Product pricing method, system and device based on data analysis

Abstract

The invention discloses a product pricing method, system and device based on data analysis, and the method comprises the steps: obtaining a product related data set, and obtaining a product feature data set through analysis and feature extraction; a comprehensive feature data set is obtained through the comprehensive feature data model and the product feature data set, and comprehensive feature data comprises sales comprehensive feature data, price comprehensive feature data, customer comprehensive feature data and inventory comprehensive feature data; establishing a plurality of product pricing pre-training models, and analyzing the product feature data set and the comprehensive feature data set to obtain product pricing prediction data; obtaining product pricing data through model training and model reasoning; and comprehensively analyzing the product pricing data to obtain a product pricing result. According to the invention, through comprehensive analysis of product related data, dynamic pricing of products of different time, markets, clients and categories is realized, the products are accurately priced in real time, and market demands are met.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a product pricing method, system and device based on data analysis. Background Art

[0002] Product pricing and product price adjustment are key links in product sales, directly affecting the market performance of products and the profitability of enterprises. With the development of product diversification, there are significant differences in the market demand and degree of competition for different types of products. At the same time, as the market is more sensitive to changes in product prices, product prices are becoming more transparent, and customers have a wider range of choices for product prices. Different regions and seasons will cause fluctuations in product demand, and product prices and product inventory will affect each other. Therefore, if you cannot analyze multiple product data and complete product pricing based on the analysis results, it will affect the product sales data and thus affect the company's profits.

[0003] Existing product pricing methods are usually based on experience or simple static analysis. Such pricing methods use inherent pricing formulas to add procurement costs to gross profit margins for pricing, ignoring dynamic factors such as market competition and product inventory pressure, and not analyzing the price acceptance of different customer groups, which may lead to the loss of high-value customers. In addition, existing product pricing methods cannot quickly capture changes in market conditions, such as price fluctuations of competitors or changes in demand from downstream customers, causing product price adjustments to lag behind market changes. In addition, most existing product pricing methods are based on single-dimensional data for pricing, without comprehensive consideration of multiple factors such as market, customers, regions and demand, and cannot accurately reflect market and customer demand, leading to profit and customer loss. Summary of the invention

[0004] In view of the shortcomings of the prior art, the present invention provides a product pricing method, system and device based on data analysis.

[0005] In order to solve the above technical problems, the present invention is solved by the following technical solutions: A product pricing method based on data analysis includes the following steps: Obtain product-related data sets and perform analysis and feature extraction to obtain product feature data sets, wherein product-related data include sales data, inventory data, procurement data and environmental data, and product feature data include sales feature data, inventory feature data, environmental order data, order feature data, category feature data and time feature data; Constructing a comprehensive feature data model, and combining it with the product feature data set to obtain a comprehensive feature data set, wherein the comprehensive feature data model includes a sales comprehensive data model, a price comprehensive data model, a customer comprehensive data model, and an inventory comprehensive data model, and the comprehensive feature data includes sales comprehensive feature data, price comprehensive feature data, customer comprehensive feature data, and inventory comprehensive feature data; Constructing several product pricing pre-training models, each of which includes a feature receiving layer, a feature processing layer and a result output layer, wherein the feature receiving layer receives product feature data and comprehensive feature data, the feature processing layer performs corresponding analysis on the product feature data and comprehensive feature data, and the result output layer obtains product pricing prediction data based on the analysis results; Based on the product feature data set and the comprehensive feature data set, each product pricing pre-training model is trained to obtain a corresponding product pricing model, and then obtain corresponding product pricing data; Conduct a comprehensive analysis of product pricing data to obtain product pricing results.

[0006] As an implementable embodiment, it also includes: Analyze sales data to obtain product price data and product sales data; Analyze the inventory data to obtain the current product inventory data; Analyze sales data and purchasing data to obtain product demand forecast data and product replenishment cycle; Obtain competitive price data and customer order data through sales data, and obtain industry trend data and seasonal impact data based on sales data and inventory data.

[0007] As an implementable method, the sales characteristic data is obtained by the following steps: Based on the product price data and product sales data, the average product price data and total product sales data within the preset time period are obtained, as shown below:

[0008]

[0009] By analyzing the fluctuations of product price data and product average price data, we can obtain price fluctuation data, which is shown as follows:

[0010] Preset a time threshold, analyze the change trend of product sales data within the time threshold, and obtain sales growth data, which is expressed as follows:

[0011] Analyze the impact of product price data on product sales data and obtain demand elasticity data, which is expressed as follows:

[0012] Sales characteristic data is formed through product average price data, product total sales data, price fluctuation data, sales growth data and demand elasticity data; in, Indicates the average price data of the product within the preset time period. Indicates product price data within a preset time period. Represents product sales data within a time period. Represents the total sales data of the product. Represents product sales data, Represents price fluctuation data, represents sample data, Indicates sales growth data, Indicates the product sales data within the current preset time threshold. Indicates the product sales data within the previous preset time threshold. represents the demand elasticity data, represents the percentage change in sales volume, Indicates the percentage change in price.

[0013] As an implementable method, the inventory characteristic data is obtained by the following steps: Based on the analysis of product sales data and current product inventory data, the inventory turnover rate data is obtained, which is expressed as follows:

[0014] Based on product replenishment cycle and product demand forecast data, analyze product inventory and product replenishment status to obtain safety inventory data; The average number of days from product purchase to sale is analyzed through inventory turnover rate data to obtain inventory turnover data, which is expressed as follows:

[0015] Through inventory turnover rate data, safety stock data and inventory turnover data, inventory characteristic data is formed; in, Represents inventory turnover data, Represents product sales data, Indicates the current product inventory data. Represents inventory turnover data.

[0016] As an implementable method, the environmental characteristic data is obtained by the following steps: Analyze the market demand of products through industry trend data and seasonal impact data to obtain market demand data; Obtain weather data, and analyze the impact of weather factors on product demand based on the weather data and product demand forecast data to obtain weather factor data; During holidays, product price data is adjusted, and holiday effect data is obtained based on the adjustment method and the adjusted product price data; Environmental characteristic data are formed through market demand data, weather factor data and holiday effect data.

[0017] As an implementable method, the order characteristic data is obtained by following the steps below: Analyze customer order data and product price data to obtain customer price sensitivity data, and divide different customer groups. Analyze customer types and customer price sensitivity data to obtain customer type data; Preset an order period threshold, obtain the customer's purchase frequency within the order period threshold through the customer order data, and obtain order frequency data; Obtain and analyze historical customer order data to obtain the customer's expected value, and then obtain customer life cycle data through the customer's expected value; Order characteristic data is formed through customer type data, order frequency data and customer life cycle data.

