Commodity intelligent pricing system and method based on big data

Abstract

The invention relates to the technical field of e-commerce and data mining, and discloses an intelligent commodity pricing system and method based on big data, and the system comprises a data collection module which is used for collecting multi-source data from cost data, competitive product data, user portrait data, market trend data and macroeconomic data; the data preprocessing module is connected with the data acquisition module and is used for cleaning and standardizing the multi-source data; the feature engineering module is connected with the data preprocessing module and used for extracting and quantizing feature vectors including cost features, competitive product features, user features, market features, macroscopic features and sales features from the processed data; and the hierarchical pricing calculation module is connected with the feature engineering module and is used for being based on the feature vectors. According to the method, the problems of static and lagging of traditional pricing are solved, dynamic response, accurate individuation and continuous self-optimization of prices are realized, and the profitability and market competitiveness of enterprises can be effectively improved.

Description

Technical Field

[0001] This invention relates to the fields of e-commerce and data mining technology, specifically to a smart pricing system and method for goods based on big data. Background Technology

[0002] Product pricing is one of the core business decisions that determines a company's profitability, market share, and brand status. With the rapid development of e-commerce and the arrival of the big data era, the dimensions and volume of data that companies can obtain have grown to an unprecedented level, covering multiple levels from production costs, supply chains, and market competition to user behavior and the macroeconomy.

[0003] However, current commodity pricing practices still have many shortcomings. Traditional pricing strategies, such as cost-plus pricing, mainly focus on the company's internal costs and the predetermined fixed profit margin, while ignoring dynamic external factors such as market supply and demand and competitive landscape. This makes their pricing strategies rigid and unable to respond quickly to and adapt to the ever-changing market environment. While competition-oriented pricing takes into account the behavior of competitors to some extent, it often puts companies in a passive following situation, which can easily lead to vicious price wars and fails to fully explore the differentiated value of the product for different consumer groups, thus losing potential profit margins. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a smart pricing system and method for commodities based on big data, which solves the problems of existing pricing strategies relying on human experience, resulting in subjective decision-making and slow response, and the static nature of the models failing to fully respond to dynamic market changes.

[0005] To achieve the above objectives, the first aspect of the present invention provides a big data-based intelligent pricing system for commodities, comprising: The data acquisition module is used to collect multi-source data from cost data, competitor data, user profile data, market trend data, and macroeconomic data. A data preprocessing module, connected to the data acquisition module, is used to clean and standardize the multi-source data. The feature engineering module, connected to the data preprocessing module, is used to extract and quantify feature vectors from the processed data, including cost features, competitor features, user features, market features, macro features, and sales features. A tiered pricing calculation module, connected to the feature engineering module, is used to calculate the price based on the feature vector using a three-tiered progressive pricing model, wherein the three-tiered progressive pricing model includes a basic pricing layer, a dynamic adjustment layer, and a personalized pricing layer. The price fusion calculation module, connected to the hierarchical pricing calculation module, is used to weight and fuse the prices output by the basic pricing layer, the dynamic adjustment layer, and the personalized pricing layer to generate the final price. The dynamic iterative optimization module, connected to the price fusion calculation module, is used to collect feedback data corresponding to the final pricing in real time, and iteratively update the model parameters in the three-layer progressive pricing model and the price fusion calculation module according to the preset optimization objective function.

[0006] In one embodiment, the macroeconomic feature extracted by the feature engineering module is the inflation adjustment coefficient. The user characteristics include the user's willingness to pay coefficient. and user loyalty .

[0007] In one embodiment, the base pricing layer is used to calculate the base price. The calculation process is based on the following formula: ; in, The unit total cost obtained from the aforementioned cost characteristics, To determine the target base profit margin based on the product category, This is the inflation adjustment coefficient derived from the aforementioned macroeconomic characteristics. This calculation method incorporates macroeconomic factors into the base pricing to offset the impact of inflation on costs and profit margins.

[0008] In one embodiment, the dynamic adjustment layer is used to construct the supply and demand coefficient. and competitor adjustment coefficient and the basic pricing Multiplying these two coefficients yields the dynamically adjusted pricing. The supply and demand coefficient The competitor adjustment coefficient is used to quantify the relationship between market demand and supply. It is used to adjust prices based on the deviation between the company's base price and the prices of competing products, as well as the stability of the competing product market.

