify users' reactions to price fluctuations. If the gap continuously increases within a certain time period, it is marked as a positive gradient. Generate an expected difference gradient analysis report to describe users' price sensitivity and its characteristics over time. Combine the potential interest deviation vector with the positive and negative gradient curves of users' expected differences to design a prediction model, with input variables including user interest deviation and price gap gradient. Use a linear regression model to predict the interest transfer risk value. Train the model to identify potential interest transfer risks. Suppose the risk prediction value output by the model is 0.75, indicating a relatively high interest transfer risk.Record the predicted values of interest transfer for each user, generate an interest transfer risk analysis report, and describe the interest change risks of different user groups. According to the predicted values of interest transfer risk, use the survival analysis model to calculate the churn probability of users. Assume that the churn probability model obtains the churn probability formula from historical data: Churn probability = 1−. ; where λ is the churn rate and t is the time. Record the churn probabilities at different time points and generate a multi-time point user churn probability curve. The churn probability at time point 1 is 0.2 and rises to 0.4 at time point 2. Generate a churn probability analysis report, which details the risks of user churn and its trend over time, providing data support for subsequent user retention strategies.

[0033] In this embodiment, step S5 includes the following steps: Define the market competition situation and the enterprise profit margin, and perform multi-objective optimization design based on the multi-time point user churn probability curve to obtain a multi-objective price optimization function; Perform dynamic constraint optimization and solution on the multi-objective price optimization function according to the price adjustment time sequence optimization strategy to obtain a price adjustment parameter set; Conduct a user acceptance evaluation simulation on the price adjustment parameter set to extract the optimal price adjustment plan; Make a user personalized price push decision on the optimal price adjustment plan and construct a personalized price push strategy.

[0034] In this embodiment, pricing information, product features, and market shares of competitors are collected, and data is obtained through industry reports, market research, user feedback, etc. Assume that the market share of competitor A is 30% and the price is $50; the market share of competitor B is 25% and the price is $55. Calculate the company's own market share and profit margin. Assume that the company's current price is $45, the sales volume is 1000 units, and the cost is $30. Then the profit margin is: Profit margin = (Selling price - Cost) / Selling price = (45 - 30) / 45 ≈ 33.33%; Combine market data to draw a market competition situation map, analyze the advantages and disadvantages of competitors, identify potential market opportunities and threats, and form a market competition analysis report. Determine the optimization objectives, including maximizing profit, minimizing churn rate, and increasing market share. Set the objective function, for example: Objective function = max(Profit) and min(Churn probability); Use the user churn probability curve data to analyze the churn probability at different price levels. When the price is $45, the churn probability is 20%; when the price is $50, the churn probability rises to 30%. Use the multi-objective optimization algorithm to solve the objective function to obtain the multi-objective price optimization function. The optimization result may show that within the price range of $45 - $55, the best pricing is $50, which balances profit and churn risk. Record the optimization result, generate a multi-objective optimization analysis report, and describe the optimization process and the final recommended pricing strategy. Set dynamic constraint conditions, such as market demand changes, user feedback, and competitor price adjustments. Assume that at a specific point in time, the competitor's price is adjusted to $52. According to these constraint conditions, dynamically adjust the multi-objective price optimization function and use the linear programming method to solve the optimization problem. The new price adjustment parameter set obtained by solving is: New price 1: $48; New price 2: $50; Record the parameters of each adjustment and their corresponding churn probabilities and profit changes to ensure competitiveness in a dynamic market environment. Generate a price adjustment strategy report describing the dynamic optimization process and the final set of price adjustment parameters. Based on historical data and user behavior models, construct a user acceptance evaluation simulation model to simulate user purchase behavior at different prices. Set the new price at $48 and predict a user acceptance rate of 70%. Conduct multiple simulations using the Monte Carlo simulation method and record user feedback and purchase rates at different prices. At $48, the expected sales volume is 800 units; at $50, the expected sales volume is 600 units. Analyze the simulation results, extract the optimal price adjustment plan, record the user acceptance and sales expectations, and ensure the feasibility of the new price in the market. Generate a user acceptance evaluation report detailing the simulation process and the final recommended price adjustment plan. Segment users based on user behavior data (such as purchase history, browsing records) to identify the price sensitivity of different user groups. User A is more sensitive to price, while User B has high brand loyalty. Develop personalized price push strategies for different user groups. Push a discounted price of $45 to User A and provide a brand recommendation of $50 to User B. Use a push system (such as emails, APP notifications) to send personalized price information to target users to ensure timely delivery. Record the purchase situations of users after the push, evaluate the effectiveness of the personalized strategy, and generate a personalized push strategy report describing the implementation effect and subsequent optimization suggestions.

