E-commerce transaction method and system

Abstract

The invention provides an e-commerce transaction method and system, and relates to the technical field of e-commerce transactions, and the method comprises the steps: obtaining commodity sales data, carrying out the feature selection, selecting a feature vector according to a feature value sequence, and carrying out the optimization of features; inputting the optimized feature matrix, and training a sales prediction model; forming a sales willingness feature set according to the sales willingness of the merchant, generating a predicted sales volume of each region in each time period, and generating an inventory amount according to the supply condition of the merchant so as to obtain the total inventory amount; obtaining the commodity inventory cost, optimizing the inventory, and obtaining the optimal inventory with the minimum cost; according to the system, a sales volume prediction model is generated according to historical sales data to generate sales volume prediction data for merchants, and cost-saving inventory preparation data is provided for the merchants according to the sales volume, so that the operation cost of the merchants is saved.

Description

Technical Field

[0001] The present invention relates to the technical field of e-commerce transactions, and in particular to an e-commerce transaction method and system. Background Art

[0002] In the e-commerce environment, inventory management is crucial for merchants. Traditional inventory management methods rely on experience and simple statistical forecasts, which can easily lead to excess or shortages of inventory, thereby affecting the profitability and customer satisfaction of the company. At the same time, with the rapid changes in the market and the uncertainty of consumer demand, merchants face more complex inventory management challenges and need more accurate forecasting and optimization methods to improve inventory management efficiency.

[0003] In the prior art, publication number CN202011203414.6 discloses an intelligent inventory management system for e-commerce platform stores based on cloud computing, including an information acquisition module, an image acquisition module, an image processing module, a product model acquisition module, a product model analysis module, a manual quality inspection module, a defective area management module, a stock area management module, an analysis server, an e-commerce platform and a storage database; the present invention collects images of each returned product, determines whether the returned product is consistent with the product information in the return order, and constructs a model of each returned product with consistent information, analyzes the model corresponding to each returned product, and analyzes whether the appearance and function of each returned product are qualified, puts each qualified returned product into the inventory position of the corresponding model, counts the number of returned products of each model in storage, calculates the real-time inventory and real-time total inventory of each model in the store, and displays them.

[0004] Although the above information disclosed in the background technology section realizes the management of inventory by the trading platform, it lacks predictive foresight, cannot respond to upcoming sales in a timely manner, and cannot save costs on commodity storage. Summary of the invention

[0005] The object of the present invention is to provide an e-commerce transaction method and system to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] An e-commerce transaction method, the specific steps include:

[0008] Step 1: Obtain product sales data and extract features, pre-process the features and aggregate them into a feature matrix, identify the eigenvalues ​​of each feature in the feature matrix by constructing a covariance matrix, sort them according to the eigenvalue values ​​and perform feature screening to form an optimized feature matrix;

[0009] Step 2: Input the optimized feature matrix into the long short-term memory network model to build a sales forecasting model;

[0010] Step 3: A sales intention feature set is formed based on the merchant's sales intention, where the sales intention feature is the feature that passes the feature screening. The sales intention feature set is input into the sales forecasting model to obtain the sales forecasting matrix. The inventory in each time period and each region is obtained based on the sales forecasting matrix and the merchant's supply capacity, thereby obtaining the total inventory in all regions.

[0011] Step 4: Obtain the cost data of commodity inventory, optimize the total inventory data through Lagrange number multiplication, obtain the optimal inventory quantity and feedback it to the merchant.

