How to Take Your Demand Planning Skills to the Next Level

Data Science will allow demand planners to bring their forecast accuracy to unprecedented levels. Learning data science is possible for anyone — but will require time.

Nicolas Vandeput
Analytics Vidhya
Published in
5 min readJul 14, 2021

--

This article was initially published in Business Forecasting: The Emerging Role of Artificial Intelligence and Machine Learning (Wiley and SAS Business Series)

Source

Based on my experience as a consultant working with clients on forecasting models and teaching forecasting to master students, I see three main ideas that will take any demand planner to the next level. After discussing those ideas, I will show you how to get started using machine learning for demand forecasting.

Taking Demand Planning Skills to the Next Level

1. Machine Learning is easy to use. You can do it.

While the first machine learning models date back to the 1960s with the work of Morgan & Sonquist (1963) on decision trees, and Rosenblatt (1957) on perceptrons (the inner neurons of a neural network), the revolution of machine learning in forecasting started in the mid-2010s.

Conventional wisdom attributes this revolution to the increase in computation power along with the availability of bigger datasets. But this would be missing two other trends that made this revolution possible. First, there are better machine algorithms developed in the last years, as we find in papers by authors such as Kingma & Ba (2015) who proposed a better optimization technique for neural networks, Ke et al. (2017) and Tianqi & Guestrin (2016), who proposed efficient and powerful methods for boosting decision trees. And second, there is an improved ease-of-use that made machine learning usable by non-experts, for example with scikit-learn (Pedregosa et al., 2011).

Today, as shown in my book Data Science for Supply Chain Forecasting (Vandeput, 2021), you only need around 60 lines of code in Python to format your data, train, and optimize a neural network (using only free and open software). With online learning platforms developing since the mid-2010s (such as edx and coursera), learning new skills…

--

--

Nicolas Vandeput
Analytics Vidhya

Consultant, Trainer, Author. I reduce forecast error by 30% 📈 and inventory levels by 20% 📦. Contact me: linkedin.com/in/vandeputnicolas