E-commerce order analysis and control method and system

Abstract

The present invention proposes an e-commerce order analysis and control method and system, the method includes: S1: building a big data platform to collect, clean and store e-commerce data in real time; S2: automatically crawling e-commerce data, integrating and standardizing the e-commerce data; S3: using a cloud data analysis platform to analyze the data in real time; S4: using an order volume prediction model to predict future order volumes in advance, and combining real-time data analysis to achieve active prediction and dynamic adjustment of e-commerce orders. The present invention uses machine learning algorithms to analyze order data, mine user behavior patterns, and predict order trends. Develop data analysis models for order optimization and risk assessment.

Description

Technical Field

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

[0002] With the rapid development of e-commerce, the number of orders has surged, and the traditional order management method can no longer meet the needs of enterprises. In order to improve order processing efficiency, reduce operating costs, and enhance user experience, it is urgent to develop a set of functional and easy-to-operate order management systems. The continuous growth of e-commerce transaction volume and the surge in order volume have put forward an urgent need for an efficient order management system. Enterprises hope to improve order processing efficiency, reduce operating costs, improve user experience, and enhance competitiveness through order management systems. The traditional manual order management method is inefficient, prone to errors, and cannot meet the growing demand for order volume. The high cost of manual management and the difficulty of expansion cannot meet the needs of enterprise development. Summary of the invention

[0003] According to the problems faced by the prior art, the present invention proposes an e-commerce order analysis and control method and system to solve the problem of low order management efficiency, and the technical solutions adopted are as follows:

[0004] An e-commerce order analysis and control method, the method comprising:

[0005] S1: Build a big data platform to collect, clean and store e-commerce data in real time;

[0006] S2: Automatically capture e-commerce data, and integrate and standardize the e-commerce data;

[0007] S3: Use a cloud data analysis platform to perform real-time analysis on the data;

[0008] S4: Use the order volume prediction model to predict future order volumes in advance, and combine it with real-time data analysis to achieve proactive prediction and dynamic adjustment of e-commerce orders.

[0009] Preferably, the S1 includes:

[0010] S11: Build a big data platform based on Hadoop, collect and clean the e-commerce data in real time, and store the data in HDFS to provide raw data for subsequent data warehouses;

[0011] S12: Building a data warehouse, and performing dimensional modeling on the data warehouse, layering data of different granularities, and performing offline analysis and data visualization on the e-commerce data.

[0012] Preferably, S2 includes:

[0013] S21: Realize real-time data synchronization and seamless integration by connecting to the API interface provided by the e-commerce platform, and use intelligent web crawlers to automatically capture e-commerce data on the e-commerce platform;

[0014] S22: The standardized equipment automatically cleans, converts and loads the e-commerce data.

[0015] Preferably, S3 includes:

[0016] S31: Use cloud computing technology to build a cloud data analysis platform and deploy a data storage system on the cloud data analysis platform;

[0017] S32: Configure Spark Streaming on the cloud data analysis platform to process the real-time data stream and store the processed data in the data storage system.

[0018] Preferably, the S4 includes:

[0019] S41: Collecting logistics freight order information through a freight information collection module, and merging the logistics freight order information into a unified data set;

[0020] S42: training the data set using the LSTM model, and dividing the data set into a training set, a validation set, and a test set, and training the model using the training set to find optimal model parameters, tuning the model using the validation set data, and evaluating the prediction performance of the model using the test set data, and calculating an error index;

[0021] S43: During the forecasting process, the production strategy is dynamically adjusted based on the feedback of real-time data and changes in forecast results.

[0022] Preferably, the API interface includes:

[0023] Order related interface, the order related interface is used to obtain the user's order information, the order information includes order status, order time, delivery address and logistics information.