[0012] Furthermore, obtain commodity sales data, which includes commodity sales, sales timestamps of each commodity, and destinations to which it is sold. Count the sales in each region in each time period to form a data set, clean the data set, and fill in missing data through interpolation. The formula is as follows:

[0013]

[0014] Among them, x filled is the filled data point, x i-1 and x i+1 are the data before and after the missing value, respectively, i is the data retrieval variable, i∈N, 1≤i≤n;

[0015] The commodity sales data is subjected to feature extraction, and the time features and geographic features in the data are extracted. The time features include seasonal features and working day features, and the geographic features include climate features, regional economic features, and regional population features. The geographic features are collected from the Internet based on the commodity sales timestamp and the destination. The features are normalized to form a feature matrix based on the following formula:

[0016]

[0017] Where: ∑X is the covariance matrix, n is the number of samples in the feature, X standard is the feature matrix, is the transpose of the feature matrix;

[0018] Solve the covariance matrix to obtain the eigenvalues ​​and the corresponding eigenvectors according to the following formula:

[0019] ∑Xv j =λ j v j

[0020] Where: ∑X is the covariance matrix, v jis the jth eigenvector, λ j is the jth eigenvalue, j is the feature retrieval variable, j∈N, 1≤j≤n;

[0021] Sort the eigenvectors according to the eigenvalue size, select and retain the top k eigenvectors, and the formula for obtaining the k value is as follows:

[0022]

[0023] Among them, λ u is the eigenvalue ranked u, k is the number of selected features, m is the number of original features, j is the eigenvalue retrieval variable, u∈N, 1≤u≤k, j≤N, 1≤j≤n, k≤m;

[0024] The eigenvectors corresponding to the first k eigenvalues ​​obtained are retained, the eigenvectors are sorted and numbered according to the size of the eigenvalues, and the eigenvectors corresponding to the eigenvalues ​​ranked after the kth are discarded to form an optimized feature matrix.

[0025] Furthermore, 80% of the optimized feature matrix is ​​divided into a training set and 20% is divided into a test set, and training is performed in a long short-term memory network (LSTM) according to the following formula:

[0026]

[0027] in, To predict sales, v u is the eigenvector corresponding to the top k eigenvalues, α u For each feature vector, the model coefficient is corresponding to b, and the intercept is randomly selected from the optimized feature matrix. The samples are input into the long short-term memory network (LSTM) to obtain the initial values ​​of each model coefficient and intercept. u is the retrieval variable of the feature vector, u∈N, 1≤u≤k.

[0028] The mean square error is used as the loss function, and the formula is as follows:

[0029]

[0030] Among them, MSE is the loss function value, y i For actual sales, To predict sales, n is the number of samples, i is the sample retrieval variable, i∈N, 1≤i≤n;

[0031] Adjust the parameters according to the overall loss function MSE. Minimizing the loss function MSE is the optimal parameter configuration, and the model that passes the test is calibrated as the sales forecast model.

[0032] Furthermore, according to the merchant's sales intention, the time period and sales area data that the merchant wants to focus on are obtained, and the time features and geographical features are extracted and screened according to the optimized feature matrix to form a sales intention feature set, which is input into the sales forecasting model to obtain the predicted sales volume. The predicted sales volume is the sales volume of the product in each region in each time period to form a predicted sales volume matrix;

[0033] Obtain the average supply cycle, replenishment cycle data and the expected service level index from the merchants to obtain the standard deviation of the supply cycle. The formula is as follows:

[0034]

[0035] Among them, σ L is the standard deviation of the supply cycle, M is the number of times the merchant supplies goods, and L j is the actual delivery period for the jth delivery, is the average value of the supply cycle, j is the retrieval variable of the supply cycle, j∈N, 1≤j≤M;

[0036] Get the safety stock of inventory in a single area based on the following formula:

[0037]

[0038] Among them, SS is the safety stock, Z is the standard normal distribution coefficient corresponding to the service level index that the merchant expects to achieve, σ D is the standard deviation of sales forecast in a certain area, T is the replenishment cycle, σ L is the standard deviation of the supply cycle, L is the average supply cycle;

[0039] Generate the recommended total inventory in a single region within a fixed time period based on the following formula:

[0040]

[0041] Among them, I is the recommended total inventory, S p-i is the predicted sales volume in the i-th time period, β is the demand growth rate, T is the replenishment cycle, SS is the safety stock, i∈N, 1≤i≤n;

