Chemical and plastic e-commerce management platform and management method based on data analysis

Abstract

The invention discloses a chemical and plastic e-commerce management platform and management method based on data analysis, and relates to the technical field of e-commerce management. The method comprises the following steps: acquiring multi-dimensional data of a plasticized product and forming a standardized data set; analyzing the standardized data set through an LSTM-GARCH prediction model, and outputting predicted market data, including: analyzing conventional data through an LSTM model, and obtaining a long-term change rule of a plasticizing product market; analyzing the fluctuation data through a GARCH model, and judging whether there is an emergency influence or not; and outputting predicted market data according to an analysis result and generating a supply chain management decision. According to the invention, the LSTM-GARCH model is applied, and the long-term law of the plastic market and the emergency response mechanism are deeply fused, so that the intelligence and the dynamics of the plastic supply chain prediction system are realized, and the system can capture market changes in real time, optimize foreign trade management strategies and enhance the international competitiveness of enterprises.

Description

Technical Field

[0001] The present invention belongs to the technical field of e-commerce management, and in particular relates to a chemical and plastic e-commerce management platform and management method based on data analysis. Background Art

[0002] The chemical and plastics product supply chain faces the challenges of volatile market demand and difficult forecasting, which leads to difficulties in production planning and inventory management, and inefficient supply chains. Traditional forecasting methods struggle to accurately grasp market changes, especially for data types with volatile fluctuations, resulting in insufficient forecast accuracy. Furthermore, the characteristics of different types of data vary greatly, making it difficult for a single model to address both. Furthermore, supply chain decisions involve multiple links, and effectively translating forecast results into specific production and inventory strategies is a major challenge. The lack of interpretability in forecast results makes it difficult for decision makers to formulate reasonable supply chain adjustment plans. Given the large regional disparities, demand and cost distribution vary across regions, and formulating targeted regional strategies is also a thorny issue.

[0003] These factors combined have led to the complexity of optimizing the supply chain of chemical and plastic products. There is an urgent need for a supply chain optimization method that can comprehensively consider multi-dimensional data, take into account the characteristics of different types of data, provide explainable prediction results and support regional decision-making, so as to cope with market fluctuations and improve the overall operational efficiency of the supply chain. Summary of the Invention

[0004] The purpose of the present invention is to provide a chemical and plastic e-commerce management platform and management method based on data analysis, which can be achieved through the following technical solutions: In a first aspect, an embodiment of the present application provides a method for managing a plastics e-commerce business based on data analysis, comprising the following steps: Obtain multi-dimensional data on chemical and plastic products on e-commerce platforms, including regular data reflecting market trends and fluctuation data reflecting unexpected events; Preprocessing the multidimensional data to form a standardized data set; The standardized data set is analyzed using a pre-trained LSTM-GARCH forecasting model to output predicted market data, specifically including: Inputting the standardized data set into the LSTM-GARCH prediction model; the LSTM-GARCH model includes a GARCH model and an LSTM model; Analyze the conventional data by using the LSTM model in the LSTM-GARCH prediction model to obtain the long-term change pattern of the chemical and plastic product market; Analyze the fluctuation data by the GARCH model in the LSTM-GARCH prediction model to determine whether there is an impact of an emergency; Outputting the predicted market data according to the analysis results; generating supply chain management decisions based on the predicted market data; The LSTM model has the function of capturing long-term trends; the GARCH model has the function of quantifying volatility risks; The predicted market data is expressed as: y t = LSTM( )+ ·GARCH( ); Represents the weight of the LSTM model; represents the weight of the GARCH model and is 0 when there is no abnormal fluctuation in the fluctuation data; Represents the prediction result output by the LSTM model; Represents the forecast results output by the GARCH model.

[0005] Preferably, the obtaining of multi-dimensional data of plastic products on the e-commerce platform includes: Collect the multi-dimensional data through an e-commerce program interface; Classifying the multi-dimensional data into regular data and fluctuating data, wherein the regular data includes sales data of chemical and plastic products, raw material demand data, and inventory change data; and the fluctuating data includes crude oil price data, exchange rate data, and tariff data; The regular data and the fluctuation data are stored separately for subsequent retrieval.

