分享興趣，傳播快樂，增長見聞，留下美好。
親愛的您，
這裡是LearningYard學苑！
今天小編為大家帶來文章
“平學（44）：精讀英文論文《大資料分析對供應鍊績效的影響因素分析》第二章 文獻綜述（1）”
歡迎您的用心通路！
Share interest, spread happiness, increase
knowledge, leave a beautiful.
Dear you,
this is LearningYard Academy.
Today, the editor brings you an article
"Ping Xue (44)：Intensive reading of English paper “Analysis of Factors Influencing Supply Chain Performance by Big Data Analytics” Chapter2 Literature review "
Welcome to your visit!
一、内容摘要（Summary of content）
今天小編将從思維導圖、精讀内容、知識補充三個闆塊為大家帶來英文期刊論文《Impact of big data analytics on supply chain performance an analysis of influencing factors》的第二章 文獻綜述的第一部分。
Today, I will bring you the Chapter2 Literature review of the English journal paper ‘Analysis of Factors Influencing Supply Chain Performance by Big Data Analytics’ from the three sections of Thinking Maps, Intensive Reading Content, and Knowledge Supplement.
二、思維導圖（Mind mapping）
三、精讀内容（Intensive reading content）
1. 簡要概述(Brief outline)
作者首先描述了大資料分析在供應鍊管理中的研究現狀和趨勢，指出，從2010年到2021年，Web of Science資料庫中包含“大資料分析”關鍵詞的期刊論文數量達到5203篇，其中與制造供應鍊、服務供應鍊和零售供應鍊相關的論文分别為480篇、265篇和226篇。這表明大資料分析在供應鍊管理領域的研究興趣正在不斷增長。盡管如此，目前還沒有明确的研究針對零售供應鍊績效衡量标準（如供應商內建、客戶內建、成本、産能使用率、靈活性、需求管理和時間與價值）來選擇最佳的大資料分析技術（包括資料科學、神經網絡、ERP、雲計算、機器學習、資料挖掘、RFID、區塊鍊、物聯網和商業智能）。
The authors firstly describe the current research status and trend of big data analytics in supply chain management, and point out that from 2010 to 2021, the number of journal articles in the Web of Science database containing the keyword “big data analytics” reaches 5203, of which 480 and 226 articles are related to manufacturing supply chain, service supply chain, and retail supply chain, respectively. The number of journal papers containing the keyword “big data analytics” in the Web of Science database reaches 5203, of which 480, 265 and 226 are related to manufacturing supply chain, service supply chain and retail supply chain, respectively. This indicates that the research interest in big data analytics in supply chain management is growing. Nonetheless, there is no clear research on selecting the best big data analytics techniques (including data science, neural networks, ERP, cloud computing, machine learning, data mining, RFID, blockchain, IoT, and business intelligence) for retail supply chain performance measures (e.g., supplier integration, customer integration, cost, capacity utilization, agility, demand management, and time and value).
2. 印度零售供應鍊情況(Indian retail supply chain scenario)
通過梳理現有研究，印度71%的零售銷售來自快速消費品。但由于資料管理不當、庫存不準确和供應商管理問題，快消品零售存在嚴重問題，需要及時解決。是以，為了克服這一問題，需要在零售供應鍊管理中引入戰略采購、供應鍊網絡設計、産品設計與開發、需求規劃、采購、生産、庫存、物流、分銷和供應等領域的知識。現有研究中，有些從監督式和無監督的資料中發現特征，以提高供應鍊的品質。有些通過自我組織的能力來識别潛在客戶。有些實時跟蹤跨管道的活動，從一個單一平台進行改進資訊共享并最小化決策延遲。還有些通過分析實時資料和曆史傳遞記錄，幫助供應鍊經理優化車隊路線，減少駕駛時間，降低成本并提高生産率。
By combing through the available research, 71% of retail sales in India come from FMCG. However, there are serious problems in FMCG retailing due to improper data management, inaccurate inventory and supplier management issues which need to be addressed in a timely manner. Therefore, to overcome this problem, knowledge in the areas of strategic sourcing, supply chain network design, product design and development, demand planning, procurement, production, inventory, logistics, distribution and supply needs to be introduced in retail supply chain management. Some of the existing studies identify features from supervised and unsupervised data to improve the quality of the supply chain. Some identify potential customers through self-organized capabilities. Some track cross-channel activities in real time, from a single platform to improve information sharing and minimize decision-making delays. Still others help supply chain managers optimize fleet routes, reduce driving time, lower costs and increase productivity by analyzing real-time data and historical delivery records.
2. 大資料實踐(Big data practice)
2.1 資料科學(Data science)
在零售供應鍊中，由于顧客增加和技術進步，原始資料量急劇增長。通過将統計學和計算技術與資料科學相結合，可以從大量原始資料中提取、改進、存儲和監控有意義的資料，以支援決策。這種方法較為複雜且耗時，需要使用機器學習、資料挖掘和人工智能技術來提高資料準确性。盡管實施成本較高，但通過在零售供應鍊中應用資料科學，可以獲得更多的收益。
In the retail supply chain, the amount of raw data has grown dramatically due to increased customers and technological advances. By combining statistical and computational techniques with data science, meaningful data can be extracted, improved, stored, and monitored from large amounts of raw data to support decision making. This approach is more complex and time-consuming and requires the use of machine learning, data mining and artificial intelligence techniques to improve data accuracy. Despite the high cost of implementation, more can be gained by applying data science to the retail supply chain.