[0042] The demand growth rate is based on the following formula:

[0043]

[0044] Among them, S p-i is the predicted sales volume in the i-th time period, i is the search variable of the time period, i∈N, 1≤i≤n-1, so as to obtain the total inventory of all regions, based on the following formula:

[0045]

[0046] Among them, I i is the total recommended inventory of the i-th region, i is the region search variable, n is the total number of regions, i∈N, 1≤i≤n;

[0047] Furthermore, the cost data of the inventory of a single commodity is obtained, and the cost data includes the purchase cost, the holding cost, and the out-of-stock cost. The total inventory is optimized by Lagrange number multiplication, and the formula is as follows:

[0048]

[0049] in, is the Lagrangian function, C h is the unit inventory holding cost, C o is the unit out-of-stock cost, C p is the unit purchase cost, S is the out-of-stock quantity, Q is the purchase quantity, and the purchase quantity value is the average of the last sales forecast values ​​of all regions in this fixed time period;

[0050] Set the objective function as:

[0051] Z=C h II+C o ·S+C p Q

[0052] Among them, Z is the total cost;

[0053] Set constraints:

[0054]

[0055] Among them, Z SL is the standard normal distribution coefficient of the service level index that the merchant expects to achieve, σ D is the standard deviation of demand, σ L is the standard deviation of the supply cycle, T is the replenishment cycle, and L is the average supply cycle;

[0056] Solve the Lagrangian function to obtain the optimal inventory I * And feed back the optimal inventory data to the merchant.

[0057] The present invention also discloses an e-commerce transaction system, which is used to execute the above-mentioned e-commerce transaction method, including:

[0058] The information collection and feature selection module is used to obtain commodity sales data and extract features, pre-process the features and aggregate them into a feature matrix, identify the eigenvalues ​​of each feature in the feature matrix by constructing a covariance matrix, sort them according to the eigenvalue values ​​and perform feature screening to form an optimized feature matrix;

[0059] Model training module, used to input the optimized feature matrix into the long short-term memory network model to build a sales forecasting model;

[0060] The inventory statistics module is used to form a sales intention feature set based on the merchant's sales intention, where the sales intention feature is a feature that passes feature screening, input the sales intention feature set into the sales forecast model, obtain the sales forecast matrix, and obtain the inventory of each time period and each region based on the sales forecast matrix and the merchant's supply capacity, thereby obtaining the total inventory of all regions;

[0061] The cost optimization module is used to obtain the cost data of commodity inventory, optimize the total inventory data through Lagrange number multiplication, obtain the optimal inventory quantity and feed it back to the merchant.

[0062] Compared with the prior art, the present invention has the following beneficial effects:

[0063] The present invention generates a sales volume forecasting model based on historical sales data to generate sales volume forecasting data for merchants, and provides cost-saving inventory preparation data for merchants based on sales volume, thereby saving the operating costs of merchants. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 It is a schematic diagram of the overall method flow of the present invention;

[0065] Figure 2 It is a schematic diagram of the overall structure of the present invention. DETAILED DESCRIPTION

[0066] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0067] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0068] Example:

[0069] See also Figure 1 , the present invention provides a technical solution:

[0070] An e-commerce transaction method, the specific steps include:

[0071] Step 1: Obtain product sales data and extract features, pre-process the features and aggregate them into a feature matrix, identify the eigenvalues ​​of each feature in the feature matrix by constructing a covariance matrix, sort them according to the eigenvalue values ​​and perform feature screening to form an optimized feature matrix;

[0072] The step 1 includes the following contents:

[0073] Step 101: Obtain commodity sales data, which includes commodity sales, sales timestamps of each commodity, and destinations to which the commodity is sold. Count the sales in each region in each time period to form a data set, clean the data set, and fill in missing data through interpolation. The formula is as follows:

[0074]