[0035] In this embodiment, step S6 includes the following steps: Perform user group targeted price push based on the personalized price push strategy, construct a differential price push engine, and collect user price response feedback information; Calculate the price adjustment revenue index based on the user price response feedback information and extract the price adjustment revenue index; Evaluate the price adjustment effect of the user price response feedback information to obtain a price adjustment effect evaluation report; Conduct feedback optimization iteration analysis on the price adjustment effect evaluation report and the price adjustment revenue index to construct an adaptive revenue adjustment-effect feedback model; Based on the adaptive revenue adjustment-effect feedback model, perform dynamic pricing matching optimization on the differential price push engine to construct an intelligent price matching optimization model to execute the commodity price recommendation operation; Generate a real-time price index for the target commodity based on the personalized price push strategy, and perform three-dimensional form analysis and VR visualization display processing on the target commodity.

[0036] In this embodiment, collect user behavior data, including purchase history, browsing habits, price sensitivity, etc., to form a user feature dataset. Assume the extracted features include: User A: Number of views 150, price sensitivity 0.8; User B: Number of views 80, price sensitivity 0.4; Through K-means clustering analysis, set the number of clusters K (for example, 3), and divide users into high-sensitivity users, medium-sensitivity users, and low-sensitivity users. The results may show: Group 1: High-sensitivity users; Group 2: Medium-sensitivity users; Group 3: Low-sensitivity users; For each user group, design a personalized price push strategy. Push discounted prices to highly sensitive users and keep the original price for less sensitive users. Record the results of user segmentation and their corresponding targeted price push strategies to generate a user segmentation report. The goal of the differential price push engine is to automatically push personalized price information according to the characteristics of different user groups and collect user feedback. Implement it using recommendation system technology and user feedback mechanisms. Develop a differential price push engine, integrate the results of user segmentation and personalized price strategies, and design the push logic. Push a discount of $30 to highly sensitive user A, while push a price of $50 to less sensitive user B. Push price information through multiple channels (such as email, APP push notifications, etc.) and set up a feedback mechanism to encourage users to provide feedback on their price perception, such as through "satisfaction rating" or "purchase or not" options. Collect user price response feedback information, for example, the acceptance rate of user A for the $30 price is 90%, while the acceptance rate of user B for the $50 price is 40%. Record the feedback data of all users to form a feedback information database, providing a basis for subsequent analysis. Calculate the revenue metrics for each price adjustment based on the user price response feedback information. Assume that the purchase volume of user A after the push is 150 units and that of user B is 50 units. Calculate the total revenue. Assume that when the price is $30, the total revenue of user A is: Revenue = selling price × purchase volume = 30 × 150 = 4500; at the same time, calculate the profit, considering the cost (assume it is $20), the profit is: Profit = (selling price - cost) × purchase volume = (30 - 20) × 150 = 1500; Integrate the revenue and profit metrics of all users to generate a price adjustment revenue metrics report, detailing the revenue situation of each user group. Analyze the user price response data and calculate the purchase conversion rate before and after the adjustment. Assume that the conversion rate of user A before the adjustment is 40% and it rises to 60% after the adjustment. Record the user feedback and behavior changes after each price adjustment and analyze their impact on the purchase decision. The overall change in the purchase volume of the user group. Generate a price adjustment effect evaluation report, describing the impact of price adjustment on user behavior and its effects, pointing out which price strategies are effective and which need to be optimized. Combine the data in the price adjustment effect evaluation report with the revenue metrics to identify well-performing strategies and areas for improvement. Find that the strategy with a price of $30 performs excellently among highly sensitive users, while the strategy with a price of $50 has poor effects on less sensitive users. Use adaptive optimization algorithms (such as genetic algorithms or particle swarm optimization) for iterative analysis to adjust the price strategy to maximize the overall revenue. The adjusted price strategy for less sensitive users is $48. Record the results of each iteration to generate a feedback optimization analysis report, describing the optimization process and the final strategy recommendations. Integrate the output of the adaptive revenue adjustment - effect feedback model to construct an intelligent price matching optimization model to monitor market changes and user feedback in real time.Set dynamic pricing rules, such as moderately increasing the price when user acceptance is high and promptly adjusting the price when user feedback is poor. Implement the dynamic pricing strategy in the differential price push engine to push the optimal price to users in real time. User A is pushed a price of $30 when acceptance is high and adjusted to $28 when acceptance is low. Record the real-time feedback and purchase situations of users, evaluate the effectiveness of the dynamic pricing strategy, and generate a dynamic pricing optimization report to describe the implementation effect of the pricing strategy and the future optimization direction.