[0024] Preferably, the method of dynamically adjusting the production strategy in S43 is as follows:

[0025] The logistics order information collection system combines LSTM model prediction and dynamic scheduling to obtain comprehensive scheduling efficiency, where the comprehensive scheduling efficiency is obtained through the following formula:

[0026]

[0027] Among them, C represents the comprehensive scheduling efficiency, P represents the prediction error of the LTSM model, Indicates the difference between the real-time collected order volume data and the predicted order volume data. represents the maximum value of the difference between the order quantity data collected historically by the logistics order information collection system and the predicted order quantity data, S represents the scheduling adjustment range, Indicates the maximum adjustment range that can be achieved. represents the minimum value of the adjustment amplitude, and The value of is 0, and the adjustment range depends on the upper limit of production capacity; α represents the forecast error index weight, β represents the adjustment range weight, and α+β=1;

[0028] Furthermore, the prediction error of the LTSM model is obtained by the following formula:

[0029]

[0030] in, represents the actual order volume in the i-th time period, represents the predicted order volume in the i-th time period;

[0031] When 0.9<C≤1, it means that the prediction accuracy is high, the real-time data feedback is highly consistent with the prediction results, the scheduling strategy adjustment is accurate and efficient, and the existing strategy and system configuration should be maintained, and long-term performance should be monitored to confirm stability;

[0032] When 0.5<C≤0.89, it means that the overall performance is moderate. Appropriate improvement measures should be formulated for the poorly performing aspects of P, D, and S. For example, if the prediction accuracy is low, the parameters or training data of the LSTM model may need to be adjusted; if the real-time data feedback varies greatly, the data collection and processing process may need to be improved; if the scheduling strategy adjustment effect is not good, the scheduling algorithm may need to be optimized or the sensitivity to real-time data may need to be increased;

[0033] When 0.1<C≤0.49, it means that the overall performance of the system is poor and cannot meet business needs. An emergency repair plan should be formulated and implemented for specific problems, including quickly adjusting model parameters, optimizing data processing processes, and modifying scheduling strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 This is an e-commerce order analysis and control method described in the present invention. DETAILED DESCRIPTION

[0035] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0036] One embodiment of the present invention, as Figure 1An e-commerce order analysis and control method is shown, the method comprising:

[0037] S1: Build a big data platform to collect, clean and store e-commerce data in real time;

[0038] S2: Automatically capture e-commerce data, and integrate and standardize the e-commerce data;

[0039] S3: Use a cloud data analysis platform to perform real-time analysis on the data;

[0040] S4: Use the order volume prediction model to predict future order volumes in advance, and combine it with real-time data analysis to achieve proactive prediction and dynamic adjustment of e-commerce orders.

[0041] The working principle and effect of the above technical solution are: collect e-commerce data in real time through the big data platform, and clean and store it. At the same time, the data is integrated and standardized to ensure data quality and availability. The processed data is analyzed in real time using the cloud data analysis platform, and the data value is deeply mined through statistical analysis, machine learning and other methods. The analysis results are displayed in the form of visual charts and other forms to provide decision support for e-commerce companies. Based on the order volume prediction model and real-time data analysis results, e-commerce companies can realize active prediction and dynamic adjustment of e-commerce orders. By predicting changes in order volume in advance and taking corresponding regulatory measures, the operational efficiency and profitability of e-commerce companies can be improved.

[0042] In one embodiment of the present invention, the S1 includes:

[0043] S11: Build a big data platform based on Hadoop, collect and clean the e-commerce data in real time, and store the data in HDFS to provide raw data for subsequent data warehouses;

[0044] S12: Building a data warehouse, and performing dimensional modeling on the data warehouse, layering data of different granularities, and performing offline analysis and data visualization on the e-commerce data.