[0075] Among them, x filled is the filled data point, x i-1 and x i+1 are the data before and after the missing value, respectively, i is the data retrieval variable, i∈N, 1≤i≤n;

[0076] Step 102: extract features from the commodity sales data, extract time features and geographic features from the data, the time features include season features and working day features, the geographic features include climate features, regional economic features and regional population features, the geographic features are collected from the Internet based on the commodity sales timestamp and the destination, and the features are normalized to form a feature matrix, based on the formula:

[0077]

[0078] Where: ∑X is the covariance matrix, n is the number of samples in the feature, X standard is the feature matrix, is the transpose of the feature matrix;

[0079] Solve the covariance matrix to obtain the eigenvalues ​​and the corresponding eigenvectors according to the following formula:

[0080] ∑Xv j =λ j v j

[0081] Where: ∑X is the covariance matrix, v j is the jth eigenvector, λ j is the jth eigenvalue, j is the feature retrieval variable, j∈N, 1≤j≤n;

[0082] Step 103: Sort the eigenvectors according to the eigenvalues, select and retain the top k eigenvectors, and the formula for obtaining the k value is as follows:

[0083]

[0084] Among them, λ u is the eigenvalue ranked u, k is the number of selected features, m is the number of original features, j is the eigenvalue retrieval variable, u∈N, 1≤u≤k, j≤N, 1≤j≤n, k≤m;

[0085] The eigenvectors corresponding to the first k eigenvalues ​​obtained are retained, the eigenvectors are sorted and numbered according to the size of the eigenvalues, and the eigenvectors corresponding to the eigenvalues ​​ranked after the kth are discarded to form an optimized feature matrix.

[0086] Through feature selection, features that are not important for the prediction results are discarded, and features that are important for the results are focused on, which reduces the amount of calculation when training the model and avoids interference from individual features.

[0087] Step 2: Input the optimized feature matrix into the long short-term memory network model to build a sales forecasting model;

[0088] The step 2 includes the following contents:

[0089] 80% of the optimized feature matrix is ​​divided into a training set and 20% is divided into a test set, and training is performed in a long short-term memory network (LSTM) according to the following formula:

[0090]

[0091] in, To predict sales, v u is the eigenvector corresponding to the top k eigenvalues, α u For each feature vector, the model coefficient is corresponding to b, and the intercept is randomly selected from the optimized feature matrix. The samples are input into the long short-term memory network (LSTM) to obtain the initial values ​​of each model coefficient and intercept. u is the retrieval variable of the feature vector, u∈N, 1≤u≤k.

[0092] The mean square error is used as the loss function, and the formula is as follows:

[0093]

[0094] Among them, MSE is the loss function value, y i For actual sales, To predict sales, n is the number of samples, i is the sample retrieval variable, i∈N, 1≤i≤n;

[0095] Adjust the parameters according to the overall loss function MSE. Minimizing the loss function MSE is the optimal parameter configuration, and the model that passes the test is calibrated as the sales forecast model.

[0096] Step 3: A sales intention feature set is formed based on the merchant's sales intention, where the sales intention feature is the feature that passes the feature screening. The sales intention feature set is input into the sales forecasting model to obtain the sales forecasting matrix. The inventory in each time period and each region is obtained based on the sales forecasting matrix and the merchant's supply capacity, thereby obtaining the total inventory in all regions.