[0009] In one embodiment, the personalized pricing layer is used to construct a user pricing coefficient. And generate personalized pricing based on this coefficient. The user pricing coefficient is calculated according to the following formula: ; in, The user's willingness to pay coefficient is obtained from the user characteristics. To obtain user loyalty from the user characteristics, Based on the coefficient, and As weighted coefficients corresponding to the user's willingness to pay coefficient and user loyalty, respectively, this calculation method quantifies the user's individual spending power and repurchase tendency, and applies them to price calculation.

[0010] In one embodiment, the price fusion calculation module calculates the base price using a weighted average method. Dynamically adjusted pricing and personalized pricing The final pricing is obtained by merging the components. .

[0011] In one embodiment, the dynamic iterative optimization module uses maximizing a profit function as the optimization objective function, the profit function being determined by the final pricing. Unit total cost And actual sales as feedback data collection It is confirmed that the dynamic iterative optimization module updates the price weights of each layer in the price fusion calculation module, as well as the coefficients within the dynamic adjustment layer and the personalized pricing layer, using the gradient descent method.

[0012] A second aspect of this invention provides a smart pricing method for goods based on big data, comprising the following steps: Collect multi-source data from cost data, competitor data, user profile data, market trend data, and macroeconomic data; The multi-source data is cleaned and standardized. The processed data is extracted and quantified to form a feature vector that includes cost characteristics, competitor characteristics, user characteristics, market characteristics, macro characteristics, and sales characteristics; Based on the feature vector, the price is calculated through a three-layer progressive pricing model, which includes a basic pricing layer, a dynamic adjustment layer, and a personalized pricing layer. The prices output by the basic pricing layer, the dynamic adjustment layer, and the personalized pricing layer are weighted and fused to generate the final price. The system collects feedback data corresponding to the final pricing in real time, and iteratively updates the model parameters in the three-layer progressive pricing model and price fusion calculation steps according to the preset optimization objective function.

[0013] This invention provides a smart pricing system and method for commodities based on big data. It has the following beneficial effects: 1. This invention integrates five types of data sources: cost data, competitor data, user profile data, market trend data, and macroeconomic data, and constructs feature vectors containing corresponding characteristics. This provides a comprehensive basis for pricing calculation. This solution solves the technical problem that existing technologies rely on only a single dimension of data, which leads to a disconnect between pricing and users' actual willingness to pay and the market environment. This enables the final pricing to match the ever-changing business environment.

[0014] 2. This invention sets up a dynamic iterative optimization module to collect feedback data under the final pricing in real time at a minute-level frequency, and updates the model parameters in the three-layer progressive pricing model and price fusion calculation at an hour-level frequency. This closed-loop feedback and high-frequency iteration mechanism solves the technical problem that existing technologies cannot respond to market changes in real time due to the lag in price adjustments, and shortens the delay time from market changes to price adjustments.

[0015] 3. This invention sets up a personalized pricing layer in a three-tiered progressive pricing model. Based on user willingness to pay coefficients and user loyalty extracted from user characteristics, a user pricing coefficient is constructed and a differentiated price is generated. This solution solves the technical defect of the existing technology that adopts uniform pricing and cannot meet the needs of different user groups, so that the pricing strategy can be adjusted for specific user groups. Attached Figure Description

[0016] Figure 1 This is a system architecture diagram of the present invention; Figure 2 This is a schematic diagram of the method flow of the present invention; Figure 3 This is a schematic diagram of the internal structure and data flow of the tiered pricing calculation module of the present invention; Figure 4 This is a schematic diagram of the dynamic iterative optimization closed loop of the present invention; Figure 5 This is a schematic diagram of the data acquisition module of the present invention interacting with multi-source data.

[0017] The module includes: 10. Data acquisition module; 20. Data preprocessing module; 30. Feature engineering module; 40. Tiered pricing calculation module; 41. Basic pricing layer; 42. Dynamic adjustment layer; 43. Personalized pricing layer; 50. Price fusion calculation module; and 60. Dynamic iterative optimization module. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Example

[0019] Please see the appendix Figure 1 -Appendix Figure 5 This invention provides a big data-based intelligent pricing system for goods, comprising: Data acquisition module 10 is used to collect multi-source data from cost data, competitor data, user profile data, market trend data and macroeconomic data; The function of the data acquisition module 10 is to provide comprehensive and timely raw data input for subsequent preprocessing, feature engineering and pricing calculation. This module periodically acquires the data required for pricing by establishing data links with multiple heterogeneous data sources.