[0045] The working principle and effect of the above technical solution are as follows: First, build a big data platform based on Hadoop. Hadoop is an open source distributed computing framework that allows distributed processing of large-scale data sets using simple programming models. On this platform, components in multiple Hadoop ecosystems can be integrated, and e-commerce data can be collected in real time using real-time data collection tools on the Hadoop platform. These tools can efficiently capture data streams from different data sources. The collected raw data often contains problems such as noise, duplication, incompleteness or inconsistent format, so data cleaning is required. The data cleaning process includes removing duplicate data, processing missing values, eliminating outliers, unifying formats, etc. to ensure the quality and availability of data. The cleaned data is stored in HDFS. HDFS is a distributed file system of Hadoop, which has the characteristics of high throughput, high fault tolerance and scalability, and is very suitable for storing large-scale data sets. Storing data in HDFS provides raw data support for subsequent data warehouses. Build a data warehouse on the Hadoop platform to store e-commerce data that has been integrated, cleaned and converted. Data warehouses are usually modeled using star schema or snowflake schema to facilitate data query and analysis. Dimensional modeling of the data warehouse means dividing the data into different fact tables and dimension tables according to business needs and analysis objectives. Data of different granularities are processed in layers to form multiple layers of the data warehouse. Data at each layer undergoes specific processing and conversion to meet the needs of different business needs and analysis scenarios. Data layering helps improve the flexibility and scalability of the data warehouse. Offline analysis is performed using data in the data warehouse to extract valuable information and insights through SQL queries, report generation, data mining and other means. At the same time, the analysis results are displayed through data visualization tools and presented to decision makers and management in the form of intuitive charts. Offline analysis and data visualization help e-commerce companies better understand key information such as market dynamics, user behavior and product performance, so as to formulate more accurate marketing strategies and business decisions. Hadoop and its distributed file system HDFS ensure data reliability and fault tolerance through data block replication and distributed storage. Even if some nodes fail, data will not be lost, ensuring the integrity and security of e-commerce data. HDFS can easily expand storage capacity. Storage space can be expanded by simply adding nodes. It supports PB-level data storage and meets the needs of rapid growth of e-commerce data. Hadoop uses distributed computing to process large amounts of data in parallel, supports high-concurrency read and write operations, provides high-throughput data access performance, and accelerates the processing and analysis of e-commerce data. Data visualization tools are used to intuitively display analysis results in the form of charts and reports, helping e-commerce companies quickly understand key information such as market trends, user behavior, and product performance.

[0046] In one embodiment of the present invention, S2 includes:

[0047] S21: Realize real-time data synchronization and seamless integration by connecting to the API interface provided by the e-commerce platform, and use intelligent web crawlers to automatically capture e-commerce data on the e-commerce platform;

[0048] S22: The standardized equipment automatically cleans, converts and loads the e-commerce data.

[0049] The working principle and effect of the above technical solution are as follows: First, select a suitable e-commerce platform API service according to business needs. Register a developer account on the selected e-commerce platform, fill in the necessary enterprise or personal information, and pass the platform's review. After successful registration, the e-commerce platform will provide an API key to verify the identity of the API call to ensure that only authorized developers can access and use the relevant API services. According to the API document, develop an interface call program to achieve real-time synchronization and seamless integration of data with the e-commerce platform. Through the API interface, you can obtain a variety of real-time data such as product information, order data, user behavior, etc., and integrate it into your own data system. Using intelligent web crawler technology, by writing a program to simulate browser behavior, automatically access the e-commerce platform website, and capture e-commerce data on the web page. After the crawler program captures the original data from the e-commerce platform web page, it needs to perform data parsing, extract the required information, and store it in a local database or file. Perform quality checks on the captured e-commerce data, identify and handle missing values, outliers, duplicate values, and other issues. Convert data from different sources and formats into a standard format for subsequent data processing and analysis. Convert text data from one encoding format to another to ensure the correct display and processing of data. Aggregate or split data to meet different analysis needs. Load the cleaned and converted data into a data warehouse or data lake for subsequent offline analysis and data mining. Index the data in the data warehouse to improve the efficiency of data query. Back up data regularly to ensure data security. When data is lost or damaged, it can be quickly restored. By connecting to the API interface of the e-commerce platform, real-time synchronization of data can be achieved to ensure the freshness and timeliness of data. This is crucial for e-commerce companies, as it can help them respond quickly to market changes and develop effective marketing strategies. The data provided by the API interface is usually strictly verified and quality controlled by the e-commerce platform, and has high accuracy and reliability. In addition, the data captured by intelligent web crawlers can also be verified and cleaned through certain rules and technical means to improve the accuracy of the data. The entire data processing process is highly automated, from data capture, cleaning, conversion to loading, reducing manual intervention and error rates. This can not only improve work efficiency, but also reduce labor costs. The automated processing of data by standardized equipment can significantly improve the processing speed, especially for large-scale data processing, which can complete data cleaning, conversion and loading faster, providing strong support for subsequent data analysis. Both the API interface of the e-commerce platform and the use of intelligent web crawlers are highly flexible and can be flexibly configured and adjusted according to business needs. In addition, different API interfaces can be selected or crawler rules can be modified to capture specific types of data as needed. Automated data processing by standardized equipment can reduce security risks during data transmission and storage.At the same time, encrypted storage and access control of sensitive data can ensure data security. The automated cleaning process can remove noise, outliers, duplicate values, and other problems in the data to improve data quality. At the same time, data verification and deduplication can ensure data accuracy and uniqueness.