[0097] The step 3 includes the following contents:

[0098] Step 301: According to the merchant's sales intention, the time period and sales area data that the merchant wants to focus on are obtained, and the time features and geographic features are extracted and screened according to the optimized feature matrix to form a sales intention feature set, which is input into the sales forecasting model to obtain the predicted sales volume. The predicted sales volume is the sales volume of the product in each region in each time period, forming a predicted sales volume matrix;

[0099] Obtain the average supply cycle, replenishment cycle data and the expected service level index from the merchants to obtain the standard deviation of the supply cycle. The formula is as follows:

[0100]

[0101] Among them, σ L is the standard deviation of the supply cycle, M is the number of times the merchant supplies goods, and L j is the actual delivery period for the jth delivery, is the average value of the supply cycle, j is the retrieval variable of the supply cycle, j∈N, 1≤j≤M;

[0102] Get the safety stock of inventory in a single area based on the following formula:

[0103]

[0104] Among them, SS is the safety stock, Z is the standard normal distribution coefficient corresponding to the service level index that the merchant expects to achieve, σ D is the standard deviation of sales forecast in a certain area, T is the replenishment cycle, σ L is the standard deviation of the supply cycle, L is the average supply cycle;

[0105] Step 302: Generate the recommended total inventory in a single region within a fixed time period, based on the following formula:

[0106]

[0107] Among them, I is the recommended total inventory, S p-i is the predicted sales volume in the i-th time period, β is the demand growth rate, T is the replenishment cycle, SS is the safety stock, i∈N, 1≤i≤n;

[0108] The demand growth rate is based on the following formula:

[0109]

[0110] Among them, S p-i is the predicted sales volume in the i-th time period, i is the search variable of the time period, i∈N, 1≤i≤n-1

[0111] The total inventory of all regions is obtained according to the following formula:

[0112]

[0113] Among them, I i is the total recommended inventory of the i-th region, i is the region search variable, n is the total number of regions, i∈N, 1≤i≤n;

[0114] Based on the merchants' ideas, we implement sales forecasts, prepare stocks based on the merchants' own capabilities and their supporting supply capabilities, and help them complete key deployments.

[0115] Step 4: Obtain the cost data of commodity inventory, optimize the total inventory data through Lagrange number multiplication, obtain the optimal inventory quantity and feedback it to the merchant.

[0116] The step 4 includes the following contents:

[0117] Obtain the cost data of a single commodity inventory, including purchase cost, holding cost, and out-of-stock cost, and optimize the total inventory by Lagrange number multiplication. The formula is:

[0118]

[0119] in, is the Lagrangian function, C h is the unit inventory holding cost, C o is the unit out-of-stock cost, C p is the unit purchase cost, S is the out-of-stock quantity, Q is the purchase quantity, and the purchase quantity value is the average of the last sales forecast values ​​of all regions in this fixed time period;

[0120] Set the objective function as:

[0121] z=C h II+C o ·S+C p Q

[0122] Among them, Z is the total cost;

[0123] Set constraints:

[0124]

[0125] Among them, Z SL is the standard normal distribution coefficient of the service level index that the merchant expects to achieve, σ D is the standard deviation of demand, σ L is the standard deviation of the supply cycle, T is the replenishment cycle, and L is the average supply cycle;

[0126] Solve the Lagrangian function to obtain the optimal inventory I * And feed back the optimal inventory data to the merchant.

[0127] Based on the cost of the merchant’s inventory and from the merchant’s perspective, we strive to achieve the best service status at the lowest cost to avoid property losses caused by the blind pursuit of service.

[0128] See also Figure 2 The present invention also discloses an e-commerce transaction system, which is used to execute the above-mentioned e-commerce transaction method, including:

[0129] The information collection and feature selection module is used to obtain commodity sales data and extract features, pre-process the features and aggregate them into a feature matrix, identify the eigenvalues ​​of each feature in the feature matrix by constructing a covariance matrix, sort them according to the eigenvalue values ​​and perform feature screening to form an optimized feature matrix;

[0130] Model training module, used to input the optimized feature matrix into the long short-term memory network model to build a sales forecasting model;

[0131] The inventory statistics module is used to form a sales intention feature set based on the merchant's sales intention, where the sales intention feature is a feature that passes feature screening, input the sales intention feature set into the sales forecast model, obtain the sales forecast matrix, and obtain the inventory of each time period and each region based on the sales forecast matrix and the merchant's supply capacity, thereby obtaining the total inventory of all regions;

[0132] The cost optimization module is used to obtain the cost data of commodity inventory, optimize the total inventory data through Lagrange number multiplication, obtain the optimal inventory quantity and feed it back to the merchant.