[0006] Preferably, the preprocessing of the multi-dimensional data to form a standardized data set includes: Performing data cleaning on the multi-dimensional data to remove outliers; Filling missing values in the multidimensional data; Performing a unit standardization operation on the multi-dimensional data to unify the data format; The standardized dataset is generated and stored for calling by the LSTM-GARCH prediction model, ensuring the consistency of the standardized dataset in time zone and unit.

[0007] Preferably, analyzing the standardized data set by a pre-trained LSTM-GARCH prediction model comprises: If no abnormal fluctuation is detected in the fluctuation data, predicting the normal data by using the LSTM model and outputting predicted second normal data; If abnormal fluctuations are detected in the fluctuation data, the normal data are first predicted using the GARCH model to obtain first normal data, and the first normal data and the second normal data are then calibrated to obtain predicted third normal data; The final predicted market data is generated according to the third regular data or the second regular data.

[0008] Preferably, analyzing the fluctuation data by the GARCH model in the LSTM-GARCH prediction model includes: Obtaining a current data fluctuation trend when abnormal fluctuations occur in the fluctuation data; Acquire similar emergencies from historical emergencies whose similarity to the current data fluctuation trend exceeds a preset threshold; Extracting the characteristic values of the fluctuation data of the abnormal fluctuation and the characteristic values of the historical fluctuation data corresponding to the similar emergency respectively, and matching the two types of extracted characteristic values. If the two characteristics match, obtaining the benchmark parameters of the similar emergency; Based on the benchmark parameters, the obtained normal data is input into the GARCH model to obtain the volatility of the normal data under the influence of the fluctuating data with abnormal fluctuations, and the predicted first normal data; The risk level of the current emergency event is determined based on the volatility.

[0009] Preferably, the first conventional data is expressed as: ;in, Indicates the mean corresponding to the normal data; represents the residual term.

[0010] Preferably, the volatility of the current regular data is expressed as: ; in, represents a constant term; represents the square of the previous residual; represents the prior conditional variance; represents the ARCH term coefficient; is the GARCH term coefficient; represents the lag order of the squared residual; represents the lag order of the conditional variance.

[0011] Preferably, in the LSTM-GARCH prediction model, the GARCH model is trained using transfer learning, including: Obtain historical transaction data for chemical and plastic products and screen for emergencies. Use data from events where crude oil prices, exchange rates, and tariffs fluctuate abnormally as training data samples, annotate emergencies associated with fluctuations, and establish an emergencies database. Extract features from historical fluctuation data of different types of emergencies, including the number of extreme points, volatility, and trend slope; Pre-train the GARCH model based on the fluctuation data characteristics of different types of emergencies to obtain benchmark parameters; When abnormal fluctuations are detected in the fluctuation data, similar emergencies are matched in real time through the benchmark parameters and parameter migration is performed; Result prediction is performed according to the GARCH model, the prediction result is calibrated and fitted using the historical transaction data, and the benchmark parameters are further optimized until the training of the GARCH model is completed; and the emergency event database is updated using the detection of new emergency events.

[0012] Preferably, the second conventional data is expressed as: ; in, Represents general data at the current moment; Indicates general data of the previous moment.

[0013] In a second aspect, the embodiments of the present application provide a chemical and plastic e-commerce management platform based on data analysis, including: Data collection module: used to obtain multi-dimensional data of chemical and plastic products on e-commerce platforms, including regular data reflecting market trends and fluctuation data reflecting unexpected events; Data processing module: used for preprocessing the multi-dimensional data to form a standardized data set; Data analysis module: used to analyze the standardized data set through a pre-trained LSTM-GARCH prediction model and output predicted market data, specifically including: inputting the standardized data set into the LSTM-GARCH prediction model; the LSTM-GARCH model includes a GARCH model and an LSTM model; analyzing the conventional data through the LSTM model in the LSTM-GARCH prediction model to obtain the long-term change pattern of the chemical plastic product market; analyzing the fluctuation data through the GARCH model in the LSTM-GARCH prediction model to determine whether there is an impact of an emergency; and outputting the predicted market data based on the analysis results; Decision generation module: generates supply chain management decisions based on the predicted market data.

[0014] The beneficial effects of the present invention are: (1) By applying the LSTM-GARCH model, the present invention deeply integrates the long-term laws of the chemical and plastics market with the emergency response mechanism, realizing the intelligence and dynamism of the chemical and plastics supply chain forecasting system, enabling the system to capture market changes in real time, optimize foreign trade management strategies, and enhance the international competitiveness of enterprises.