2.2 神經網絡(Neural network)
在零售和電商供應鍊中，由于顧客數量龐大且需求多變，預測需求和供應成為難題。神經網絡技術通過模拟人腦工作方式，可以有效預測銷售和顧客行為，廣泛應用于市場預測。然而，這種方法需要大量資料和較長的算法開發時間，計算成本高，并且無法解釋預測結果的具體原因，降低了對該技術的信任度。
In retail and e-commerce supply chains, predicting demand and supply becomes a challenge due to the large number of customers and their changing needs. Neural network technology is widely used for market forecasting as it can effectively predict sales and customer behavior by simulating the way the human brain works. However, this approach requires large amounts of data and long algorithm development times, has high computational costs, and is unable to explain the specific reasons for the prediction results, reducing the level of trust in the technique.
2.3 企業資源規劃(EPR)
随着零售行業中顧客數量的增加，産品種類必須根據顧客的口味和偏好進行擴充，導緻零售商規模擴大。然而，需求預測的困難使得零售商難以在正确的時間以正确的價格向正确的顧客銷售合适的産品。為了解決這一問題，零售商使用ERP系統來連接配接各部門，確定前後端從業人員緊密協作。對于産品混合和配置設定，零售商使用分銷需求計劃（DRP）子產品，幫助處理進出庫物流和庫存管理。ERP系統有助于零售商更好地了解庫存、訂單狀态和生産率等資訊，但安裝成本高且完全功能化可能需要多年時間，其成功依賴于員工的技能水準。
As the number of customers increases in the retail industry, the product assortment has to expand according to customer tastes and preferences, leading to an increase in the size of retailers. However, the difficulty of forecasting demand makes it difficult for retailers to sell the right product to the right customer at the right time at the right price. To solve this problem, retailers use ERP systems to connect departments and ensure that front- and back-end staff work closely together. For product mixing and distribution, retailers use Distribution Requirements Planning (DRP) modules to help handle inbound and outbound logistics and inventory management.ERP systems help retailers better understand information such as inventory, order status, and productivity, but they are expensive to set up and can take years to become fully functional, and their success relies on the skill level of staff.
2.4 雲計算(Cloud computing)
雲計算是一種新興的資源系統，因其無需使用者直接參與操作和具備大規模資料存儲能力而廣受歡迎。其主要特點包括按需服務、廣泛的網絡通路、資源池化、快速彈性伸縮和計量服務。這些特性簡化了供應鍊的操作和性能，降低了市場準入成本，使小企業能更容易進入市場。雲計算尤其對零售商和電子商務行業有益，但也存在帶寬問題和資料存儲安全問題。確定雲計算的安全需要額外投資，但從長遠看，雲計算能提升組織的财務表現。
Cloud computing is an emerging resource system that has gained popularity due to its ability to operate without direct user involvement and to have large-scale data storage capabilities. Its key features include on-demand services, broad network access, resource pooling, rapid elastic scaling and metered services. These features simplify supply chain operations and performance, reduce market entry costs, and enable small businesses to enter the market more easily. Cloud computing is especially beneficial to retailers and the e-commerce industry, but there are bandwidth issues and data storage security concerns. Securing the cloud requires additional investment, but in the long run, cloud computing can improve an organization's financial performance.
2.5 機器學習(Machine learning)
機器學習是一種利用機器模拟人腦解決問題的技術，通過結合大資料和算法進行疊代學習，進而識别資料模式并預測顧客需求。在零售和電商行業中，由于顧客數量增加和資料量增大，手動預測顧客需求和産品銷售變得非常困難。機器學習可以簡化複雜資料，使其易于處理，并輕松預測未來銷售，進而優化産品成本。然而，這種方法在導入高品質資料方面存在一定的局限性。
Machine learning is a technique that uses machines to simulate the human brain for problem solving by combining big data and algorithms for iterative learning in order to recognize data patterns and predict customer demand. In the retail and e-commerce industry, manually predicting customer demand and product sales has become very difficult due to the increase in the number of customers and volume of data. Machine learning can simplify complex data, making it easy to process and easily predict future sales, which in turn optimizes product costs. However, this approach has limitations in importing high quality data.
2.6 資料挖掘(Data mining)
資料挖掘是一個将大量資料精煉化，進而識别出用于問題解決和預測的模式的過程。它首先将大資料簡化，了解資料模式的趨勢。然後，它将資訊分享給行業内的所有部門，這意味着它消除了不需要的資料，使用對資源和營運表現更重要的資料。在零售和電子商務行業，資料挖掘技術可以精煉從顧客那裡收集的資料。但也有局限性，比如侵犯使用者隐私，這不安全也不準确。
Data mining is a process of refining large amounts of data to identify patterns for problem solving and prediction. It begins by simplifying big data to understand trends in data patterns. It then shares the information across all sectors of the industry, which means it eliminates unneeded data and uses data that is more important to resources and operational performance. In the retail and e-commerce industry, data mining techniques can refine the data collected from customers. However, there are limitations, such as invasion of user privacy, which is not secure or accurate.
