Introduction [chapter]

2021 Data Science for Supply Chain Forecasting  
In the same way electricity revolutionized the second half of the 19th century, allowing industries to produce more with less, artificial intelligence (AI) will drastically impact the decades to come. While some companies already use this new electricity to cast new light upon their business, others are still using old oil lamps or even candles, using manpower to manually change these candles every hour of the day to keep the business running. As you will discover in this book, AI and machine
more » ... arning (ML) are not just a question of coding skills. Using data science to solve a problem will require more a scientific mindset than coding skills. We will discuss many different models and algorithms in the later chapters. But as you will see, you do not need to be an IT wizard to apply these models. There is another more important story behind these: a story of experimentation, observation, and questioning everything-a truly scientific method applied to supply chain. In the field of data science as well as supply chain, simple questions do not come with simple answers. To answer these questions, you need to think like a scientist and use the right tools. In this book, we will discuss how to do both. Supply Chain Forecasting Within all supply chains lies the question of planning. The better we evaluate the future, the better we can prepare ourselves. The question of future uncertainty, how to reduce it, and how to protect yourself against this unknown has always been crucial for every supply chain. From negotiating contract volumes with suppliers to setting safety stock targets, everything relates to the ultimate question: What Is Tomorrow Going to Be Like? Yesterday, big companies provided forecasting software that allowed businesses to use a statistical forecast as the backbone of their S&OP 2 process. These statistical forecast models were proposed sixty years ago by Holt and Winters 3 and haven't changed much since: at the core of any statistical forecast tool, you still find exponential smoothing. Software companies sell the idea that they can add a bit of extra 1 Andrew Ng is the co-founder of Coursera, the leading online-classes platform. 2 The sales and operations planning (S&OP) process focuses on aligning mid-and long-term demand and supply. 3 See Section 3.3 for more information about Holt-Winters models.
doi:10.1515/9783110671124-205 fatcat:shxhwox4kbflffxc7xi6zob7em