[0037] Integrate the output of the personalized price strategy, the real-time price index, and the three-dimensional visual expression of the commodity to form the final user-side visual experience. The generation of the price index combines the user segmentation results and current market dynamic factors to output a real-time dynamic price attractiveness index (range 0 - 1). This index is dynamically normalized and mapped to the three-dimensional model of the commodity through color, transparency, or texture. The three-dimensional shape analysis reuses the aforementioned point cloud modeling and reconstruction technology to display the detailed commodity structure using high-precision point clouds and surface mesh models. The VR display is developed using the Unity3D engine and accesses the terminal head-mounted display device (such as the Oculus Quest 3) through the WebXR or OpenXR interface. During the model rendering process, the price index generates visual effects on the model surface through dynamic Shaders. For example, hot price parts are highlighted in red, and cold commodity areas are hinted at in blue. In addition, the system supports users to interact with the commodity in the VR space. Clicking on any part of the commodity can view price-related recommended explanations (such as material impact, origin cost, competitor comparison, etc.). The experimental results show that users' perception ability of price fluctuations and commodity value in the VR environment has increased by more than 30%, and the willingness to click and purchase has increased by 22%.

[0038] In this embodiment, the specific steps for generating the real-time price index of the target commodity based on the personalized price push strategy and performing three-dimensional shape analysis and VR visualization display processing on the target commodity are as follows: Generate the real-time price index of the target commodity based on the personalized price push strategy; Obtain multi-angle display images of the target commodity; Mark the structural feature points on the multi-angle display images of the target commodity and extract multiple structural feature points; Perform three-dimensional shape analysis on multiple structural feature points to obtain the three-dimensional shape characteristics of the target commodity; Perform three-dimensional point cloud modeling on the three-dimensional shape characteristics of the target commodity to construct a three-dimensional point cloud structure model; Perform dynamic index mapping on the three-dimensional point cloud structure model according to the real-time price index of the target commodity and perform VR visualization display processing to construct a holographic VR model of the price and shape of the target commodity; Perform user visualization display based on the holographic VR model of the price and shape of the target commodity to execute the VR visualization task.