[0050] In one embodiment of the present invention, S3 includes:

[0051] S31: Use cloud computing technology to build a cloud data analysis platform and deploy a data storage system on the cloud data analysis platform;

[0052] S32: Configure Spark Streaming on the cloud data analysis platform to process the real-time data stream and store the processed data in the data storage system.

[0053] The working principle and effect of the above technical solution are as follows: cloud computing technology provides computing resources, storage resources and data processing services as a service to users through the Internet. It has the characteristics of flexibility, scalability and high availability, and is an ideal choice for building a cloud data analysis platform. Design the architecture of the cloud data analysis platform according to business needs and the amount of data. It usually includes a data collection layer, a data processing layer, a data storage layer and a data analysis layer. Deploy the corresponding hardware and software resources on the platform of the cloud service provider, including virtual servers, cloud databases, data analysis tools, etc., to build a complete cloud data analysis platform. Select appropriate data storage technology based on factors such as data type, scale and access mode. Commonly used data storage technologies include relational databases, NoSQL databases, distributed file systems, etc. Design the architecture of data storage, including data partitioning, indexing, backup and recovery strategies, etc., to ensure data security and efficiency. Install the Spark framework on the cloud data analysis platform and configure the Spark Streaming component. SparkStreaming is an extension of the Spark core API for processing real-time streaming data. Configure Spark Streaming to connect to a real-time data source. Spark Streaming receives real-time data streams from the data source and divides the data streams into multiple small batches according to the set time interval. Each small batch of data is processed using Spark's batch processing operations and the processed data is output to the specified storage system. Cloud computing allows enterprises to deploy computing resources and storage resources on demand according to actual business needs, avoiding the problem of excess or insufficient resources in traditional IT infrastructure. The Spark framework uses in-memory computing technology, which can significantly improve the data processing speed compared to traditional disk I / O operations and meet application scenarios with high real-time requirements. Spark Streaming divides the real-time data stream into multiple small batches for processing, combining the advantages of batch processing and stream processing, which not only ensures the real-time nature of the data, but also improves processing efficiency. By utilizing cloud computing resources, enterprises do not need to purchase and maintain a large number of hardware devices, reducing hardware investment costs. Cloud computing platforms and Spark Streaming provide a wealth of API interfaces for easy integration and expansion with other systems and tools. Cloud data storage systems can support the storage and management of massive data to meet the needs of the e-commerce industry for data scale. Through the cloud data analysis platform, users can access and process data anytime and anywhere, improving the convenience and efficiency of data utilization.

[0054] In one embodiment of the present invention, the S4 includes:

[0055] S41: Collecting logistics freight order information through a freight information collection module, and merging the logistics freight order information into a unified data set;

[0056] S42: training the data set using the LSTM model, and dividing the data set into a training set, a validation set, and a test set, and training the model using the training set to find optimal model parameters, tuning the model using the validation set data, and evaluating the prediction performance of the model using the test set data, and calculating an error index;

[0057] S43: During the forecasting process, the production strategy is dynamically adjusted based on the feedback of real-time data and changes in forecast results.