2.7 RFID
RFID用于跟蹤庫存、生産率、産品在不同部門間的移動，并存儲和共享資訊，還能防止盜竊。它在倉庫管理、運輸管理、生産排程、訂單管理、庫存管理和資産管理等方面發揮了重要作用，提高了勞動生産率，減少了庫存損失，進而優化了供應鍊的整體性能。盡管RFID技術在安全性上有優勢，但相比條形碼，它存在準确性、可靠性和成本上的局限性，安裝也需要更多時間，并需要驗證是否适合特定行業。
RFID is used to track inventory, productivity, movement of products between different departments and to store and share information and also to prevent theft. It plays an important role in warehouse management, transportation management, production scheduling, order management, inventory management, and asset management, which improves labor productivity and reduces inventory losses, thus optimizing the overall performance of the supply chain. Despite its security advantages, RFID technology has accuracy, reliability and cost limitations compared to barcodes, and installation requires more time and verification of suitability for specific industries.
2.8 區塊鍊和物聯網(Blockchain and the Internet of Things)
區塊鍊在供應鍊中主要用于跟蹤各部門的功能，并在無需審計的情況下監控不可篡改的行為。在零售和電商行業中，通過區塊鍊和物聯網，可以更有效地控制缺陷并提高性能，但不會改變供應鍊的關鍵活動。然而，區塊鍊技術存在存儲、私鑰管理、網絡安全、過度依賴網際網路和可能導緻勞工失業等局限性。
Blockchain is mainly used in the supply chain to track the functions of various departments and monitor tamper-proof behavior without the need for auditing. In the retail and e-commerce industry, defects can be controlled more effectively and performance can be improved through blockchain and IoT without changing the key activities of the supply chain. However, blockchain technology has limitations such as storage, private key management, cybersecurity, over-reliance on the Internet and the potential for worker unemployment.
2.9 商業智能(Business intelligence)
在競争激烈的環境中，企業面臨資料管理、市場變化和業務營運的挑戰。商業智能通過分析、整合和收集資料，幫助企業擷取準确及時的資訊，監控市場趨勢和客戶行為變化。然而，這項技術存在成本高昂、技術僵化和實施耗時的問題。此外，依賴曆史資料模組化也成為限制，因為市場頻繁變化，企業不再專注于曆史資料。
In a competitive environment, organizations face challenges in data management, market changes and business operations. Business Intelligence (BI) helps organizations obtain accurate and timely information by analyzing, integrating and collecting data to monitor market trends and changes in customer behavior. However, this technology suffers from high costs, technical rigidity and time-consuming implementation. In addition, relying on historical data for modeling has become a limitation as the market changes frequently and organizations no longer focus on historical data.
四、知識補充（Knowledge supplementation）
什麼是有監督式資料，什麼是無監督式資料？
What is supervised data and what is unsupervised data?
有監督式資料是指帶有标簽的資料，而無監督式資料則是指沒有标簽的資料。在有監督學習中，訓練資料被标記為已知類别的樣本，模型通過學習這些樣本的輸入和輸出關系來預測新的未知資料。常見的有監督學習任務包括分類和回歸，例如在圖像識别任務中，使用已經标記的圖檔來訓練模型，使其能夠識别新圖檔中的物體。無監督學習則是在沒有标簽的情況下進行的學習，模型通過分析未标記的資料來發現資料中的結構和關系。常見的無監督學習任務包括聚類、降維和特征提取等，例如在市場細分分析中，無監督學習可以通過對消費者的行為資料進行聚類分析，将消費者劃分為不同的群體。
Supervised data refers to labeled data, while unsupervised data refers to unlabeled data. In supervised learning, the training data is labeled with samples of known categories, and the model predicts new unknown data by learning the input and output relationships of these samples. Common supervised learning tasks include classification and regression, such as in image recognition tasks, where already labeled images are used to train the model to recognize objects in new images. Unsupervised learning, on the other hand, is learning without labeling, where the model analyzes unlabeled data to discover structure and relationships in the data. Common unsupervised learning tasks include clustering, dimensionality reduction and feature extraction, for example, in market segmentation analysis, unsupervised learning can be used to classify consumers into different groups by clustering and analyzing their behavioral data.
今天的分享就到這裡了。
如果您對文章有獨特的想法，
歡迎給我們留言，
讓我們相約明天。
祝您今天過得開心快樂！
That's all for today's sharing.
If you have a unique idea about the article
please leave us a message
Let us meet tomorrow.
I wish you a happy day today!
參考資料：Google翻譯，百度，通義千問
參考文獻：Gopal P R C, Rana N P, Krishna T V, et al. Impact of big data analytics on supply chain performance: an analysis of influencing factors [J]. Annals of Operations Research, 2024, 333(2): 769-797.
本文由LearningYard學苑整理發出，如有侵權請在背景留言！
文字 | Ping
排版 | Ping
稽核 | yyz