[0015] (2) The present invention obtains multi-dimensional historical data and pre-processes it, and uses the generalized autoregressive conditional heteroskedasticity model (GARCH model) and the long short-term memory network model (LSTM model) of sample transfer learning to analyze and predict the fluctuation type data and the regular type data, and then obtains the comprehensive prediction data through weighted fitting. Based on the prediction results, sales, demand and inventory forecasts are generated, the market fluctuation trend is analyzed and the direction of supply chain adjustment is determined, and then production plans and inventory management recommendations are formulated. Finally, the prediction data and optimization strategy are displayed through a visual interface, and a spatiotemporal heat map is generated to determine the regional demand and cost distribution, providing support for supply chain decision-making. The present invention can effectively respond to market fluctuations, optimize production planning and inventory management, and improve supply chain operation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] For better understanding and implementation, the technical solution of the present application is described in detail below with reference to the accompanying drawings.

[0017] Figure 1 A flowchart of a method for managing a plastics e-commerce business based on data analysis provided in an embodiment of the present application; Figure 2 A flowchart of the steps for analyzing fluctuation data using a GARCH model provided in an embodiment of the present application; Figure 3 A flowchart of the steps for training a GARCH model using transfer learning provided in an embodiment of the present application; Figure 4 A structural diagram of a chemical and plastic e-commerce management platform based on data analysis provided in an embodiment of the present application. DETAILED DESCRIPTION

[0018] To further illustrate the technical means and effectiveness of the present invention in achieving its intended purpose, exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all possible embodiments consistent with the present application. Rather, they are merely examples of methods and systems consistent with certain aspects of the present application, as detailed in the appended claims.

[0019] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The singular forms "a," "an," "the," and "the" used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to any or all possible combinations of one or more of the associated listed items.

[0020] The following describes in detail the specific implementation methods, features and effects of the present invention in conjunction with the accompanying drawings and preferred embodiments.

[0021] Example 1 See also Figure 1 The present application provides a method for managing a plastics e-commerce business based on data analysis, comprising the following steps: S101, obtaining multi-dimensional data of chemical and plastic products on an e-commerce platform, wherein the multi-dimensional data includes regular data reflecting market trends and fluctuation data reflecting unexpected events; S102, preprocessing the multi-dimensional data to form a standardized data set; S103, analyzing the standardized data set using a pre-trained LSTM-GARCH prediction model and outputting predicted market data, specifically including: Inputting the standardized data set into the LSTM-GARCH prediction model; the LSTM-GARCH model includes a GARCH model and an LSTM model; Analyze the conventional data by using the LSTM model in the LSTM-GARCH prediction model to obtain the long-term change pattern of the chemical and plastic product market; Analyze the fluctuation data by the GARCH model in the LSTM-GARCH prediction model to determine whether there is an impact of an emergency; Outputting the predicted market data according to the analysis results; S104, generating a supply chain management decision based on the predicted market data; The LSTM model has the function of capturing long-term trends; the GARCH model has the function of quantifying volatility risks; The predicted market data is expressed as: y t = LSTM( )+ ·GARCH( ); Represents the weight of the LSTM model; represents the weight of the GARCH model and is 0 when there is no abnormal fluctuation in the fluctuation data; Represents the prediction result output by the LSTM model; Represents the forecast results output by the GARCH model.

[0022] Specifically, in the present application, since the chemical and plastic product market is easily affected by external events such as crude oil prices, exchange rate changes, and tariffs, in order to better conduct dynamic analysis of chemical and plastic product market data, this application adopts the LSTM (primary)-GARCH (secondary) model for data prediction and analysis. The LSTM model is suitable for processing relationships in long sequence data and can be used to capture the long-term trend patterns of cross-border trade in chemical and plastic products, but it cannot quantify the impact of emergencies. Therefore, it is used for chemical and plastic market data analysis in the absence of emergencies; the GARCH model performs robustly in volatility prediction tasks and can be used to quantify sudden volatility risks, that is, data fluctuations caused by emergencies. The present application creatively integrates the LSTM model and the GARCH model, and introduces the GARCH model on the basis of the LSTM model prediction for weighted fitting when an emergency occurs. Based on this, the present application defines two types of data: volatility data and regular data. Fluctuating data refers to data that reflects the occurrence of unexpected events that impact the chemical and plastics market, such as tariff adjustments or financial crises, such as crude oil price data, exchange rate data, and tariff data. Specifically, when an unexpected event occurs, data such as crude oil price data, exchange rate data, and tariff data will fluctuate significantly. Conventional data, on the other hand, refers to market data generated during cross-border trade of chemical and plastics products, such as sales data, raw material demand data, and inventory change data. Fluctuating data is used to detect the occurrence of unexpected events, while conventional data is used to reflect market changes and analyze market trends. This application uses an LSTM model to predict and analyze market data when there are no unexpected events, predicting and outputting future conventional data (the GARCH model weight is 0 at this time). When an unexpected event is detected, a GARCH model is introduced to fit and correct the output of the LSTM model, thereby generating more accurate forecasts, optimizing foreign trade management strategies, and assisting enterprise decision-makers in supply chain optimization, thereby accurately grasping changes in the chemical and plastics e-commerce foreign trade market.