[0058] The working principle and effect of the above technical solution are as follows: The freight information collection module is responsible for collecting logistics freight order information from various logistics links in real time or regularly. This information may include but is not limited to order number, delivery time, delivery time, delivery location, delivery location, cargo type, quantity, weight, volume, etc. Before the collected information is merged into a data set, it is necessary to clean the data to remove duplicate, erroneous or inconsistent data. This helps to ensure the quality and accuracy of subsequent model training. The cleaned logistics freight order information is merged into a unified data set. This data set will serve as the basis for LSTM model training. The unified data set is divided into training set, validation set and test set according to a certain proportion (60% training set, 20% validation set, 20% test set). These three sets should remain independent and non-overlapping in data to ensure the fairness and objectivity of model evaluation. The LSTM model is trained using the training set data. During the training process, the model learns the time patterns and features in the logistics freight order data to predict future order trends or status. The trained LSTM model is tuned using the validation set data. By adjusting the parameters of the model, the model configuration that performs best on the validation set is found. Finally, the prediction performance of the tuned model is evaluated using the test set data. The LSTM model is deployed in the production environment to predict the real-time collected logistics freight order data. A feedback mechanism is established to compare the prediction results with the actual order situation. By analyzing the prediction error and deviation, the prediction ability of the model can be continuously optimized. According to the feedback of real-time data and the changes in the prediction results, enterprises can dynamically adjust production strategies. For example, production plans, inventory management, staffing, etc. are adjusted according to the prediction results of the order volume to better meet market demand and optimize operational efficiency. The LSTM model can effectively capture and preserve long-term dependencies in sequence data through its unique gating structure, which makes it more accurate in predicting the future trend of logistics freight orders. Compared with traditional prediction methods, the LSTM model can better handle the complexity and nonlinear relationships in time series data. The LSTM model can learn sequence data of different lengths and adapt to various change patterns, which enables it to maintain a high prediction accuracy when facing large fluctuations in the logistics freight order volume. Through the freight information collection module, the logistics freight order information is collected in real time, and the LSTM model can make predictions based on the latest data, thereby ensuring the timeliness and accuracy of the prediction results. Based on the feedback from real-time data and changes in forecast results, companies can dynamically adjust production strategies, such as production planning, inventory management, and staffing, to better meet market demand and optimize operational efficiency. This dynamic adjustment capability enables companies to flexibly respond to market changes and improve market competitiveness. With accurate forecast results, companies can more reasonably arrange various links such as production planning, inventory management, and logistics transportation, thereby avoiding waste of resources and increased costs.By predicting the changing trend of order volume in advance, enterprises can prepare in advance and reduce delays and errors caused by sudden increase or decrease in order volume. By automatically collecting and processing logistics freight order information, enterprises can reduce manual intervention and errors and improve processing efficiency and accuracy.

[0059] In one embodiment of the present invention, the method of dynamically adjusting the production strategy in S43 is as follows:

[0060] The logistics order information collection system combines LSTM model prediction and dynamic scheduling to obtain comprehensive scheduling efficiency, where the comprehensive scheduling efficiency is obtained through the following formula:

[0061]

[0062] Among them, C represents the comprehensive scheduling efficiency, P represents the prediction error of the LTSM model, Indicates the difference between the real-time collected order volume data and the predicted order volume data. represents the maximum value of the difference between the order quantity data collected historically by the logistics order information collection system and the predicted order quantity data, S represents the scheduling adjustment range, Indicates the maximum adjustment range that can be achieved. represents the minimum value of the adjustment amplitude, and The value of is 0, and the adjustment range depends on the upper limit of production capacity; α represents the forecast error index weight, β represents the adjustment range weight, and α+β=1;

[0063] Furthermore, the prediction error of the LTSM model is obtained by the following formula:

[0064]

[0065] in, represents the actual order volume in the i-th time period, represents the predicted order volume in the i-th time period;

[0066] When 0.9<C≤1, it means that the prediction accuracy is high, the real-time data feedback is highly consistent with the prediction results, the scheduling strategy adjustment is accurate and efficient, and the existing strategy and system configuration should be maintained, and long-term performance should be monitored to confirm stability;