[0133] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0134] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0135] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0136] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. An electronic commerce transaction method, characterized in that: The specific steps include: Step 1: Obtain product sales data and extract features, pre-process the features and aggregate them into a feature matrix, identify the eigenvalues ​​of each feature in the feature matrix by constructing a covariance matrix, sort them according to the eigenvalue values ​​and perform feature screening to form an optimized feature matrix; Step 2: Input the optimized feature matrix into the long short-term memory network model to build a sales forecasting model; Step 3: A sales intention feature set is formed based on the merchant's sales intention, where the sales intention feature is the feature that passes the feature screening. The sales intention feature set is input into the sales forecasting model to obtain the sales forecasting matrix. The inventory in each time period and each region is obtained based on the sales forecasting matrix and the merchant's supply capacity, thereby obtaining the total inventory in all regions. Step 4: Obtain the cost data of commodity inventory, optimize the total inventory data through Lagrange number multiplication, obtain the optimal inventory quantity and feedback it to the merchant.

2. An electronic commerce transaction method according to claim 1, characterized in that: Obtain product sales data, which includes product sales, sales timestamps of each product, and destinations to which it is sold. Count sales in each region in each time period to form a data set. Clean the data set and fill in missing data through interpolation. The formula is as follows: Among them, x filled is the filled data point, x i-1 and x i+1 are the data before and after the missing value, respectively, i is the data retrieval variable, i∈N, 1≤i≤n; The commodity sales data is subjected to feature extraction, and the time features and geographic features in the data are extracted. The time features include seasonal features and working day features, and the geographic features include climate features, regional economic features, and regional population features. The geographic features are collected from the Internet based on the commodity sales timestamp and the destination. The features are normalized to form a feature matrix based on the following formula: Where: ∑X is the covariance matrix, n is the number of samples in the feature, X standard is the feature matrix, is the transpose of the feature matrix; Solve the covariance matrix to obtain the eigenvalues ​​and the corresponding eigenvectors according to the following formula: ∑Xv j =λ j v j Where: ∑X is the covariance matrix, v j is the jth eigenvector, λ j is the jth eigenvalue, j is the feature retrieval variable, j∈N, 1≤j≤n; Sort the eigenvectors according to the eigenvalue size, select and retain the top k eigenvectors, and the formula for obtaining the k value is as follows: Among them, λ u is the eigenvalue ranked u, k is the number of selected features, m is the number of original features, j is the eigenvalue retrieval variable, u∈N, 1≤u≤k, j∈N, 1≤j≤n, k≤m; The eigenvectors corresponding to the first k eigenvalues ​​obtained are retained, the eigenvectors are sorted and numbered according to the size of the eigenvalues, and the eigenvectors corresponding to the eigenvalues ​​ranked after the kth are discarded to form an optimized feature matrix.

3. An electronic commerce transaction method according to claim 2, characterized in that: 80% of the optimized feature matrix is ​​divided into a training set and 20% is divided into a test set, and training is performed in a long short-term memory network (LSTM) according to the following formula: in, To predict sales, v u is the eigenvector corresponding to the top k eigenvalues, α u For each feature vector, the model coefficient is corresponding to b, and the intercept is randomly selected from the optimized feature matrix. The samples are input into the long short-term memory network (LSTM) to obtain the initial values ​​of each model coefficient and intercept. u is the retrieval variable of the feature vector, u∈N, 1≤u≤k. The mean square error is used as the loss function, and the formula is as follows: Among them, MSE is the loss function value, y i For actual sales, To predict sales, n is the number of samples, i is the sample retrieval variable, i∈N, 1≤i≤n; Adjust the parameters according to the overall loss function MSE. Minimizing the loss function MSE is the optimal parameter configuration, and the model that passes the test is calibrated as the sales forecasting model.