[0023] It should be noted that the LSTM (Long Short-Term Memory) model used in this application can capture both long-term and short-term dependencies and is suitable for processing complex nonlinear time series data, especially when the data exhibits seasonality, trends, and cyclical characteristics. The GARCH (Generalized Autoregressive Conditional Heteroskedasticity) model, on the other hand, captures the temporal variation in volatility in a time series by modeling conditional heteroskedasticity. It is primarily used to model time series volatility, particularly in the financial sector, to help analyze and predict market volatility (such as in the stock market and exchange rates).

[0024] It can be understood that in this application, long-term time series data includes sales data of chemical and plastic products, raw material demand data, inventory change data, and customer behavior data; while short-term time series data includes crude oil price data, exchange rate data, and tariff data, etc.

[0025] In step S101, big data is collected using an e-commerce application programming interface (API) to quickly and efficiently obtain multi-dimensional data from a specific platform. Because the collected data may contain inconsistent time zones and units, missing values, and outliers, data preprocessing is performed through data cleaning, outlier processing, missing value filling, and unit standardization.

[0026] In one embodiment, obtaining multi-dimensional data of plastic products on an e-commerce platform includes: Collect the multi-dimensional data through an e-commerce program interface; Classifying the multi-dimensional data into regular data and fluctuating data, wherein the regular data includes sales data of chemical and plastic products, raw material demand data, and inventory change data; and the fluctuating data includes crude oil price data, exchange rate data, and tariff data; The regular data and the fluctuation data are stored separately for subsequent retrieval.

[0027] In one embodiment provided in the present application, preprocessing the multi-dimensional data to form a standardized data set includes: Performing data cleaning on the multi-dimensional data to remove outliers; Filling missing values in the multidimensional data; Performing a unit standardization operation on the multi-dimensional data to unify the data format; The standardized dataset is generated and stored for calling by the LSTM-GARCH prediction model, ensuring the consistency of the standardized dataset in time zone and unit.

[0028] In one embodiment provided in the present application, the standardized data set is analyzed by a pre-trained LSTM-GARCH prediction model, including: If no abnormal fluctuation is detected in the fluctuation data, predicting the normal data by using the LSTM model and outputting predicted second normal data; If abnormal fluctuations are detected in the fluctuation data, the normal data are first predicted using the GARCH model to obtain first normal data, and the first normal data and the second normal data are then calibrated to obtain predicted third normal data; The final predicted market data is generated according to the third regular data or the second regular data.

[0029] like Figure 2 As shown, in one embodiment provided in the present application, analyzing the fluctuation data by the GARCH model in the LSTM-GARCH prediction model includes: Obtaining a current data fluctuation trend when abnormal fluctuations occur in the fluctuation data; Acquire similar emergencies from historical emergencies whose similarity to the current data fluctuation trend exceeds a preset threshold; Extracting the characteristic values of the fluctuation data of the abnormal fluctuation and the characteristic values of the historical fluctuation data corresponding to the similar emergency respectively, and matching the two types of extracted characteristic values. If the two characteristics match, obtaining the benchmark parameters of the similar emergency; Based on the benchmark parameters, the obtained normal data is input into the GARCH model to obtain the volatility of the normal data under the influence of the fluctuating data with abnormal fluctuations, and the predicted first normal data; The risk level of the current emergency event is determined based on the volatility.