[0018] As an implementable method, the category characteristic data is obtained by analyzing the influence of product type, product size, product grade and product life cycle on product price data, wherein the category characteristic data includes product type data, product size data, quality grade data and product life data; The time characteristic data is obtained by performing seasonal change analysis and cyclical change analysis on product price data.

[0019] As an implementable method, the comprehensive feature data model is constructed and combined with the product feature data set to obtain the comprehensive feature data set, including the following steps: A comprehensive sales data model is constructed to analyze the total product sales data, price fluctuation data, demand elasticity data, and current product inventory data to obtain comprehensive sales feature data, which is expressed as follows:

[0020] A comprehensive price data model is constructed to analyze demand elasticity data, inventory turnover data, competitive price data, current product price data and market demand data to obtain comprehensive price feature data, which is expressed as follows:

[0021] Build a comprehensive customer data model, obtain single order amount data and customer loyalty data through customer order data, and analyze it in combination with order frequency data and customer life cycle data to obtain comprehensive customer feature data, which is expressed as follows:

[0022] Construct a comprehensive inventory data model, analyze the current product inventory data, market demand data and inventory turnover rate data, and obtain comprehensive inventory feature data, which is expressed as follows:

[0023] A comprehensive feature data set is formed by using sales comprehensive feature data, price comprehensive feature data, customer comprehensive feature data and inventory comprehensive feature data; in, Represents sales comprehensive characteristic data, Represents the total sales data of the product. Represents price fluctuation data, represents the demand elasticity data, Represents product inventory data, , , , represents the product feature weight, Represents comprehensive price feature data, Represents inventory turnover data, represents competitive price data, Represents the current product price data, Represents market demand data, , , , represents the price feature weight, Represents comprehensive customer feature data, Represents customer lifecycle data, represents order frequency data, Indicates the single order amount data. represents customer loyalty data, , , , represents the customer feature weight, Represents comprehensive inventory feature data, Indicates the current product inventory data. , Represents the inventory feature weight.

[0024] As an implementable embodiment, the feature receiving layer receives product feature data and comprehensive feature data; The corresponding feature processing layer analyzes the importance of product feature data and comprehensive feature data, and obtains feature weights based on the importance analysis results; The result output layer analyzes the feature weights, product feature data, and comprehensive feature data to obtain product pricing prediction data, which is expressed as follows:

[0025] The product loss function and the comprehensive loss function are constructed through the product pricing prediction data, feature weights and real product pricing data, and the pricing linear loss function is formed through the product loss function and the comprehensive loss function. The product loss function and the comprehensive loss function are expressed as follows:

[0026]

[0027] The pricing linear loss function is gradient-solved, and based on the gradient solution result, the feature weights of the product pricing pre-training model are updated using the gradient descent method to obtain the product pricing model. The gradient solution and weight update are expressed as follows:

[0028] The product pricing model is used to analyze and infer the product feature data and comprehensive feature data to be predicted, and obtain product pricing data; in, represents the product loss function, represents the comprehensive loss function, represents the pricing linear loss function, represents the number of samples, Indicates The real pricing data of samples, represents product pricing forecast data, Represents the importance parameter of comprehensive feature data, Indicates the quantity of product feature data, represents the number of comprehensive feature data, Indicates The feature weight of each product feature data, Indicates The feature weight of the comprehensive feature data, represents the weight adjustment factor, Indicates bias, Indicates The sample Product feature data, Indicates The sample Comprehensive feature data, represents the weight update, Indicates the weight of product feature data or the weight of comprehensive feature data, represents the learning rate, Indicates the dynamic adjustment coefficient.

[0029] As an implementable embodiment, the feature receiving layer receives product feature data and comprehensive feature data; The corresponding feature processing layer recursively splits the product feature data and the comprehensive feature data, analyzes the information changes of the product feature data and the comprehensive feature data before and after the split, obtains the information gain, and obtains the split node for the next split based on the information gain, where the information gain is expressed as follows:

[0030] The category probabilities of the product feature data and comprehensive feature data after splitting are obtained, and the feature Gini index is obtained based on the category probability. The recursive splitting is fed back through the feature Gini index until the optimal splitting point is obtained. The feature Gini index is expressed as follows:

[0031] The output layer analyzes the product feature data and comprehensive feature data through the optimal split point to obtain product pricing data; in, represents information gain, represents the information entropy of the split node of the last split, represents the Gini index, Indicates that product feature data or comprehensive feature data belongs to the category The probability of Represents a collection of product feature data and comprehensive feature data, Indicates the first subsets.

[0032] As an implementable embodiment, the feature receiving layer receives product feature data and comprehensive feature data; The corresponding feature processing layer constructs a pricing learner, analyzes and predicts product feature data and comprehensive feature data based on the pricing learner, and obtains pricing residuals based on the product pricing prediction data of the previous iteration. It iterates and constructs a new pricing learner through pricing residuals and learning rates; The output layer obtains the current product pricing prediction data through the new pricing learner, learning rate and the product pricing prediction data of the previous iteration, which is expressed as follows:

[0033] Through product pricing prediction data, real pricing data and regularization terms, a pricing regression loss function is constructed, which is expressed as follows:

[0034] The feature importance of product feature data and comprehensive feature data is analyzed through pricing residuals to obtain feature gain. The product pricing pre-training model is trained through feature gain and pricing regression loss function to obtain the product pricing model. Based on the product pricing model, the product feature data and comprehensive feature data to be predicted are analyzed to obtain product pricing data; in, represents the pricing regression loss function, Indicates Product pricing forecast data, represents the regularization term, represents the prediction of the current tree, represents the number of trees in the model, represents the number of samples, represents the learning rate, represents product pricing forecast data, express Keshu’s product pricing prediction data.

[0035] As an implementable method, the comprehensive analysis of product pricing data to obtain product pricing results includes the following steps: A comprehensive pricing pre-training model is constructed. The product pricing data is linearly combined through the comprehensive pricing pre-training model and combined with the real product pricing data and feature coefficient vector to construct a comprehensive pricing loss function, which is expressed as follows:

[0036] The feature coefficient vector is updated through the comprehensive pricing loss function to obtain the optimal feature coefficient vector, and then the comprehensive pricing model is obtained. The update process is expressed as follows:

[0037]

[0038] The product pricing data corresponding to several product pricing models are analyzed and inferred through the comprehensive pricing model to obtain the product pricing results, which are expressed as follows:

[0039] in, represents the comprehensive pricing loss function, represents real product pricing data, Indicates product characteristic data or comprehensive characteristic data, Indicates The characteristic coefficient vector, represents the number of characteristic coefficient vectors, represents the regularization parameter, represents the number of samples, represents the number of iterations, Indicates the product characteristic data or the first of the comprehensive characteristic data data, Represents the product pricing result, represents the comprehensive pricing model, Indicates Product pricing data, represents the number of base learners, Indicates product characteristic data or comprehensive characteristic data.