[0039] In this embodiment, by collecting the real-time price data of the target product on the e-commerce platform and combining multi-dimensional information such as the user's historical purchase behavior, browsing preferences, and consumption ability, a machine learning model is used to dynamically generate a personalized price index. The specific approach is as follows: First, a user profile is constructed. Using the collaborative filtering algorithm and the neural collaborative filtering (NCF) model in deep learning, the acceptance degree of users for different price ranges is predicted. In the experiment, the user data dimension is usually above 100 dimensions, including user browsing time, click-through rate, conversion rate, etc., and the training data volume exceeds 100,000 to ensure the generalization ability of the model. The price index calculation fuses factors such as real-time product price, historical price fluctuations, competitor prices, and promotion intensity through weighting, and uses weighted moving average (WMA) and time series prediction (such as LSTM network) for trend prediction. Specific experimental parameters are as follows: the WMA window length is set to 7 days, the number of LSTM network layers is 2, the number of hidden units is 64, and the number of training iterations is set to 50 rounds. The model outputs a price index ranging from 0 to 1, reflecting the competitiveness and attractiveness of the current product price. This index is dynamically updated, combined with user characteristics, to achieve personalized adjustment of price push and improve the user's purchase conversion rate. The acquisition of multi-angle images is the basis for subsequent three-dimensional shape analysis and point cloud modeling, and it is necessary to ensure the integrity and high quality of the images. The adopted approach is to use an automatic rotating shooting device or a multi-camera array to collect 360-degree continuous images of the product. In the experiment, usually 12 - 36 pictures are set to cover the full range of views of the product, and the image resolution is at least 1920×1080 to ensure clear details. The shooting environment needs to control the lighting uniformity, reduce shadows and reflections, and use a ring soft light and a diffuser for auxiliary lighting. To ensure image alignment and subsequent feature matching, camera calibration technology is used to obtain internal and external parameters to ensure the geometric consistency between images from different perspectives. The calibration error is controlled below the pixel level (for example, the reprojection error < 0.5 pixels). After image acquisition, through image denoising, color correction, and geometric correction processing, a high-quality image dataset suitable for structural feature point extraction is output. The multi-angle images lay the foundation for subsequent 3D reconstruction and are the key data source for realizing the three-dimensional shape recognition of products. The extraction of structural feature points uses key point detection algorithms in computer vision, mainly including SIFT (Scale-Invariant Feature Transform), SURF (Speeded-Up Robust Features), and popular deep learning methods in recent years such as the SuperPoint network. In the experiment, usually, thousands of key points are automatically detected in multi-view images through the SIFT algorithm, and further, the RANSAC algorithm is used to eliminate incorrect matches to ensure the stability and robustness of the feature points.

[0040] Key point detection is performed on each image to obtain local descriptors, and then a feature matching algorithm (such as the FLANN matcher) is used to achieve cross-view point correspondence. Typical experimental parameter settings: the SIFT detection threshold is 0.04, the number of key points is controlled between 1000 and 3000, and the matching error threshold is within 3 pixels. For the deep learning method SuperPoint, real commodity images and synthetic data are used for augmented training during training, achieving a matching accuracy of over 85%. Through the matching results of the structural feature points in the multi-view images, a multi-view stereo (MVS) algorithm is adopted to map the two-dimensional feature points to the three-dimensional space and restore the three-dimensional morphological features of the commodity. The specific approach is to combine the camera calibration data and the feature point matching results, and calculate the three-dimensional coordinates of each corresponding feature point through triangulation.

[0041] The selected MVS algorithms include Patch-based Multi-View Stereo (PMVS) and COLMAP, both of which support dense point cloud generation. Taking COLMAP as an example, input multi-view images and camera parameters, run feature matching, incremental camera pose estimation, and dense reconstruction, and finally generate tens of thousands of three-dimensional point coordinates. The reconstruction accuracy depends on the image resolution and the view baseline. In the experiment, the camera view spacing is generally controlled between 15° and 30°, ensuring that the depth recovery error is less than 2 mm. The three-dimensional morphological features not only include the point cloud coordinates but also the normal vectors and color information of the points, which are used to enhance the detail level of the morphological expression. These three-dimensional features lay a solid data foundation for subsequent point cloud modeling and morphological analysis. Filtering, resampling, and fusion processing are performed on the original point cloud to generate a high-quality three-dimensional structure model. The filtering methods include Statistical Outlier Removal (SOR) and Radius Outlier Removal (ROR). In the experiment, the number of neighborhood points for SOR is set to 50, and the standard deviation multiplier is set to 1.0 to ensure that noise points are effectively removed. Resampling uses a voxel grid filter, and the voxel size is set to 1 - 2 mm to balance detail and computational efficiency. Point cloud fusion uses a method based on Moving Least Squares (MLS) to smooth the point cloud surface and enhance the continuity and integrity of the model.

[0042] The point cloud can be converted into a mesh model (such as a triangular mesh) for better morphological analysis and rendering. Using the Poisson Surface Reconstruction algorithm, a smooth surface is generated by combining the normal information of the points. Finally, the three-dimensional point cloud structure model contains hundreds of thousands to millions of points, which can present the complex morphological features of the commodity in detail and meet the high-precision requirements of VR visualization. Based on the existing three-dimensional point cloud structure model, the real-time price index generated in the first step is mapped as a dynamic parameter into the attributes of the point cloud model. The specific method is to use the point cloud color mapping technology to map the price index onto the point cloud surface through color gradients (such as warm and cold color systems), so that the price information is directly reflected in the three-dimensional model visually. In the experiment, the price index is normalized to the range of 0-1 and mapped to a color gradient bar from blue (low price) to red (high price). The dynamic index mapping is updated based on real-time data. Through WebSocket or other real-time data push mechanisms, the color attributes of the model are adjusted in real time. In the VR environment, engines such as Unity3D or Unreal Engine are used to load the point cloud model, and custom shaders are used to achieve dynamic color changes, combined with animation of price index changes to enhance the user interaction experience.