[0030] like Figure 3 As shown, in one embodiment provided in the present application, in the LSTM-GARCH prediction model, the GARCH model is trained using transfer learning, including: Obtain historical transaction data for chemical and plastic products and screen for emergencies. Use data from events where crude oil prices, exchange rates, and tariffs fluctuate abnormally as training data samples, annotate emergencies associated with fluctuations, and establish an emergencies database. Extract features from historical fluctuation data of different types of emergencies, including the number of extreme points, volatility, and trend slope; Pre-train the GARCH model based on the fluctuation data characteristics of different types of emergencies to obtain a benchmark parameter set; When abnormal fluctuations are detected in the fluctuation data, similar emergencies are matched in real time through the benchmark parameter set and parameter migration is performed; Result prediction is performed according to the GARCH model, the prediction result is calibrated and fitted using the historical transaction data, and the benchmark parameters are further optimized until the training of the GARCH model is completed; and the emergency event database is updated using the detection of new emergency events.

[0031] Specifically, the present application collects conventional type data and fluctuation type data of cross-border trade of chemical and plastic products through the e-commerce program interface to obtain a multi-dimensional original data set. The original data set is subjected to data cleaning, outlier processing, missing value filling and unit standardization to obtain a preprocessed data set. Fluctuation type data, including crude oil price data, exchange rate data and tariff data, are extracted from the preprocessed data set to obtain a fluctuation type data set. If it is detected that the fluctuation type data has a single-day fluctuation of more than 5% low, 10% medium, or 20% high, the risk level is judged according to the degree of fluctuation, and the characteristic value and risk level of the fluctuation event are obtained. According to the characteristic value of the fluctuation event, the fluctuation type data of similar historical events are matched, the corresponding historical benchmark parameters are obtained, and the benchmark parameter set required for transfer learning is obtained. The sample transfer learning method is adopted, and the benchmark parameter set is used to train the generalized autoregressive conditional heteroskedasticity model (GARCH model) to obtain a trained fluctuation prediction model. Through the trained fluctuation prediction model, the conventional type data and fluctuation type data of the current market are input to predict the impact of the fluctuation type data on the conventional type data, and the fluctuation impact prediction result is obtained. Based on the volatility impact forecast results and the calibration fit of historical data, we optimized the baseline parameters of the generalized autoregressive conditional heteroskedasticity model to obtain an optimized parameter set. Using this optimized parameter set, we re-ran the generalized autoregressive conditional heteroskedasticity model to predict the trend of conventional data in the future period and obtain the final forecast results.

[0032] like Figure 3 As shown, in one embodiment provided in the present application, regular type data is extracted from the preprocessed data set, and a long short-term memory network model is used for training to obtain an LSTM model. The long-term trend patterns of regular type data in the period without fluctuation events are captured, and the regular trend prediction results are obtained by inputting the current market data for prediction.

[0033] Specifically, regular data from periods without volatility events are extracted from the preprocessed dataset to obtain training data samples. This training data sample is then trained using a long short-term memory (LSTM) model to capture the trend patterns of regular data during periods without volatility events. Current market regular data is input based on the captured trend patterns, resulting in the LSTM model outputting predicted regular data. Historical records of volatility data are obtained, and events with single-day fluctuations exceeding 5% are detected to determine whether there is a volatility impact. If a volatility impact is detected, the GARCH model is used to predict the volatility variance and output the adjusted regular data. If no volatility impact is detected, the predicted regular data output by the LSTM model is directly used. A comprehensive forecast is obtained by fitting the regular data output by the LSTM model with the regular data output by the GARCH model.

[0034] In another embodiment provided in the present application, weighted fitting is performed on the conventional trend prediction results and the fluctuation impact prediction results, and a preset weight allocation rule is adopted. If an abnormal change in the fluctuation type data is detected, the weight of the GARCH model is increased, otherwise the weight is allocated to the LSTM model to obtain comprehensive prediction data.