[0067] When 0.5<C≤0.89, it means that the overall performance is moderate. Appropriate improvement measures should be formulated for the poorly performing aspects of P, D, and S. For example, if the prediction accuracy is low, the parameters or training data of the LSTM model may need to be adjusted; if the real-time data feedback varies greatly, the data collection and processing process may need to be improved; if the scheduling strategy adjustment effect is not good, the scheduling algorithm may need to be optimized or the sensitivity to real-time data may need to be increased;

[0068] When 0.1<C≤0.49, it means that the overall performance of the system is poor and cannot meet business needs. An emergency repair plan should be formulated and implemented for specific problems, including quickly adjusting model parameters, optimizing data processing processes, and modifying scheduling strategies.

[0069] The working principle and effect of the above technical solution are as follows: The freight information collection module is responsible for collecting logistics freight order information from various logistics links in real time or regularly. This information may include but is not limited to order number, delivery time, delivery time, delivery location, delivery location, cargo type, quantity, weight, volume, etc. Before the collected information is merged into a data set, it is necessary to clean the data to remove duplicate, erroneous or inconsistent data. This helps to ensure the quality and accuracy of subsequent model training. The cleaned logistics freight order information is merged into a unified data set. This data set will serve as the basis for LSTM model training. The unified data set is divided into training set, validation set and test set according to a certain proportion (60% training set, 20% validation set, 20% test set). These three sets should remain independent and non-overlapping in data to ensure the fairness and objectivity of model evaluation. The LSTM model is trained using the training set data. During the training process, the model learns the time patterns and features in the logistics freight order data to predict future order trends or status. The trained LSTM model is tuned using the validation set data. By adjusting the parameters of the model, the model configuration that performs best on the validation set is found. Finally, the prediction performance of the tuned model is evaluated using the test set data. The LSTM model is deployed in the production environment to predict the real-time collected logistics freight order data. A feedback mechanism is established to compare the prediction results with the actual order situation. By analyzing the prediction error and deviation, the prediction ability of the model can be continuously optimized. According to the feedback of real-time data and the changes in the prediction results, enterprises can dynamically adjust production strategies. For example, production plans, inventory management, staffing, etc. are adjusted according to the prediction results of the order volume to better meet market demand and optimize operational efficiency. The LSTM model can effectively capture and preserve long-term dependencies in sequence data through its unique gating structure, which makes it more accurate in predicting the future trend of logistics freight orders. Compared with traditional prediction methods, the LSTM model can better handle the complexity and nonlinear relationships in time series data. The LSTM model can learn sequence data of different lengths and adapt to various change patterns, which enables it to maintain a high prediction accuracy when facing large fluctuations in the logistics freight order volume. Through the freight information collection module, the logistics freight order information is collected in real time, and the LSTM model can make predictions based on the latest data, thereby ensuring the timeliness and accuracy of the prediction results. Based on the feedback from real-time data and changes in forecast results, companies can dynamically adjust production strategies, such as production planning, inventory management, and staffing, to better meet market demand and optimize operational efficiency. This dynamic adjustment capability enables companies to flexibly respond to market changes and improve market competitiveness. With accurate forecast results, companies can more reasonably arrange various links such as production planning, inventory management, and logistics transportation, thereby avoiding waste of resources and increased costs.By predicting the changing trend of order volume in advance, enterprises can prepare in advance and reduce delays and errors caused by sudden increase or decrease in order volume. By automatically collecting and processing logistics freight order information, enterprises can reduce manual intervention and errors and improve processing efficiency and accuracy.