[0040] A product pricing system based on data analysis, including a data acquisition module, a feature extraction module, a model building module, a model training module and a comprehensive prediction module; The data acquisition module acquires product-related data sets and performs analysis and feature extraction to obtain product feature data sets, wherein the product-related data includes sales data, inventory data, procurement data and environmental data, and the product feature data includes sales feature data, inventory feature data, environmental order data, order feature data, category feature data and time feature data; The feature extraction module constructs a comprehensive feature data model and obtains a comprehensive feature data set by combining the product feature data set, wherein the comprehensive feature data model includes a sales comprehensive data model, a price comprehensive data model, a customer comprehensive data model and an inventory comprehensive data model, and the comprehensive feature data includes sales comprehensive feature data, price comprehensive feature data, customer comprehensive feature data and inventory comprehensive feature data; The model building module builds a plurality of product pricing pre-training models, each of which includes a feature receiving layer, a feature processing layer and a result output layer. The feature receiving layer receives product feature data and comprehensive feature data, the feature processing layer performs corresponding analysis on the product feature data and comprehensive feature data, and the result output layer obtains product pricing prediction data based on the analysis results. The model training module trains each product pricing pre-training model based on the product feature data set and the comprehensive feature data set to obtain a corresponding product pricing model, and then obtain corresponding product pricing data; The comprehensive prediction module performs comprehensive analysis on product pricing data to obtain product pricing results.

[0041] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following method is implemented: Obtain product-related data sets and perform analysis and feature extraction to obtain product feature data sets, wherein product-related data include sales data, inventory data, procurement data and environmental data, and product feature data include sales feature data, inventory feature data, environmental order data, order feature data, category feature data and time feature data; Constructing a comprehensive feature data model, and combining it with the product feature data set to obtain a comprehensive feature data set, wherein the comprehensive feature data model includes a sales comprehensive data model, a price comprehensive data model, a customer comprehensive data model, and an inventory comprehensive data model, and the comprehensive feature data includes sales comprehensive feature data, price comprehensive feature data, customer comprehensive feature data, and inventory comprehensive feature data; Constructing several product pricing pre-training models, each of which includes a feature receiving layer, a feature processing layer and a result output layer, wherein the feature receiving layer receives product feature data and comprehensive feature data, the feature processing layer performs corresponding analysis on the product feature data and comprehensive feature data, and the result output layer obtains product pricing prediction data based on the analysis results; Based on the product feature data set and the comprehensive feature data set, each product pricing pre-training model is trained to obtain a corresponding product pricing model, and then obtain corresponding product pricing data; Conduct a comprehensive analysis of product pricing data to obtain product pricing results.

[0042] A product pricing device based on data analysis includes a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the following method when executing the computer program: Obtain product-related data sets and perform analysis and feature extraction to obtain product feature data sets, wherein product-related data include sales data, inventory data, procurement data and environmental data, and product feature data include sales feature data, inventory feature data, environmental order data, order feature data, category feature data and time feature data; Constructing a comprehensive feature data model, and combining it with the product feature data set to obtain a comprehensive feature data set, wherein the comprehensive feature data model includes a sales comprehensive data model, a price comprehensive data model, a customer comprehensive data model, and an inventory comprehensive data model, and the comprehensive feature data includes sales comprehensive feature data, price comprehensive feature data, customer comprehensive feature data, and inventory comprehensive feature data; Constructing several product pricing pre-training models, each of which includes a feature receiving layer, a feature processing layer and a result output layer, wherein the feature receiving layer receives product feature data and comprehensive feature data, the feature processing layer performs corresponding analysis on the product feature data and comprehensive feature data, and the result output layer obtains product pricing prediction data based on the analysis results; Based on the product feature data set and the comprehensive feature data set, each product pricing pre-training model is trained to obtain a corresponding product pricing model, and then obtain corresponding product pricing data; Conduct a comprehensive analysis of product pricing data to obtain product pricing results.

[0043] The present invention has significant technical effects due to the adoption of the above technical solution: The present invention obtains product-related data sets and performs feature extraction, constructs a comprehensive feature data model, combines product feature data sets to obtain a comprehensive feature data set, constructs several product pricing pre-training models, analyzes product feature data sets and comprehensive feature data sets, obtains product pricing prediction data, obtains product pricing data through model training and data reasoning, and then obtains product pricing results. The present invention solves the problem that the existing methods cannot quickly adjust prices and are difficult to adapt to rapid market changes. At the same time, through multi-dimensional data analysis, it accurately reflects the needs of the market and customers, avoiding the disadvantages of profit loss or customer loss. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0045] Figure 1 It is a schematic flow diagram of the method of the present invention; Figure 2 It is an overall schematic diagram of the system of the present invention. DETAILED DESCRIPTION

[0046] The present invention is further described in detail below in conjunction with embodiments. The following embodiments are for explanation of the present invention but the present invention is not limited to the following embodiments.

[0047] Embodiment 1: A product pricing method based on data analysis, such as Figure 1 As shown, the following steps are included: S100, obtaining a product-related data set and performing analysis and feature extraction to obtain a product feature data set, wherein the product-related data includes sales data, inventory data, procurement data, and environmental data, and the product feature data includes sales feature data, inventory feature data, environmental order data, order feature data, category feature data, and time feature data; S200, constructing a comprehensive feature data model, and combining it with a product feature data set to obtain a comprehensive feature data set, wherein the comprehensive feature data model includes a sales comprehensive data model, a price comprehensive data model, a customer comprehensive data model, and an inventory comprehensive data model, and the comprehensive feature data includes sales comprehensive feature data, price comprehensive feature data, customer comprehensive feature data, and inventory comprehensive feature data; S300, constructing several product pricing pre-training models, each of which includes a feature receiving layer, a feature processing layer and a result output layer, wherein the feature receiving layer receives product feature data and comprehensive feature data, the feature processing layer performs corresponding analysis on the product feature data and comprehensive feature data, and the result output layer obtains product pricing prediction data based on the analysis results; S400, training each product pricing pre-training model based on the product feature data set and the comprehensive feature data set to obtain a corresponding product pricing model, and then obtain corresponding product pricing data; S500: Comprehensively analyze product pricing data to obtain product pricing results.