[0043] In addition, to enhance the immersion, dynamic particle effects or light effects of price fluctuation trends can be added to the model. Combining the user's perspective and interaction feedback, a holographic fusion display of price and form is achieved. The entire process needs to ensure that the rendering frame rate is stable above 60FPS to ensure a smooth VR experience. This holographic VR model becomes an intuitive tool for users to understand the relationship between commodity value and form. Interactive VR visualization display for users. Using a head-mounted display (HMD) such as Oculus Quest 2, HTC Vive or Valve Index to achieve an immersive experience. Users can freely rotate and scale the three-dimensional commodity model through the handle, gesture recognition or voice interaction, and observe the morphological details of the commodity and the changes in the price index. To improve the user experience, a friendly UI interface is designed to display the historical trend of the price index, price composition and promotion information. Combining the 3D space interaction logic, operations such as "selecting a commodity part to view price influencing factors" are realized. In the experiment, the UI response time is controlled within 100 milliseconds, and the interaction delay is less than 20 milliseconds to ensure smooth and natural user operations. In addition, the system background integrates an AI recommendation module, which adjusts the commodity display order and price strategy in real time based on user behavior to achieve a personalized recommendation closed loop. In the VR display operation, multi-modal data fusion technology is used to integrate interaction means such as voice, gesture and gaze tracking to enrich the user interaction dimension.

[0044] In this embodiment, a user commodity price recommendation and VR visualization system based on AI analysis is provided, including: A price fluctuation analysis module that obtains the real-time price data stream of the target commodity based on an AI model, conducts time-series price fluctuation trend analysis, and constructs a time-frequency characteristic graph of price fluctuations; A user usage behavior module that collects user usage behavior logs from various platforms, conducts dynamic mining of user consumption trajectories, and evaluates user price acceptance based on the time-frequency characteristic graph of price fluctuations to construct a user price elasticity model; An optimal price adjustment module that identifies the evolution of multi-point sensitivity based on the user price elasticity model, then predicts the optimal price adjustment timing, and generates an optimized time-series price adjustment strategy; A churn probability calculation module that predicts user interest transfer based on the user usage behavior logs and calculates the time-series churn probability to construct a multi-point user churn probability curve; A constraint optimization solving module that dynamically constraint-optimizes and solves the multi-point user churn probability curve according to the optimized time-series price adjustment strategy and makes user personalized price push decisions to construct a personalized price push strategy; A VR visualization module that executes commodity price recommendation operations based on the personalized price push strategy; and conducts 3D form VR visualization display operations of the target commodity.

[0045] Therefore, from any perspective, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Accordingly, all changes falling within the meaning and scope of the equivalent elements of the application documents are intended to be embraced by the present invention.

[0046] As described above, these are only the specific implementation manners of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather will conform to the broadest scope consistent with the principles and novel features disclosed herein.

Claims

1. A user commodity price recommendation and VR visualization method based on AI analysis, characterized in that It includes the following steps: Step S1: Obtain the real-time price data stream of the target commodity based on the AI model, conduct time-series price fluctuation trend analysis, and construct a time-frequency characteristic map of price fluctuations; Step S2: Collect the user usage behavior logs of multiple platforms, conduct dynamic mining of user consumption trajectories, and evaluate the user price acceptance based on the time-frequency characteristic map of price fluctuations to construct a user price elasticity model; Step S3: Identify the evolution of multi-point sensitivity based on the user price elasticity model, then predict the optimal price adjustment time, and generate a time-series optimization strategy for price adjustment; Step S4: Predict the transfer of user interests based on the user usage behavior logs, calculate the time-series churn probability, and construct a multi-point user churn probability curve; Step S5: Dynamically constrain and optimize the multi-point user churn probability curve according to the time-series optimization strategy for price adjustment and make a decision on personalized price push for users to construct a personalized price push strategy; Step S6: Execute the commodity price recommendation operation based on the personalized price push strategy; And conduct a 3D form VR visualization display operation of the target commodity.