[0020] In a specific embodiment, the cost data acquired by the data acquisition module 10 originates from the enterprise resource planning (ERP) system within the enterprise. The data acquisition module 10 uses the API interface provided by the ERP system to periodically capture structured data related to product costs on a daily basis. Specifically, the collected data fields include, but are not limited to: The raw material costs, unit production and manufacturing costs, and unit warehousing and logistics costs from the warehouse to the user for a single product constitute the cost benchmark for subsequent basic pricing calculations.

[0021] For competitor data, data acquisition module 10 uses a distributed web crawler system based on the Scrapy framework. This crawler system is configured to target one or more mainstream e-commerce platforms and runs at a high frequency of 5 minutes. The fields collected include: Real-time sales prices, promotional information, inventory status, and number of user reviews for all identified products. This high-frequency data collection ensures the system can promptly grasp dynamic changes in the competitive landscape.

[0022] For user profile data, the data acquisition module 10 interfaces in real time with the enterprise's internal Customer Relationship Management (CRM) system. By accessing the CRM database or calling its service interface, the data acquisition module 10 obtains various user behavior and attribute data organized around a unique user ID. The specific data fields collected include: The user's historical spending records, historical average order value, cumulative repeat purchases, most recent login time, and other user activity metrics, as well as user tags pre-generated by the CRM system.

[0023] For market trend data, the data acquisition module 10 obtains data from one or more third-party data service platforms via an API interface. The acquisition cycle is set to once a week, and the collected data mainly consists of quantitative indicators, such as: The industry demand index of the target product category, the seasonality coefficient reflecting the strength of demand in a specific period, and the search popularity index of keywords related to the product on major search engines are all used to measure the macro supply and demand situation of the market in the dynamic adjustment layer.

[0024] For macroeconomic data, the data acquisition module 10 obtains data by connecting to the API interface of the official data release agency. The acquisition cycle is set to once a month. The core data fields collected are the Consumer Price Index (CPI) and the Producer Price Index (PPI), as well as the officially published inflation rate. Among them, the CPI data is used to calculate the inflation adjustment coefficient to correct costs and pricing and cope with macroeconomic fluctuations.

[0025] The data preprocessing module 20 is connected to the data acquisition module 10 and is used to clean and standardize multi-source data. In this embodiment, when missing values ​​are detected in the input data, the data preprocessing module 20 first performs a missing value imputation operation. The data preprocessing module 20 adopts a classification imputation strategy, applying different imputation methods for different types of data. Specifically, when a field (e.g., raw material cost) in the cost data is missing, the module uses the time series mean imputation method, using the arithmetic mean of the field over the past 7 collection periods to imput it. When competitor price data is missing, the module first identifies other competitors that belong to the same category as the product and have similar functions and specifications, then calculates the average price of these similar competitors in the current time window, and uses the average to imput the missing position. When a numerical field in the user profile data is missing, the module classifies the user into a specific user group based on the user tag, and calculates the median of all other users in the group for that field to imput the missing value.

[0026] After handling missing values, the data preprocessing module 20 performs outlier detection and processing on the numerical features in the dataset. This module 20 first calculates the mean of each numerical feature across the entire dataset. and standard deviation Then, the module iterates through each data point under this feature. If the value of a data point deviates from its mean by more than three standard deviations, that is, its value does not fall within the interval... If the data point is found to be out of range, it will be removed from the dataset to prevent extreme data from interfering with subsequent model calculations.

[0027] To eliminate the impact of differences in units and numerical ranges between different features on model calculations, the data preprocessing module 20 performs data standardization after completing the above steps. In one embodiment, this module uses the min-max standardization method to linearly map all numerical features involved in model calculations to an interval. The transformation process is performed according to the following formula: ; in, The original values ​​of the features. and These are the minimum and maximum values ​​of this feature across all values ​​in the current dataset, respectively. The result values ​​are obtained after standardization. After this step, all numerical features are normalized to form the final dataset and output to the feature engineering module 30.

[0028] The feature engineering module 30, connected to the data preprocessing module 20, is used to extract and quantify feature vectors from the processed data, including cost features, competitor features, user features, market features, macro features, and sales features. In this embodiment, the feature engineering module 30 constructs six types of core features: Cost characteristics: This module directly extracts the total unit cost from the preprocessed cost data. Its value is the sum of raw material costs, production and manufacturing costs, and warehousing and logistics costs. This feature is the benchmark input for the calculation of the basic pricing layer.

[0029] Competitor Features: This module calculates two key features based on preprocessed competitor data. The first is the average price of competitors. The first is the arithmetic mean of the sales prices of all monitored competitors within a specific time window; the second is the competitor price volatility coefficient. This is used to quantify the stability of competitor prices, and its calculation process is based on the following formula: ; in, This represents the maximum price of all competing products within that time window. This is the minimum price of all competing products. The larger the value of this coefficient, the more volatile the market prices of competing products.