[0035] Volatility data, including crude oil prices, exchange rates, and tariff data, is obtained. Data cleaning is performed to address time zone inconsistencies, non-standardized units, missing values, and outliers, resulting in a standardized volatility dataset. Based on this standardized volatility dataset, daily fluctuations are calculated. If the crude oil price fluctuations exceed 5%, 10%, or 20%, or if the exchange rate or tariff data meet corresponding thresholds, the data is classified as low, medium, or high volatility, and the volatility level is determined. Based on the volatility level, historical scenarios with similar volatility characteristics are matched to the historical event database to obtain the baseline parameters for the corresponding historical events and generate the parameter set for the GARCH model. This parameter set is then used to input current market data into a pre-trained GARCH model. The mean and residual terms of the data under the influence of volatility are predicted to obtain the volatility impact prediction results. Regular data on chemical and plastic products from the past 5-10 years, during periods without volatility events, is obtained to train an LSTM model to capture trends in regular cross-border trade data. The trained model is then completed. The trained LSTM model is fed with regular data from the current and previous moments to predict regular data in the absence of volatility events, yielding a regular trend forecast. Based on whether the volatility data has experienced anomalies, if so, the GARCH model weight is assigned a preset high value and the LSTM model weight is assigned a low value. Otherwise, the LSTM model weight is assigned a 1 and the GARCH model weight is assigned a 0. A weighting rule is then determined. Using this weighting rule, a weighted fit is performed on the LSTM model's regular trend forecast and the GARCH model's volatility impact forecast to produce a composite forecast. This composite forecast is then calibrated and fitted using historical data to optimize the baseline parameters of the GARCH and LSTM models, resulting in updated model parameters.

[0036] In an embodiment provided in this application, the first conventional data is represented as: ;in, Indicates the mean value corresponding to the current regular data; represents the residual term.

[0037] In one embodiment provided in this application, the volatility of the current regular data is expressed as: ; in, represents a constant term; represents the square of the previous residual; represents the prior conditional variance; represents the ARCH term coefficient; is the GARCH term coefficient; represents the lag order of the squared residual; represents the lag order of the conditional variance.

[0038] Specifically, this embodiment uses sample transfer learning to train a generalized autoregressive conditional heteroskedasticity model (GARCH model). Based on historical data, this model analyzes fluctuations in crude oil prices, exchange rates, and tariffs when unexpected events cause fluctuations in volatile data. For example, a single-day fluctuation in crude oil prices exceeding 5% (low), 10% (medium), or 20% (high) is detected, with different values for each category. The risk level is determined based on the degree of fluctuation, and regular data for future forecasts is output.

[0039] Specifically, when crude oil prices, exchange rates, and tariffs fluctuate, the system predicts the impact of these fluctuations on regular data. When a change in fluctuation data is detected, the system matches the characteristic values of the fluctuation data with historical emergencies with similar fluctuations. Based on the benchmark parameters of these historical emergencies, the system inputs regular market data, predicts the volatility variance of this event, determines the risk level, and outputs the changing trend of regular data under the influence of the event.

[0040] In short, this application uses the GARCH model to determine whether the current emergency event is similar to the emergency events that have occurred in historical data. If the two are similar, the baseline parameters of the emergency events in the historical data are used as the baseline parameters of the current emergency event to predict the risk level of this emergency event.

[0041] In one embodiment provided in the present application, the long short-term memory network is used to train the LSTM model, including: Perform a screening of no emergencies based on the historical transaction data, and use the historical regular data of chemical and plastic products when no emergencies occur as training data samples; The LSTM model is trained based on training data samples to capture the trend patterns of regular data of chemical and plastic products when there are no emergencies, and an optimized LSTM model is obtained.

[0042] In an embodiment provided in this application, the second regular data is expressed as: ;in, Represents general data at the current moment; Indicates general data of the previous moment.

[0043] Specifically, this embodiment adopts a long short-term memory network (LSTM) model, trains the LSTM model based on historical regular data, captures the trend patterns of regular types of cross-border trade data of chemical and plastic products when there are no fluctuation events, and outputs predicted regular data based on the input of regular data of the recent market.

[0044] Specifically: Use the regular data of chemical and plastic products in the period without fluctuation events in historical data as training data samples to train the LSTM model, capture the trend patterns of regular type data of cross-border trade of chemical and plastic products when there are no fluctuation events, input the current regular data of the current dynamic market, and output the predicted regular data based on the trend patterns captured during training.

[0045] In another embodiment of the present application, sales data forecasts, raw material demand data forecasts, and inventory change data forecasts for chemical and plastic products are generated based on the comprehensive forecast data to obtain a forecast dataset. By analyzing the sales data, raw material demand data, and inventory change data in the forecast dataset, market demand fluctuation trends are calculated and supply chain adjustment directions are determined.