2. The method for recommending user commodity prices based on AI analysis and VR visualization according to claim 1, wherein, The specific steps of Step S1 are as follows: Conduct real-time price tracking of the target commodity on multiple platforms based on the AI model to construct a real-time price data stream of the target commodity; Conduct adaptive identification of outliers for the real-time price data stream of the target commodity, and conduct outlier removal processing to obtain an outlier-optimized price data stream; Conduct dynamic time window segmentation on the outlier-optimized price data stream to obtain a sequence of commodity price data streams with multiple time windows; Conduct time-series price fluctuation trend analysis on the sequence of commodity price data streams to obtain price fluctuation trend characteristics; Extract the periodic characteristics, mutation characteristics, and trend characteristics of the price fluctuation trend characteristics, and conduct multi-level decomposition to construct a time-frequency characteristic map of price fluctuations.

3. The method for recommending user commodity prices based on AI analysis and VR visualization according to claim 1, characterized in that, The specific steps of Step S2 are as follows: Obtain the user usage behavior logs of multiple platforms; extract the behavior data of users' browsing, clicking, favoriting, adding to the shopping cart, and purchasing according to the user usage behavior logs, and mark all the usage behavior particle sequences of users; Conduct dynamic mining of user consumption trajectories based on the usage behavior particle sequences to extract user consumption trajectories; Conduct multi-behavior inertia and momentum estimation on the user consumption trajectories to obtain the consumption inertia and momentum under the influence of different behavior particles; Conduct user value sensitivity analysis according to the consumption inertia and momentum to construct a user price sensitivity curve; Conduct user price acceptance evaluation according to the sensitivity curve and the time-frequency characteristic map of price fluctuations to construct a user price elasticity model.

4. The method for recommending user commodity prices based on AI analysis and VR visualization according to claim 3, wherein The specific steps of conducting user price acceptance evaluation according to the sensitivity curve and the time-frequency characteristic map of price fluctuations to construct a user price elasticity model are as follows: Conduct price fluctuation mutation identification on the time-frequency characteristic map of price fluctuations and mark the price fluctuation inflection points; Calculate the fluctuation time length and fluctuation gradient of the price fluctuation inflection points; Extract the initial occurrence timestamp of the price fluctuation inflection points; Conduct spatio-temporal matching mapping of the user price sensitivity curve based on the initial occurrence timestamp, and conduct sensitivity correlation response evolution according to the fluctuation time length and fluctuation gradient to construct a price fluctuation-sensitivity response correlation network; Based on the price fluctuation-sensitivity response correlation network, perform in-depth sensitive diffusion topology analysis to extract the user price sensitivity propagation path; Evaluate the user price acceptance according to the user price sensitivity propagation path, and construct a user price acceptance elasticity model.

5. The method for user commodity price recommendation and VR visualization based on AI analysis according to claim 1, wherein, The specific steps of step S3 are as follows: Based on the user usage behavior log, perform multi-dimensional clustering of user groups to obtain multiple user group clustering clusters; Perform dynamic time-series evolution on the user price elasticity model to extract the user price sensitivity critical point; Based on the user price sensitivity critical point, perform adaptive matching on multiple user group clustering clusters, and perform vectorization processing of price acceptance to obtain price preference characteristics of different user groups; Identify the multi-time point sensitivity evolution according to the price preference characteristics of different user groups to obtain the sensitivity change characteristics under the user price preference at different time points; Predict the optimal price adjustment timing based on the sensitivity change characteristics under the user price preference, and generate a time-series optimization strategy for price adjustment.