[0048] The present invention extracts features from product-related data sets, obtains a comprehensive feature data set based on a comprehensive feature data model, analyzes the comprehensive feature data set through a number of product pricing pre-training models, obtains product pricing prediction data, trains the product pricing pre-training model to obtain a product pricing model, and then obtains corresponding product pricing data and obtains product pricing results through comprehensive analysis. The present invention efficiently analyzes and mines product-related data sets, realizes dynamic adjustment of product prices, dynamically adjusts pricing for different times, markets, customers and product categories, implements accurate pricing for products and increases the profit margin of enterprises.

[0049] Taking glass products as an example, this embodiment collects a large amount of data generated in the process of glass procurement and sales for analysis to obtain product-related data, among which historical sales prices, sales quantities, customer information and sales areas are obtained to form sales data; inventory data of various glass products, expired glass products and glass loss data are obtained to form inventory data; glass procurement costs, supplier information and glass procurement batches are obtained to form procurement data; and glass industry supply and demand data and external factors such as weather environment are obtained to form environmental data.

[0050] To obtain a product feature data set, you first need to analyze product-related data, including: analyzing sales data to obtain product price data and product sales data; analyzing inventory data to obtain current product inventory data; analyzing sales data and procurement data to obtain product demand forecast data and product replenishment cycle; obtaining competitive price data and customer order data through sales data, and obtaining industry trend data and seasonal impact data based on sales data and inventory data. Then, the product feature data set is obtained through feature extraction, where product feature data includes sales feature data, inventory feature data, environmental order data, order feature data, category feature data, and time feature data. Sales feature data is obtained through the following steps: Step 1: Preset a time period and obtain the average sales price of glass products within the time period to reflect the market acceptance of the product. By analyzing the product price data and product sales data, the product average price data and product total sales data within the time period are obtained. Among them, the sales characteristic data includes product average price data, product total sales data, price fluctuation data, sales growth data and demand elasticity data, including the following steps:

[0051]

[0052] Step 2: Obtain the standard deviation of the product's historical sales price to reflect the instability of the product price. Analyze the fluctuations of the product price data and the product average price data to obtain the price fluctuation data, which is expressed as follows:

[0053] Step 3: Preset a time threshold and analyze the glass sales growth rate within the time threshold to identify the changing trend of product demand. By analyzing the product sales data of the current period and the previous period, the sales growth data is obtained, which is expressed as follows:

[0054] Step 4: Analyze the impact of product price changes on product sales and obtain demand elasticity data, which is expressed as follows:

[0055] in, Indicates the average price data of the product within the preset time period. Indicates product price data within a preset time period. Represents product sales data within a time period. Represents the total sales data of the product. Represents product sales data, Represents price fluctuation data, represents sample data, Indicates sales growth data, Indicates the product sales data within the current preset time threshold. Indicates the product sales data within the previous preset time threshold. represents the demand elasticity data, represents the percentage change in sales volume, Indicates the percentage change in price.

[0056] Feature extraction is performed on product-related data sets to obtain inventory feature data, which is used to describe the relationship between product prices and inventory data. The inventory feature data includes inventory turnover rate data, safety stock data, and inventory turnover data. The specific steps include: Step 1: Calculate the ratio between product sales data and current product inventory data to measure the turnover rate of inventory products and obtain inventory turnover rate data, which is expressed as follows:

[0057] Step 2: Analyze and calculate the product demand forecast data and product replenishment cycle to obtain safety stock data. The safety stock data is used to indicate the probability of product inventory being out of stock, which serves as a constraint condition for product pricing strategy. Step 3: Analyze the average number of days from purchase to sales of each product's inventory to obtain inventory turnover data. A smaller inventory turnover data means that the product circulates faster, and the product price can be adjusted based on this. For example, the price of the product with faster circulation can be increased to cope with changes in market demand. The inventory turnover data is expressed as follows:

[0058] in, Represents inventory turnover data, Represents product sales data, Indicates the current product inventory data. Represents inventory turnover data.

[0059] Feature extraction is performed on product-related data sets to obtain environmental feature data, where the environmental feature data includes market demand data, weather factor data, and holiday effect data. Specifically, the following steps are included: Step 1: Analyze the impact of the current industry and season on product prices through industry trend data and seasonal impact data, and then analyze the market demand for glass products to obtain market demand data; Step 2: The demand for glass products is closely related to weather changes. For example, the demand for window glass may increase in winter or during extreme climates. Obtain current weather data and analyze the impact of current weather data on the demand for glass products through weather data and product demand data to obtain weather factor data. Step 3: During holidays, adjust the prices of glass products through promotional activities using pricing strategies, and analyze the price adjustment methods and adjusted product price data to obtain holiday effect data.

[0060] Feature extraction is performed on product-related data sets to obtain order feature data, where the order feature data includes customer type data, order frequency data, and customer life cycle data. Specifically, the following steps are included: Step 1: Analyze customer order data and distinguish different types of customers (such as glass processing plants, regional service providers, original glass plants, etc.). Different customer groups have different sensitivities to product price data. Analyze the sensitivity of different types of customers to product price data to obtain customer price sensitivity data, and then construct different customer categories to obtain customer type data; Step 2: Preset the order period threshold, obtain the customer's purchase frequency within the order period threshold, analyze the impact of purchase frequency on product price data, and obtain order frequency data. For example, customers who frequently purchase products can enjoy discounts, which affects product pricing strategies. Step 3: Assess the customer's purchasing power through historical order data combined with order frequency data, and measure the customer's contribution to the company throughout the life cycle to obtain the customer's expected value, and then obtain customer life cycle data, and adopt strategies such as product price discounts for high-value customers.

[0061] This embodiment analyzes the impact of different types of glass products (such as transparent glass, tempered glass, heat-insulating glass, etc.) on the price of glass products, adjusts the price according to different product types, and obtains product type data; in the production and sales process of glass products, the size of glass products often affects the price of glass products. Glass products with larger sizes often have higher costs and consume more in the procurement and transportation process. Therefore, when the product size is larger, the product price is correspondingly higher. By analyzing the impact of product size on product price, product size data is obtained; the quality grades of glass products (such as conventional glass, durable glass, fine glass, etc.) are distinguished. High-quality glass products are priced higher. The impact of quality grade on product pricing is analyzed to obtain quality grade data; the life cycle of glass products (such as new products, mature products, obsolete products, etc.) will also affect the product pricing strategy. For example, new products may need to increase market share through promotions, while obsolete products may need to clear inventory through price cuts. By analyzing the impact of product life cycles on product prices, product life data is obtained. Category feature data is formed through product type data, product size data, quality grade data, and product life data.

[0062] In addition, the product price data of glass products will be affected by time, including seasonal factors (such as a surge in demand in summer and a decline in demand in winter, etc.) and cyclical factors (such as day-night, monthly, and quarterly changes, etc.). Some glass products may sell better during the day, while others may be more in demand at night. Working days and weekends also have a certain impact on glass product price data. In addition, specific holidays may bring additional demand for glass products, such as year-end promotions, etc. Therefore, time characteristic data is obtained by analyzing seasonal and cyclical changes in product price data.