[0030] User Features: This module constructs two core user features based on preprocessed user profile data. The first is the user's willingness to pay coefficient. This is used to quantify the relative spending power of a single user, and its calculation process is based on the following formula: ; in, This is the user's historical average order value. This represents the market average order value for the product category. A coefficient greater than 1 indicates that the user's purchasing power is higher than the category average. The second factor is user loyalty. This is used to measure a user's repurchase tendency, and its calculation process is based on the following formula: ; in, This represents the user's cumulative number of repeat purchases. This represents the user's total number of purchases. The closer this coefficient is to 1, the higher the user's loyalty.

[0031] Market Characteristics: This module extracts two quantitative indicators from the preprocessed market trend data. The first is the demand index. The first value is a standardized value between 0 and 1, directly sourced from a third-party data platform. A higher value indicates stronger market demand. The second value is a seasonality coefficient. Similarly, it is a standardized value between 0 and 1, reflecting the seasonal fluctuations in commodity demand. The higher the value, the more it indicates that the current demand is in its peak season.

[0032] Macroeconomic characteristics: This module calculates the inflation adjustment coefficient based on preprocessed macroeconomic data. This coefficient is used in pricing models to hedge against the impact of inflation on costs and pricing, and its calculation is based on the following formula: ; in, The consumer price index for the current month is obtained by the data acquisition module 10. This refers to the Consumer Price Index for the same month last year.

[0033] Sales Characteristics: This module calculates two reference characteristics based on the company's own historical sales data. The first is historical sales volume. The first is the average daily sales volume of the product over the past 30 days, and the second is the historical profit margin. The calculation process is based on the following formula: ; in, This is the historical average selling price of the product over the past sales cycle.

[0034] In addition, for non-numerical features obtained from user profile data, the feature engineering module 30 performs feature quantization operations. In one embodiment, the module uses the one-hot encoding method to convert each label into a multi-dimensional binary vector, so that these discrete features can be processed by the subsequent mathematical model. All the constructed features are combined into a unified feature vector and transmitted to the hierarchical pricing calculation module 40.

[0035] The tiered pricing calculation module 40, connected to the feature engineering module 30, is used to calculate the price based on the feature vector through a three-tiered progressive pricing model, which includes a basic pricing layer, a dynamic adjustment layer, and a personalized pricing layer. In this embodiment, the tiered pricing calculation module 40 integrates three functionally independent calculation layers: a basic pricing layer 41, a dynamic adjustment layer 42, and a personalized pricing layer 43.

[0036] The function of the base pricing layer 41 is to set a benchmark price that guarantees a basic profit and mitigates macroeconomic risks. This layer receives the unit total cost provided by the feature engineering module 30. and inflation adjustment coefficient As input, base pricing The calculation process is performed according to the following formula: ; in, Total cost per unit; The target base profit margin is a pre-set parameter based on the product category; This is an inflation adjustment factor. This calculation method ensures that the base price covers all costs while dynamically adjusting the expected profit margin according to the inflation level. The calculation results... It is output to the dynamic adjustment layer 42 and the price fusion calculation module 50.

[0037] The function of the dynamic adjustment layer 42 is to enable prices to respond to changes in market supply and demand and competitive landscape. This layer receives the base price output from the base pricing layer 41. and obtain from feature engineering module 30 The calculation process is based on the following formula: ; in, This is the demand index. This is a seasonal coefficient. Based on historical sales figures, The first is an adjustable supply scaling factor used to calibrate the relationship between historical supply and current market conditions; the second is a competitor adjustment coefficient. The calculation process is based on the following formula: ; in, The average price of competing products, This is the competitor's price volatility coefficient; subsequently, this layer will set the base price. Multiplying this by the two adjustment coefficients yields the dynamically adjusted pricing. The calculation process is as follows: The result It is output to the personalized pricing layer 43 and the price fusion calculation module 50.

[0038] The function of the personalized pricing layer 43 is to implement differentiated pricing based on individual user characteristics. This layer receives the dynamically adjusted pricing output by the dynamic adjustment layer 42. It then obtains user features from feature engineering module 30. This layer first constructs a user pricing coefficient. The calculation process is based on the following linear regression formula: ; in, The user's willingness to pay coefficient For user loyalty; Based on the coefficient, and These three coefficients, corresponding to the weight coefficients of the two user features respectively, are model parameters that can be updated in the subsequent dynamic iterative optimization module 60. This layer will then dynamically adjust the post-pricing. With user pricing coefficient Multiply by these to obtain the final personalized price. The calculation process is as follows: The result It is output to the price fusion calculation module 50.