6. The method for recommending user commodity prices based on AI analysis and VR visualization according to claim 1, characterized in that The specific steps of step S4 are as follows: Extract the platform historical marketing intervention records and short-term hot spot propagation trend data based on the user usage behavior log; Perform commodity interest deviation perception on the platform historical marketing intervention records and short-term hot spot propagation trend data to obtain a potential interest deviation vector; Calculate the user expected price gap for the price fluctuation time-frequency characteristic diagram to obtain a user expected price gap tensor; Perform expected difference gradient analysis on the user expected price gap tensor to obtain the positive and negative gradient curves of the user expected difference; Predict the user interest transfer according to the potential interest deviation vector and the positive and negative gradient curves of the user expected difference to obtain an interest transfer risk prediction value; Calculate the time-series churn probability based on the interest transfer risk prediction value, and construct a multi-time point user churn probability curve.

7. The method for recommending user commodity prices based on AI analysis and VR visualization according to claim 1, characterized in that, The specific steps of step S5 are as follows: Define the market competition situation and enterprise profit rate, and perform multi-objective optimization design based on the multi-time point user churn probability curve to obtain a multi-objective price optimization function; Perform dynamic constraint optimization solution on the multi-objective price optimization function according to the time-series optimization strategy for price adjustment to obtain a price adjustment parameter set; Perform user acceptance evaluation simulation on the price adjustment parameter set to extract the optimal price adjustment plan; Make a user personalized price push decision on the optimal price adjustment plan, and construct a personalized price push strategy.

8. The user commodity price recommendation and VR visualization method based on AI analysis according to claim 1, characterized in that The specific steps of step S6 are as follows: Perform user group targeted price push based on the personalized price push strategy, construct a differential price push engine, and collect user price response feedback information; Calculate the price adjustment income index according to the user price response feedback information, and extract the price adjustment income index; Evaluate the price adjustment effect on the user price response feedback information to obtain a price adjustment effect evaluation report; Perform feedback optimization iteration analysis on the price adjustment effect evaluation report and the price adjustment income index, and construct an adaptive income adjustment-effect feedback model; Perform dynamic pricing matching optimization on the differential price push engine based on the adaptive income adjustment-effect feedback model, and construct an intelligent price matching optimization model to execute the commodity price recommendation operation; Generate a real-time price index for target products based on a personalized price push strategy, and perform three-dimensional shape analysis and VR visualization display processing based on the target products.

9. The method for recommending user commodity prices based on AI analysis and VR visualization according to claim 8, characterized in that The specific steps for generating a real-time price index for target products based on a personalized price push strategy and performing three-dimensional shape analysis and VR visualization display processing are as follows: Generate a real-time price index for target products based on a personalized price push strategy; Obtain multi-angle display images of the target products; Mark the structural feature points of the multi-angle display images of the target products and extract multiple structural feature points; Perform three-dimensional shape analysis on the multiple structural feature points to obtain the three-dimensional shape features of the target products; Perform three-dimensional point cloud modeling on the three-dimensional shape features of the target products to construct a three-dimensional point cloud structure model; Perform dynamic index mapping on the three-dimensional point cloud structure model according to the real-time price index of the target products and perform VR visualization display processing to construct a holographic VR model of the price and shape of the target products; Perform user visualization display based on the holographic VR model of the price and shape of the target products to execute the VR visualization operation.

10. A user commodity price recommendation and VR visualization system based on AI analysis, characterized in that, Used to execute the user product price recommendation and VR visualization method based on AI analysis as described in claim 1, including: A price fluctuation analysis module that obtains the real-time price data stream of the target products based on an AI model, performs time-series price fluctuation trend analysis, and constructs a price fluctuation time-frequency characteristic diagram; A user usage behavior module that collects the user usage behavior logs of each platform, performs dynamic mining of the user consumption trajectory, and evaluates the user price acceptance based on the price fluctuation time-frequency characteristic diagram to construct a user price elasticity model; An optimal price adjustment module that identifies the evolution of multi-time point sensitivity based on the user price elasticity model, then predicts the optimal price adjustment timing, and generates a time-series optimization strategy for price adjustment; A churn probability calculation module that predicts the user interest transfer based on the user usage behavior logs and calculates the time-series churn probability to construct a multi-time point user churn probability curve; A constraint optimization solving module that performs dynamic constraint optimization solving on the multi-time point user churn probability curve according to the time-series optimization strategy for price adjustment and makes a user personalized price push decision to construct a personalized price push strategy; A VR visualization module that executes the product price recommendation operation based on the personalized price push strategy; and performs the three-dimensional shape VR visualization display operation of the target products.

Images