[0063] A sales comprehensive data model, a price comprehensive data model, a customer comprehensive data model and an inventory comprehensive data model are constructed respectively, and combined with the product feature data set to obtain sales comprehensive feature data, price comprehensive feature data, customer comprehensive feature data and inventory comprehensive feature data to form a comprehensive feature data set, including the following steps: Step 1: Construct a comprehensive sales data model. The higher the total sales data of a product, the stronger the demand for the product. The price fluctuation data indicates the price sensitivity of the product. The stronger the fluctuation, the greater the reaction of the product sales to price changes. The current product inventory data reflects the inventory capacity of the product. Excessive inventory will affect the room for product price increase. Analyze the total sales data, price fluctuation data, demand elasticity data and current product inventory data to obtain comprehensive sales feature data, which is used to measure the sales potential of the product in the future and provide a strong basis for product pricing. The comprehensive sales feature data is expressed as follows:

[0064] Step 2: Construct a comprehensive price data model. Demand elasticity data is used to measure the impact of price changes on product sales. Products with high price elasticity can achieve maximum profit by fine-tuning prices. Inventory turnover data reflects the liquidity of products. Products with strong liquidity can increase prices and reduce excessive inventory. Competitive price data is a key factor constraining product pricing data. If the product price is too high or too low, it will affect the market competitiveness of the product. Market demand data is used to reflect the strength of market demand. When demand is strong, prices can be increased, otherwise prices may need to be reduced. Demand elasticity data, inventory turnover data, competitive price data, current product price data and market demand data are analyzed to obtain comprehensive price feature data, which is expressed as follows:

[0065] Step 3: Build a comprehensive customer data model. Customer life cycle data represents the total purchase value of customers throughout their life cycle, and customer loyalty data represents the customer's repurchase rate and purchase stability. Obtain single order amount data and customer loyalty data through customer order data, and analyze them in combination with order frequency data and customer life cycle data to obtain comprehensive customer feature data, which is expressed as follows:

[0066] Step 4: Construct a comprehensive inventory data model, analyze the current product inventory data, market demand data, and inventory turnover rate data, and obtain comprehensive inventory feature data, which is expressed as follows:

[0067] in, Represents sales comprehensive characteristic data, Represents the total sales data of the product. Represents price fluctuation data, represents the demand elasticity data, Represents product inventory data, , , , represents the product feature weight, Represents comprehensive price feature data, Represents inventory turnover data, represents competitive price data, Represents the current product price data, Represents market demand data, , , , represents the price feature weight, Represents comprehensive customer feature data, Represents customer lifecycle data, represents order frequency data, Indicates the single order amount data. represents customer loyalty data, , , , represents the customer feature weight, Represents comprehensive inventory feature data, Indicates the current product inventory data. , Represents the inventory feature weight.

[0068] The comprehensive characteristic data of the product directly affects the pricing of the product. If the total sales data of the product is larger and the price fluctuation data is smaller, it means that the price is stable. If the demand elasticity data is high and the product inventory data is moderate, then the comprehensive characteristic data of sales will be higher, indicating that the product has greater room for pricing optimization. On the contrary, if the product inventory data is large, the price fluctuation data is high, and the demand elasticity data is unstable, then the sales characteristic data is low, indicating that the current product price adjustment should be cautious; if the product price elasticity is high, the inventory turnover rate data is fast, and the market demand is strong, then the comprehensive characteristic data of price is high, indicating that the product can appropriately increase the price to obtain more profit, otherwise the price comprehensive characteristic data is low, indicating that the product needs to be reduced in price or the product inventory needs to be adjusted; if the customer life cycle data is high, the order frequency is high and the order amount data is large, the corresponding customer comprehensive characteristic data is high, indicating that this type of customer is more important to the company, and personalized pricing strategies can be provided to improve customer loyalty. Otherwise, it is necessary to adjust the sales strategy or further analyze customer needs to increase customer repurchase rate; if the product inventory is too much and the demand is low, the inventory comprehensive characteristic data is negative, indicating that the product needs to reduce the price to clear the inventory, otherwise the inventory comprehensive characteristic data is positive, the product price can be increased.

[0069] In order to improve the accuracy of product pricing data, in this embodiment, several product pricing pre-training models are constructed to analyze product feature data and comprehensive feature data to obtain corresponding product pricing prediction data, wherein the product pricing pre-training model includes a feature receiving layer, a feature processing layer and a result output layer. In this embodiment, the product feature data and comprehensive feature data are received by the feature receiving layer; the feature processing layer analyzes the importance to identify the product feature data and comprehensive feature data that contribute more to the prediction of the product pricing prediction data. If the comprehensive feature data has a higher correlation with the target value, a higher feature weight is assigned, thereby obtaining the feature weights of all product feature data and comprehensive feature data; the result output layer analyzes based on the feature weights, product feature data and comprehensive feature data to obtain product pricing prediction data, which is expressed as follows:

[0070] This embodiment also includes a training process for the product pricing pre-training model. Through the product pricing prediction data, feature weights and real product pricing data, combined with L2 regularization to avoid overfitting of the model, the feature weight distribution of the comprehensive feature data and the product feature data is made smoother, and then a product loss function and a comprehensive loss function are constructed. The pricing linear loss function is formed through the product loss function and the comprehensive loss function, wherein the product loss function and the comprehensive loss function are expressed as follows:

[0071]

[0072] The pricing linear loss function is gradient-solved. Based on the gradient solution result, the feature weights in the product pricing pre-training model are updated using the gradient descent method until the model converges or meets the performance threshold, and the product pricing model is obtained. The gradient solution and weight update are expressed as follows:

[0073] The product pricing model is used to analyze and infer the product feature data and comprehensive feature data to be predicted, and obtain product pricing data; in, represents the product loss function, represents the comprehensive loss function, represents the pricing linear loss function, represents the number of samples, Indicates The real pricing data of samples, represents product pricing forecast data, Represents the importance parameter of comprehensive feature data, Indicates the quantity of product feature data, represents the number of comprehensive feature data, Indicates The feature weight of each product feature data, Indicates The feature weight of the comprehensive feature data, represents the weight adjustment factor, Indicates bias, Indicates The sample Product feature data, Indicates The sample Comprehensive feature data, represents the weight update, Indicates the weight of product feature data or the weight of comprehensive feature data, represents the learning rate, Indicates the dynamic adjustment coefficient.