[0070] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. An e-commerce order analysis and control method, characterized in that: The method comprises: S1: Build a big data platform to collect, clean and store e-commerce data in real time; S2: Automatically capture e-commerce data, and integrate and standardize the e-commerce data; S3: Use a cloud data analysis platform to perform real-time analysis on the data; S4: Using the order volume prediction model, the future order volume is predicted in advance, and combined with real-time data analysis, the active prediction of e-commerce orders and the dynamic adjustment of the generation strategy are realized, and the method of dynamically adjusting the production strategy is as follows: The logistics order information collection system combines LSTM model prediction and dynamic scheduling to obtain comprehensive scheduling efficiency, where the comprehensive scheduling efficiency is obtained through the following formula: Where C represents the comprehensive scheduling efficiency, P represents the prediction error of the LTSM model, D0 represents the difference between the order volume data collected in real time and the predicted order volume data, D1 represents the maximum difference between the order volume data collected historically by the logistics order information collection system and the predicted order volume data, S represents the scheduling adjustment range, S max Indicates the maximum adjustment range that can be achieved, S min represents the minimum value of the adjustment amplitude, and S min The value of is 0, and the adjustment range depends on the upper limit of production capacity; α represents the forecast error index weight, β represents the adjustment range weight, and α+β=1; Furthermore, the prediction error of the LTSM model is obtained by the following formula: P=∑ i＝1 (and i -and i ') 2 Among them, y i represents the actual order volume in the i-th time period, y′ i represents the predicted order volume in the i-th time period; When 0.9<C≤1, it means that the prediction accuracy is high, the real-time data feedback is highly consistent with the prediction results, the scheduling strategy adjustment is accurate and efficient, and the existing strategy and system configuration should be maintained, and long-term performance should be monitored to confirm stability; When 0.5<C≤0.89, it means that the overall performance is moderate. We should formulate corresponding improvement measures for the poor performance of P, D and S. For example, if the prediction accuracy is low, we need to adjust the parameters or training data of the LSTM model; if the real-time data feedback is very different, we need to improve the data collection and processing process; if the scheduling strategy adjustment effect is not good, we need to optimize the scheduling algorithm or increase the sensitivity to real-time data; When 0.1<C≤0.49, it means that the overall performance of the system is poor and cannot meet business needs. An emergency repair plan should be formulated and implemented for specific problems, including quickly adjusting model parameters, optimizing data processing processes, and modifying scheduling strategies.

2. The e-commerce order analysis and control method according to claim 1, characterized in that: The S1 includes: S11: Build a big data platform based on Hadoop, collect and clean the e-commerce data in real time, and store the data in HDFS to provide raw data for subsequent data warehouses; S12: Building a data warehouse, and performing dimensional modeling on the data warehouse, layering data of different granularities, and performing offline analysis and data visualization on the e-commerce data.

3. The e-commerce order analysis and control method according to claim 1, characterized in that: The S2 includes: S21: Realize real-time data synchronization and seamless integration by connecting to the API interface provided by the e-commerce platform, and use intelligent web crawlers to automatically capture e-commerce data on the e-commerce platform; S22: The standardized equipment automatically cleans, converts and loads the e-commerce data.

4. The e-commerce order analysis and control method according to claim 1, characterized in that: The S3 includes: S31: Use cloud computing technology to build a cloud data analysis platform and deploy a data storage system on the cloud data analysis platform; S32: Configure Spark Streaming on the cloud data analysis platform to process the real-time data stream and store the processed data in the data storage system.

5. The e-commerce order analysis and control method according to claim 1, characterized in that: The S4 includes: S41: Collecting logistics freight order information through a freight information collection module, and merging the logistics freight order information into a unified data set; S42: training the data set using the LSTM model, and dividing the data set into a training set, a validation set, and a test set, and training the model using the training set to find optimal model parameters, tuning the model using the validation set data, and evaluating the prediction performance of the model using the test set data, and calculating an error index; S43: During the forecasting process, the production strategy is dynamically adjusted based on the feedback of real-time data and changes in forecast results.

6. The e-commerce order analysis and control method according to claim 3, characterized in that: The API interface includes: Order related interface, the order related interface is used to obtain the user's order information, the order information includes order status, order time, delivery address and logistics information.

7. An electronic device, characterized in that: It includes a memory, a processor, and a computer program stored in the memory and executable on the memory, wherein the processor executes the program to implement an e-commerce order analysis and control method as described in any one of claims 1 to 6.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement an e-commerce order analysis and control method as described in any one of claims 1-6.

9. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instruction is executed by the processor, an e-commerce order analysis and control method as described in any one of claims 1-6 is implemented.

Images