[0039] The price fusion calculation module 50 is connected to the tiered pricing calculation module 40 and is used to perform weighted fusion of the prices output by the basic pricing layer, the dynamic adjustment layer and the personalized pricing layer to generate the final price. In this embodiment, the price fusion calculation module 50 uses a weighted average method to fuse the input prices. This module receives the base price from the base pricing layer 41. Dynamically adjusted pricing from dynamic adjustment layer 42 And personalized pricing from personalized pricing tier 43 And calculate the final price according to the following formula. : ; in, , and These are the weighting coefficients corresponding to the three input prices, and the sum of these three weighting coefficients is always 1. .

[0040] The weighting coefficients are not fixed values, but rather reflect the focus of the current pricing strategy. During system initialization, these weights can be assigned initial values ​​based on preset business objectives. For example, during the new product launch phase, to prioritize market competitiveness and cost recovery, the initial weights can be set to... For products that have entered the mature stage and accumulated a large amount of user data, the weight can be adjusted to maximize the value of individual users. These weighting coefficients are one of the core optimization objects of the subsequent dynamic iterative optimization module 60.

[0041] After calculating the final price Then, the price fusion calculation module 50 performs a boundary check. This module will calculate... Compared with the unit total cost obtained from feature engineering module 30 A comparison is made to ensure that the final output price is not lower than the cost. Specifically, the final output price is... After completing the boundary check, the module outputs the final price to the external business system (such as the product price interface of an e-commerce platform) for display and sales. At the same time, it transmits the final price and its corresponding user ID and other information to the dynamic iterative optimization module 60 as the input basis for subsequent optimization.

[0042] The dynamic iterative optimization module 60 is connected to the price fusion calculation module 50. It is used to collect feedback data corresponding to the final pricing in real time, and iteratively update the three-layer progressive pricing model and the model parameters in the price fusion calculation module 50 according to the preset optimization objective function.

[0043] In this embodiment, the module first performs feedback data collection. The module initiates query requests to the business database at a preset high-frequency period to obtain new sales orders generated within that period. For each order, the module extracts its actual transaction price and actual purchase quantity. The module also receives and temporarily stores the generated final price from the price fusion calculation module 50, along with the corresponding user ID. The module, by matching the user ID and transaction time in the order with the temporarily stored pricing records, can accurately associate each actual sales volume with the specific pricing decision that generated that sales volume.

[0044] After collecting and correlating the feedback data, the dynamic iterative optimization module 60 initiates a model parameter update process at a preset update cycle. The goal of this process is to maximize a preset profit function. In this embodiment, this profit function... Defined as the total profit accumulated within a single update cycle, its calculation is based on the following formula: ; in, This represents the total number of orders placed during that period. For the first The actual sales volume of this order This is the final price corresponding to this order. This represents the unit total cost of the product. This represents the set of all model parameters to be optimized.

[0045] To maximize profits, this module uses gradient descent to iteratively update the model parameters, which are then optimized. Including the weighting coefficients in the price fusion calculation module 50 And some coefficients within the tiered pricing calculation module 40, such as the weighting coefficients in the personalized pricing layer 43. .

[0046] In each update cycle, the module first calculates the profit function. The partial derivatives with respect to each parameter to be optimized are used to obtain the gradient of the profit function in that parameter space. This gradient indicates the direction in which the parameters should be adjusted to maximize profit growth. Subsequently, the module adjusts each parameter according to the following general update rules: ; in, This is the value of the parameter before the update. This is the updated value. It is a preset hyperparameter called the learning rate, which is used to control the step size of each update.

[0047] After completing the update calculations for all parameters, the dynamic iterative optimization module 60 will update the newly generated parameters. The updated weighting coefficients are pushed to the corresponding modules. The updated regression coefficients were sent to the price fusion calculation module 50. The personalized pricing layer 43 is sent to the tiered pricing calculation module 40. After receiving the new parameters, these modules immediately use them in the next round of pricing calculation, thereby applying the optimization results to the business in real time and forming a complete and automated closed-loop optimization process.

[0048] Reference Figure 2 According to an embodiment of the present invention, a big data-based intelligent pricing method for goods includes the following steps: S201: Collect data from multiple sources.