[0074] This embodiment also provides a product pricing pre-training model, including a feature receiving layer, a feature processing layer and a result output layer. The feature receiving layer receives product feature data and comprehensive feature data. The corresponding feature processing layer recursively splits the product feature data and comprehensive feature data through a decision tree. During the training process, the decision tree determines the importance of each feature in the recursive splitting process, analyzes the change in the amount of information of the product feature data and the comprehensive feature data before and after the split, obtains information gain, and obtains the splitting node for the next split based on the information gain, wherein the information gain is expressed as follows:

[0075] The category probabilities of the product feature data and comprehensive feature data after splitting are obtained, and the feature Gini index is obtained based on the category probability. The recursive splitting is fed back through the feature Gini index until the optimal splitting point is obtained. The feature Gini index is expressed as follows:

[0076] The output layer analyzes the product feature data and comprehensive feature data through the optimal split point to obtain product pricing data; in, represents information gain, represents the information entropy of the split node of the last split, represents the Gini index, Indicates that product feature data or comprehensive feature data belongs to the category The probability of Represents a collection of product feature data and comprehensive feature data, Indicates the first subsets.

[0077] This embodiment also provides a product pricing pre-training model as a gradient boosting algorithm, including a feature receiving layer, a feature processing layer and a result output layer, wherein the feature receiving layer receives product feature data and comprehensive feature data, the feature processing layer constructs a pricing learner, analyzes and predicts the product feature data and comprehensive feature data based on the pricing learner, obtains pricing residuals based on product pricing data of the previous iteration, performs iterative analysis through pricing residuals and learning rates, and constructs a new pricing learner; The output layer obtains the product pricing prediction data through the new pricing learner, learning rate and product pricing data of the previous iteration, which is expressed as follows:

[0078] It also includes the training process of the product pricing pre-training model. Through the product pricing prediction data, the real pricing data and the regularization term, a pricing regression loss function is constructed. The regularization term is used to control the complexity of the model and prevent the model from overfitting, which is expressed as follows:

[0079] The feature importance of product feature data and comprehensive feature data is analyzed through pricing residuals to obtain feature gain. The product pricing pre-training model is trained through feature gain and pricing regression loss function to obtain the product pricing model. Based on the product pricing model, the product feature data and comprehensive feature data to be predicted are analyzed to obtain product pricing data; in, represents the pricing regression loss function, Indicates Product pricing forecast data, represents the regularization term, represents the prediction of the current tree, represents the number of trees in the model, represents the number of samples, represents the learning rate, represents product pricing forecast data, express Keshu’s product pricing prediction data.

[0080] In this embodiment, after obtaining corresponding product pricing data through several product pricing models, the corresponding product pricing data is integrated for comprehensive analysis to better capture the relationship between the data, further optimize the accuracy of the product pricing data, obtain more accurate product pricing results, and effectively improve the accuracy and reliability of the product pricing model, including the following steps: Construct a comprehensive pricing pre-training model. The comprehensive pricing pre-training model performs linear combination on the product pricing data, obtains the characteristic coefficient vector through initialization, and constructs a comprehensive pricing loss function in combination with the real product pricing data, which is expressed as follows:

[0081] The feature coefficient vector is iteratively updated through the comprehensive pricing loss function to obtain the optimal feature coefficient vector, and then the comprehensive pricing model is obtained, where the update process is expressed as follows:

[0082]

[0083] The product pricing data corresponding to several product pricing models are analyzed and inferred through the comprehensive pricing model to obtain the product pricing results, which are expressed as follows:

[0084] in, represents the comprehensive pricing loss function, represents real product pricing data, Indicates product characteristic data or comprehensive characteristic data, Indicates The characteristic coefficient vector, represents the number of characteristic coefficient vectors, represents the regularization parameter, represents the number of samples, represents the number of iterations, Indicates the product characteristic data or the first of the comprehensive characteristic data data, Represents the product pricing result, represents the comprehensive pricing model, Indicates Product pricing data, represents the number of base learners, Indicates product characteristic data or comprehensive characteristic data.

[0085] The method of the present invention conducts multi-dimensional analysis on product-related data sets, constructs several product pricing pre-training models, and conducts model training based on product feature data sets and comprehensive feature data sets to obtain product pricing models, thereby obtaining corresponding product pricing data, and conducting comprehensive analysis on several product pricing data to obtain product pricing results. Dynamic optimization of product pricing is achieved, and product sales prices can be adjusted in real time according to market changes, inventory conditions and customer needs, ensuring that the company's product pricing is flexible and accurate, and improving the company's profitability and market competitiveness.

[0086] Embodiment 2: A product pricing system based on data analysis, such as Figure 2 As shown, it includes a data acquisition module 100, a feature extraction module 200, a model construction module 300, a model training module 400 and a comprehensive prediction module 500; The data acquisition module 100 acquires product-related data sets and performs analysis and feature extraction to obtain product feature data sets, wherein the product-related data includes sales data, inventory data, purchase data and environmental data, and the product feature data includes sales feature data, inventory feature data, environmental order data, order feature data, category feature data and time feature data; The feature extraction module 200 constructs a comprehensive feature data model and obtains a comprehensive feature data set by combining the product feature data set, wherein the comprehensive feature data model includes a sales comprehensive data model, a price comprehensive data model, a customer comprehensive data model and an inventory comprehensive data model, and the comprehensive feature data includes sales comprehensive feature data, price comprehensive feature data, customer comprehensive feature data and inventory comprehensive feature data; The model building module 300 builds a plurality of product pricing pre-training models, each of which includes a feature receiving layer, a feature processing layer and a result output layer. The feature receiving layer receives product feature data and comprehensive feature data, the feature processing layer performs corresponding analysis on the product feature data and comprehensive feature data, and the result output layer obtains product pricing prediction data based on the analysis results. The model training module 400 trains each product pricing pre-training model based on the product feature data set and the comprehensive feature data set to obtain a corresponding product pricing model, and then obtain corresponding product pricing data; The comprehensive prediction module 500 performs comprehensive analysis on the product pricing data to obtain product pricing results.

[0087] Various changes and modifications can be made without departing from the spirit and scope of the present invention, and all equivalent technical solutions also belong to the scope of the present invention.

[0088] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0089] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0090] The present invention is described with reference to the flowcharts and / or block diagrams of the method, terminal device (system), and computer program product according to the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0091] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0092] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable terminal device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0093] It should be noted that: The "one embodiment" or "embodiment" mentioned in the specification means that the specific features, structures or characteristics described in conjunction with the embodiment are included in at least one embodiment of the present invention. Therefore, the phrases "one embodiment" or "embodiment" appearing in various places throughout the specification do not necessarily refer to the same embodiment.