[0046] It also proposes production plans by analyzing the sales data, raw material demand data, and inventory change data in the forecast data set, such as: if sales are good and inventory is sufficient, then increase production; if sales are poor, then reduce production; and inventory management suggestions, such as: if raw material supply is sufficient and inventory is low, then replenish inventory. It also provides a visual interface to generate a spatiotemporal heat map. The map displays regional demand based on the amount of sales data of chemical and plastic products in different regions, and displays material costs based on the amount of raw material demand data to assist corporate decision makers in optimizing the supply chain. Analyze the fluctuation trend of market demand based on the forecast data, determine the adjustment direction of supply chain management, and obtain preliminary decision-making basis. In summary, this application realizes the intelligence and dynamism of the chemical and plastic supply chain forecasting system by applying the LSTM-GARCH model and deeply integrating the long-term laws of the chemical and plastic market with the emergency response mechanism, so that the system can capture market changes in real time, optimize foreign trade management strategies, and enhance the international competitiveness of enterprises.

[0047] Example 2 See also Figure 4 The embodiment of the present application provides a chemical and plastic e-commerce management platform based on data analysis, which applies the chemical and plastic e-commerce management method described above, including: Data collection module: used to collect historical transaction data of plastic products on the e-commerce management platform and perform pre-processing operations on the historical transaction data; Model training module: Based on pre-processed historical transaction data, transfer learning is used to train the GARCH model, and long short-term memory network is used to train the LSTM model; Data analysis module: uses the trained GARCH model to perform data analysis, specifically: analyzes the risk level when an emergency occurs and outputs the predicted first regular data; uses the trained LSTM model to perform data analysis, specifically: obtains the current regular data of the current dynamic market and inputs it into the LSTM model, and outputs the predicted second regular data; Data fitting module: used for fitting the first regular data and the second regular data to obtain an LSTM-GARCH model, and outputting predicted third regular data; Model prediction module: Generates supply chain prediction results for chemical and plastic products based on the third conventional data predicted by the LSTM-GARCH model; based on the supply chain prediction results, generates a spatiotemporal heat map using a visual interface to provide auxiliary decision-making for supply chain optimization.

[0048] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0049] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0050] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0051] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as a preferred embodiment as above, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to equivalent embodiments using the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A chemical and plastic e-commerce management method based on data analysis, characterized by: The steps include: Obtain multi-dimensional data on chemical and plastic products on e-commerce platforms, including regular data reflecting market trends and fluctuation data reflecting unexpected events; Preprocessing the multidimensional data to form a standardized data set; The standardized data set is analyzed using a pre-trained LSTM-GARCH forecasting model to output predicted market data, specifically including: Inputting the standardized data set into the LSTM-GARCH prediction model; the LSTM-GARCH model includes a GARCH model and an LSTM model; Analyze the conventional data by using the LSTM model in the LSTM-GARCH prediction model to obtain the long-term change pattern of the chemical and plastic product market; Analyze the fluctuation data by the GARCH model in the LSTM-GARCH prediction model to determine whether there is an impact of an emergency; Outputting the predicted market data according to the analysis results; generating supply chain management decisions based on the predicted market data; The LSTM model has the function of capturing long-term trends; the GARCH model has the function of quantifying volatility risks; The predicted market data is expressed as: y t = ·LSTM( )+ ·GARCH( ); Represents the weight of the LSTM model; represents the weight of the GARCH model and is 0 when there is no abnormal fluctuation in the fluctuation data; Represents the prediction result output by the LSTM model; Represents the forecast results output by the GARCH model.

2. The method for managing chemical and plastic e-commerce according to claim 1, characterized in that: The acquisition of multi-dimensional data of chemical plastic products on the e-commerce platform includes: Collect the multi-dimensional data through an e-commerce program interface; Classifying the multi-dimensional data into regular data and fluctuating data, wherein the regular data includes sales data of chemical and plastic products, raw material demand data, and inventory change data; and the fluctuating data includes crude oil price data, exchange rate data, and tariff data; The regular data and the fluctuation data are stored separately for subsequent retrieval.

3. The method for managing chemical and plastic e-commerce according to claim 2, characterized in that: Preprocessing the multi-dimensional data to form a standardized data set includes: Performing data cleaning on the multi-dimensional data to remove outliers; Filling missing values in the multidimensional data; Performing a unit standardization operation on the multi-dimensional data to unify the data format; The standardized dataset is generated and stored for calling by the LSTM-GARCH prediction model, ensuring the consistency of the standardized dataset in time zone and unit.