[0049] S202: Preprocess the collected data.

[0050] S203: Extract and construct feature vectors from the preprocessed data.

[0051] S204: Perform tiered pricing calculation based on feature vectors.

[0052] S205: Merge the prices calculated at each level.

[0053] S206: Perform dynamic iterative optimization based on feedback data.

Claims

1. A big data-based intelligent pricing system for commodities, characterized in that: include: The data acquisition module is used to collect data from multiple sources. A data preprocessing module, connected to the data acquisition module, is used to clean and standardize the multi-source data. The feature engineering module, connected to the data preprocessing module, is used to extract and quantize feature vectors from the processed data. A tiered pricing calculation module, connected to the feature engineering module, is used to calculate the commodity price based on the feature vector using a three-tiered progressive pricing model. The price fusion calculation module, connected to the hierarchical pricing calculation module, is used to perform weighted fusion of the prices output by the three-level progressive model to generate the final price.

2. The intelligent commodity pricing system based on big data according to claim 1, characterized in that, The data acquisition module is specifically used for: The cost data is collected through the enterprise ERP system API interface; The competitor data was collected using distributed web crawlers; Collect the user profile data from the user management system; The market trend data is collected through a third-party data platform.

3. The intelligent commodity pricing system based on big data according to claim 1, characterized in that, The user features extracted by the feature engineering module include the user's willingness to pay coefficient. and user loyalty The competitive product characteristics include the average price of competing products. Price volatility coefficient of competing products The macroeconomic characteristic mentioned is the inflation adjustment coefficient. .

4. The intelligent commodity pricing system based on big data according to claim 1, characterized in that, The three-tiered progressive pricing model includes a base pricing layer, a dynamic adjustment layer, and a personalized pricing layer. The base pricing layer is used to calculate the base price. The calculation formula is as follows: ； in, The unit total cost obtained from the aforementioned cost characteristics, The target base profit margin is set at 0. The inflation adjustment coefficient is obtained from the aforementioned macroeconomic characteristics.

5. The intelligent commodity pricing system based on big data according to claim 4, characterized in that, The dynamic adjustment layer is used to construct supply and demand coefficients. and competitor adjustment coefficient The output of the basic pricing layer is multiplied by the supply-demand coefficient and the competitor adjustment coefficient to obtain the dynamically adjusted pricing. .

6. The intelligent commodity pricing system based on big data according to claim 4, characterized in that, The personalized pricing layer is used to construct user pricing coefficients. And generate personalized pricing based on this coefficient. The formula for calculating the user pricing coefficient is as follows: ； in, The user's willingness to pay coefficient is obtained from the user characteristics. To obtain user loyalty from the user characteristics, Based on the coefficient, and These are the weighting coefficients corresponding to the user's willingness to pay coefficient and user loyalty, respectively.

7. The intelligent commodity pricing system based on big data according to claim 1, characterized in that, The price fusion calculation module specifically fuses the prices output by the basic pricing layer, dynamic adjustment layer, and personalized pricing layer using a weighted average method. The system also includes: The dynamic iterative optimization module, connected to the price fusion calculation module, is used to collect feedback data corresponding to the final pricing in real time, and iteratively update the model parameters in the three-layer progressive pricing model and the price fusion calculation module according to the preset optimization objective function.

8. The intelligent commodity pricing system based on big data according to claim 1, characterized in that, The objective function of the dynamic iterative optimization module is to maximize profit, and the feedback data is the actual sales volume and actual profit generated under the final pricing.

9. The intelligent commodity pricing system based on big data according to claim 8, characterized in that, The model parameters updated by the dynamic iterative optimization module include: the price weights of each layer in the price fusion calculation module, and the coefficients within the dynamic adjustment layer and the personalized pricing layer.

10. A big data-based intelligent pricing method for commodities, and a big data-based intelligent pricing system for commodities according to any one of claims 1-9, characterized in that, Includes the following steps: Collect data from multiple sources; The multi-source data is cleaned and standardized. The processed data is extracted and quantified to form a feature vector that includes cost characteristics, competitor characteristics, user characteristics, market characteristics, macro characteristics, and sales characteristics; Based on the aforementioned feature vector, the commodity price is calculated using a three-layer progressive pricing model. The prices output by the basic pricing layer, the dynamic adjustment layer, and the personalized pricing layer are weighted and fused to generate the final price. The system collects feedback data corresponding to the final pricing in real time, and iteratively updates the model parameters in the three-layer progressive pricing model and price fusion calculation steps according to the preset optimization objective function.

Images