[0094] In addition, it should be noted that the shapes and names of the parts and components of the specific embodiments described in this specification may be different. Any equivalent or simple changes made based on the structure, features and principles described in the patent concept of the present invention are included in the protection scope of the patent of the present invention. The technicians in the technical field of the present invention can make various modifications or supplements to the specific embodiments described or replace them in a similar manner, as long as they do not deviate from the structure of the present invention or exceed the scope defined by the claims, they should all fall within the protection scope of the present invention.

Claims

1. A product pricing method based on data analysis, characterized in that: The following steps are involved: Obtain product-related data sets and perform analysis and feature extraction to obtain product feature data sets, wherein product-related data include sales data, inventory data, procurement data and environmental data, and product feature data include sales feature data, inventory feature data, environmental order data, order feature data, category feature data and time feature data; Constructing a comprehensive feature data model, and combining it with the product feature data set to obtain a comprehensive feature data set, wherein the comprehensive feature data model includes a sales comprehensive data model, a price comprehensive data model, a customer comprehensive data model, and an inventory comprehensive data model, and the comprehensive feature data includes sales comprehensive feature data, price comprehensive feature data, customer comprehensive feature data, and inventory comprehensive feature data; Constructing several product pricing pre-training models, each of which includes a feature receiving layer, a feature processing layer and a result output layer, wherein the feature receiving layer receives product feature data and comprehensive feature data, the feature processing layer performs corresponding analysis on the product feature data and comprehensive feature data, and the result output layer obtains product pricing prediction data based on the analysis results; Based on the product feature data set and the comprehensive feature data set, each product pricing pre-training model is trained to obtain a corresponding product pricing model, and then obtain corresponding product pricing data; Conduct a comprehensive analysis of product pricing data to obtain product pricing results.

2. The product pricing method based on data analysis according to claim 1, characterized in that: Also includes: Analyze sales data to obtain product price data and product sales data; Analyze the inventory data to obtain the current product inventory data; Analyze sales data and purchasing data to obtain product demand forecast data and product replenishment cycle; Obtain competitive price data and customer order data through sales data, and obtain industry trend data and seasonal impact data based on sales data and inventory data.

3. The product pricing method based on data analysis according to claim 1, characterized in that: The sales characteristic data is obtained through the following steps: Based on the product price data and product sales data, the average product price data and total product sales data within the preset time period are obtained, as shown below: By analyzing the fluctuations of product price data and product average price data, we can obtain price fluctuation data, which is shown as follows: Preset a time threshold, analyze the change trend of product sales data within the time threshold, and obtain sales growth data, which is expressed as follows: Analyze the impact of product price data on product sales data and obtain demand elasticity data, which is expressed as follows: Sales characteristic data is formed through product average price data, product total sales data, price fluctuation data, sales growth data and demand elasticity data; in, Indicates the average price data of the product within the preset time period. Indicates product price data within a preset time period. Represents product sales data within a time period. Represents the total sales data of the product. Represents product sales data, Represents price fluctuation data, represents sample data, Indicates sales growth data, Indicates the product sales data within the current preset time threshold. Indicates the product sales data within the previous preset time threshold. represents the demand elasticity data, represents the percentage change in sales volume, Indicates the percentage change in price.

4. The product pricing method based on data analysis according to claim 1, characterized in that: The inventory characteristic data is obtained through the following steps: Based on the analysis of product sales data and current product inventory data, the inventory turnover rate data is obtained, which is expressed as follows: Based on product replenishment cycle and product demand forecast data, analyze product inventory and product replenishment status to obtain safety inventory data; The average number of days from product purchase to sale is analyzed through inventory turnover rate data to obtain inventory turnover data, which is expressed as follows: Through inventory turnover rate data, safety stock data and inventory turnover data, inventory characteristic data is formed; in, Represents inventory turnover data, Represents product sales data, Indicates the current product inventory data. Represents inventory turnover data.

5. The product pricing method based on data analysis according to claim 1, characterized in that: The environmental characteristic data is obtained by following the steps below: Analyze the market demand of products through industry trend data and seasonal impact data to obtain market demand data; Obtain weather data, and analyze the impact of weather factors on product demand based on the weather data and product demand forecast data to obtain weather factor data; During holidays, product price data is adjusted, and holiday effect data is obtained based on the adjustment method and the adjusted product price data; Environmental characteristic data are formed through market demand data, weather factor data and holiday effect data.

6. The product pricing method based on data analysis according to claim 1, characterized in that: The order characteristic data is obtained through the following steps: Analyze customer order data and product price data to obtain customer price sensitivity data, and divide different customer groups. Analyze customer types and customer price sensitivity data to obtain customer type data; Preset an order period threshold, obtain the customer's purchase frequency within the order period threshold through the customer order data, and obtain order frequency data; Obtain and analyze historical customer order data to obtain the customer's expected value, and then obtain customer life cycle data through the customer's expected value; Order characteristic data is formed through customer type data, order frequency data and customer life cycle data.

7. The product pricing method based on data analysis according to claim 1, characterized in that: The category characteristic data is obtained by analyzing the impact of product type, product size, product grade and product life cycle on product price data, wherein the category characteristic data includes product type data, product size data, quality grade data and product life data; The time characteristic data is obtained by performing seasonal change analysis and cyclical change analysis on product price data.

8. The product pricing method based on data analysis according to claim 1, characterized in that: The construction of the comprehensive feature data model and combining the product feature data set to obtain the comprehensive feature data set includes the following steps: A comprehensive sales data model is constructed to analyze the total product sales data, price fluctuation data, demand elasticity data, and current product inventory data to obtain comprehensive sales feature data, which is expressed as follows: A comprehensive price data model is constructed to analyze demand elasticity data, inventory turnover data, competitive price data, current product price data and market demand data to obtain comprehensive price feature data, which is expressed as follows: Build a comprehensive customer data model, obtain single order amount data and customer loyalty data through customer order data, and analyze it in combination with order frequency data and customer life cycle data to obtain comprehensive customer feature data, which is expressed as follows: Construct a comprehensive inventory data model, analyze the current product inventory data, market demand data and inventory turnover rate data, and obtain comprehensive inventory feature data, which is expressed as follows: A comprehensive feature data set is formed by using sales comprehensive feature data, price comprehensive feature data, customer comprehensive feature data and inventory comprehensive feature data; in, Represents sales comprehensive characteristic data, Represents the total sales data of the product. Represents price fluctuation data, represents the demand elasticity data, Represents product inventory data, , , , represents the product feature weight, Represents comprehensive price feature data, Represents inventory turnover data, represents competitive price data, Represents the current product price data, Represents market demand data, , , , represents the price feature weight, Represents comprehensive customer feature data, Represents customer lifecycle data, represents order frequency data, Indicates the single order amount data. represents customer loyalty data, , , , represents the customer feature weight, Represents comprehensive inventory feature data, Indicates the current product inventory data. , Represents the inventory feature weight.