4. The chemical plastic e-commerce management method according to claim 1, characterized in that: The standardized data set is analyzed by the pre-trained LSTM-GARCH prediction model, comprising: If no abnormal fluctuation is detected in the fluctuation data, predicting the normal data by using the LSTM model and outputting predicted second normal data; If abnormal fluctuations are detected in the fluctuation data, the normal data are first predicted using the GARCH model to obtain first normal data, and the first normal data and the second normal data are then calibrated to obtain predicted third normal data; The final predicted market data is generated according to the third regular data or the second regular data.

5. The chemical plastic e-commerce management method according to claim 1, characterized in that: Analyzing the fluctuation data by the GARCH model in the LSTM-GARCH prediction model includes: Obtaining a current data fluctuation trend when abnormal fluctuations occur in the fluctuation data; Acquire similar emergencies from historical emergencies whose similarity to the current data fluctuation trend exceeds a preset threshold; Extracting the characteristic values of the fluctuation data of the abnormal fluctuation and the characteristic values of the historical fluctuation data corresponding to the similar emergency respectively, and matching the two types of extracted characteristic values. If the two characteristics match, obtaining the benchmark parameters of the similar emergency; Based on the benchmark parameters, the obtained normal data is input into the GARCH model to obtain the volatility of the normal data under the influence of the fluctuating data with abnormal fluctuations, and the predicted first normal data; The risk level of the current emergency event is determined based on the volatility.

6. The chemical plastic e-commerce management method according to claim 5, characterized in that: The first conventional data is represented as: ;in, Indicates the mean corresponding to the normal data; represents the residual term.

7. The chemical plastic e-commerce management method according to claim 5, characterized in that: The volatility is expressed as: ; in, represents a constant term; represents the square of the previous residual; represents the prior conditional variance; represents the ARCH term coefficient; is the GARCH term coefficient; represents the lag order of the squared residual; represents the lag order of the conditional variance.

8. The method for managing chemical and plastic e-commerce according to claim 1, characterized in that: In the LSTM-GARCH prediction model, transfer learning is used to train the GARCH model, including: Obtain historical transaction data for chemical and plastic products and screen for emergencies. Use data from events where crude oil prices, exchange rates, and tariffs fluctuate abnormally as training data samples, annotate emergencies associated with fluctuations, and establish an emergencies database. Extract features from historical fluctuation data of different types of emergencies, including the number of extreme points, volatility, and trend slope; Pre-train the GARCH model based on the fluctuation data characteristics of different types of emergencies to obtain a benchmark parameter set; When abnormal fluctuations are detected in the fluctuation data, similar emergencies are matched in real time through the benchmark parameter set and parameter migration is performed; Result prediction is performed according to the GARCH model, the prediction result is calibrated and fitted using the historical transaction data, and the benchmark parameters are further optimized until the training of the GARCH model is completed; and the emergency event database is updated using the detection of new emergency events.

9. The chemical plastic e-commerce management method according to claim 4, characterized in that: The second conventional data is expressed as: ; in, Represents general data at the current moment; Indicates general data of the previous moment.

10. A chemical and plastic e-commerce management platform based on data analysis, applying the chemical and plastic e-commerce management method according to any one of claims 1 to 9, characterized in that: include: Data collection module: used to obtain multi-dimensional data of chemical and plastic products on e-commerce platforms, including regular data reflecting market trends and fluctuation data reflecting unexpected events; Data processing module: used for preprocessing the multi-dimensional data to form a standardized data set; Data analysis module: used to analyze the standardized data set through a pre-trained LSTM-GARCH prediction model and output predicted market data, specifically including: inputting the standardized data set into the LSTM-GARCH prediction model; the LSTM-GARCH model includes a GARCH model and an LSTM model; analyzing the conventional data through the LSTM model in the LSTM-GARCH prediction model to obtain the long-term change pattern of the chemical plastic product market; analyzing the fluctuation data through the GARCH model in the LSTM-GARCH prediction model to determine whether there is an impact of an emergency; and outputting the predicted market data based on the analysis results; Decision generation module: generates supply chain management decisions based on the predicted market data.

Images