4. An electronic commerce transaction method according to claim 3, characterized in that: According to the merchant's sales intention, the time period and sales area data that the merchant wants to focus on are obtained, and the time and geographical features are extracted and screened according to the optimized feature matrix to form a sales intention feature set, which is input into the sales forecasting model to obtain the predicted sales volume. The predicted sales volume is the sales volume of the product in each region in each time period to form a predicted sales volume matrix; Obtain the average supply cycle, replenishment cycle data and the expected service level index from the merchants to obtain the standard deviation of the supply cycle. The formula is as follows: Among them, σ L is the standard deviation of the supply cycle, M is the number of times the merchant supplies goods, and L j is the actual delivery period for the jth delivery, is the average value of the supply cycle, j is the retrieval variable of the supply cycle, j∈N, 1≤j≤M; Get the safety stock of inventory in a single area based on the following formula: Among them, SS is the safety stock, Z is the standard normal distribution coefficient corresponding to the service level index that the merchant expects to achieve, σ D is the standard deviation of sales forecast in a certain area, T is the replenishment cycle, σ L is the standard deviation of the supply cycle, L is the average supply cycle; Generate the recommended total inventory in a single region within a fixed time period based on the following formula: Where I is the recommended total inventory, Sp-i is the predicted sales volume in the i-th time period, β is the demand growth rate, T is the replenishment cycle, SS is the safety stock, i∈N, 1≤i≤n; The demand growth rate is based on the following formula: Among them, S p-i is the predicted sales volume in the i-th time period, i is the search variable of the time period, i∈N, 1≤i≤n-1 The total inventory of all regions is obtained according to the following formula: Among them, I i is the total recommended inventory of the i-th region, i is the region retrieval variable, n is the total number of regions, i∈N.1≤i≤n.

5. An electronic commerce transaction method according to claim 4, characterized in that: Obtain the cost data of a single commodity inventory, including purchase cost, holding cost, and out-of-stock cost, and optimize the total inventory by Lagrange number multiplication. The formula is: in, is the Lagrangian function, C h is the unit inventory holding cost, C o is the unit out-of-stock cost, C p is the unit purchase cost, S is the out-of-stock quantity, Q is the purchase quantity, and the purchase quantity value is the average of the last sales forecast values ​​of all regions in this fixed time period; Set the objective function as: Z=C h ·II+C o ·S+C p ·Q Among them, Z is the total cost; Set constraints: Among them, Z SL is the standard normal distribution coefficient of the service level index that the merchant expects to achieve, σ D is the standard deviation of demand, σ L is the standard deviation of the supply cycle, T is the replenishment cycle, and L is the average supply cycle; Solve the Lagrangian function to obtain the optimal inventory I * And feed back the optimal inventory data to the merchant.

6. An electronic commerce transaction system, characterized in that: The transaction system is used to execute an e-commerce transaction method according to any one of claims 1 to 5, comprising: The information collection and feature selection module is used to obtain commodity sales data and extract features, pre-process the features and aggregate them into a feature matrix, identify the eigenvalues ​​of each feature in the feature matrix by constructing a covariance matrix, sort them according to the eigenvalue values ​​and perform feature screening to form an optimized feature matrix; Model training module, used to input the optimized feature matrix into the long short-term memory network model to build a sales forecasting model; The inventory statistics module is used to form a sales intention feature set based on the merchant's sales intention, where the sales intention feature is a feature that passes feature screening, input the sales intention feature set into the sales forecast model, obtain the sales forecast matrix, and obtain the inventory of each time period and each region based on the sales forecast matrix and the merchant's supply capacity, thereby obtaining the total inventory of all regions; The cost optimization module is used to obtain the cost data of commodity inventory, optimize the total inventory data through Lagrange number multiplication, obtain the optimal inventory quantity and feed it back to the merchant.

Images