9. The product pricing method based on data analysis according to claim 1, characterized in that: The feature receiving layer receives product feature data and comprehensive feature data; The corresponding feature processing layer analyzes the importance of product feature data and comprehensive feature data, and obtains feature weights based on the importance analysis results; The result output layer analyzes the feature weights, product feature data, and comprehensive feature data to obtain product pricing prediction data, which is expressed as follows: The product loss function and the comprehensive loss function are constructed through the product pricing prediction data, feature weights and real product pricing data, and the pricing linear loss function is formed through the product loss function and the comprehensive loss function. The product loss function and the comprehensive loss function are expressed as follows: The pricing linear loss function is gradient-solved, and based on the gradient solution result, the feature weights of the product pricing pre-training model are updated using the gradient descent method to obtain the product pricing model. The gradient solution and weight update are expressed as follows: The product pricing model is used to analyze and infer the product feature data and comprehensive feature data to be predicted, and obtain product pricing data; in, represents the product loss function, represents the comprehensive loss function, represents the pricing linear loss function, represents the number of samples, Indicates The real pricing data of samples, represents product pricing forecast data, Represents the importance parameter of comprehensive feature data, Indicates the quantity of product feature data, represents the number of comprehensive feature data, Indicates The feature weight of each product feature data, Indicates The feature weight of the comprehensive feature data, represents the weight adjustment factor, Indicates bias, Indicates The sample Product feature data, Indicates The sample Comprehensive feature data, represents the weight update, Indicates the weight of product feature data or the weight of comprehensive feature data, represents the learning rate, Indicates the dynamic adjustment coefficient.

10. The product pricing method based on data analysis according to claim 1, characterized in that: The feature receiving layer receives product feature data and comprehensive feature data; The corresponding feature processing layer recursively splits the product feature data and the comprehensive feature data, analyzes the information changes of the product feature data and the comprehensive feature data before and after the split, obtains the information gain, and obtains the split node for the next split based on the information gain, where the information gain is expressed as follows: The category probabilities of the product feature data and comprehensive feature data after splitting are obtained, and the feature Gini index is obtained based on the category probability. The recursive splitting is fed back through the feature Gini index until the optimal splitting point is obtained. The feature Gini index is expressed as follows: The output layer analyzes the product feature data and comprehensive feature data through the optimal split point to obtain product pricing data; in, represents information gain, represents the information entropy of the split node of the last split, represents the Gini index, Indicates that product feature data or comprehensive feature data belongs to the category The probability of Represents a collection of product feature data and comprehensive feature data, Indicates the first subsets.

11. The product pricing method based on data analysis according to claim 1, characterized in that: The feature receiving layer receives product feature data and comprehensive feature data; The corresponding feature processing layer constructs a pricing learner, analyzes and predicts product feature data and comprehensive feature data based on the pricing learner, and obtains pricing residuals based on the product pricing prediction data of the previous iteration. It iterates and constructs a new pricing learner through pricing residuals and learning rates; The output layer obtains the current product pricing prediction data through the new pricing learner, learning rate and the product pricing prediction data of the previous iteration, which is expressed as follows: Through product pricing prediction data, real pricing data and regularization terms, a pricing regression loss function is constructed, which is expressed as follows: The feature importance of product feature data and comprehensive feature data is analyzed through pricing residuals to obtain feature gain. The product pricing pre-training model is trained through feature gain and pricing regression loss function to obtain the product pricing model. Based on the product pricing model, the product feature data and comprehensive feature data to be predicted are analyzed to obtain product pricing data; in, represents the pricing regression loss function, Indicates Product pricing forecast data, represents the regularization term, represents the prediction of the current tree, represents the number of trees in the model, represents the number of samples, represents the learning rate, represents product pricing forecast data, express Product pricing prediction data from Keshu.

12. The product pricing method based on data analysis according to claim 1, characterized in that: The comprehensive analysis of the product pricing data to obtain the product pricing results includes the following steps: A comprehensive pricing pre-training model is constructed. The product pricing data is linearly combined through the comprehensive pricing pre-training model and combined with the real product pricing data and feature coefficient vector to construct a comprehensive pricing loss function, which is expressed as follows: The feature coefficient vector is updated through the comprehensive pricing loss function to obtain the optimal feature coefficient vector, and then the comprehensive pricing model is obtained. The update process is expressed as follows: The product pricing data corresponding to several product pricing models are analyzed and inferred through the comprehensive pricing model to obtain the product pricing results, which are expressed as follows: in, represents the comprehensive pricing loss function, represents real product pricing data, Indicates product characteristic data or comprehensive characteristic data, Indicates The characteristic coefficient vector, represents the number of characteristic coefficient vectors, represents the regularization parameter, represents the number of samples, represents the number of iterations, Indicates the product characteristic data or the first of the comprehensive characteristic data data, Represents the product pricing result, represents the comprehensive pricing model, Indicates Product pricing data, represents the number of base learners, Indicates product characteristic data or comprehensive characteristic data.

13. A product pricing system based on data analysis, characterized in that: It includes data acquisition module, feature extraction module, model construction module, model training module and comprehensive prediction module; The data acquisition module acquires product-related data sets and performs analysis and feature extraction to obtain product feature data sets, wherein the product-related data includes sales data, inventory data, procurement data and environmental data, and the product feature data includes sales feature data, inventory feature data, environmental order data, order feature data, category feature data and time feature data; The feature extraction module constructs a comprehensive feature data model and obtains a comprehensive feature data set by combining the product feature data set, wherein the comprehensive feature data model includes a sales comprehensive data model, a price comprehensive data model, a customer comprehensive data model and an inventory comprehensive data model, and the comprehensive feature data includes sales comprehensive feature data, price comprehensive feature data, customer comprehensive feature data and inventory comprehensive feature data; The model building module builds a plurality of product pricing pre-training models, each of which includes a feature receiving layer, a feature processing layer and a result output layer. The feature receiving layer receives product feature data and comprehensive feature data, the feature processing layer performs corresponding analysis on the product feature data and comprehensive feature data, and the result output layer obtains product pricing prediction data based on the analysis results. The model training module trains each product pricing pre-training model based on the product feature data set and the comprehensive feature data set to obtain a corresponding product pricing model, and then obtain corresponding product pricing data; The comprehensive prediction module conducts a comprehensive analysis on the product pricing data to obtain the product pricing result.

14. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 12 is implemented.

15. A product pricing device based on data analysis, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 12 is implemented.

Images