How can Machine Learning be used in Supply chain?

Daft Punk Machine Learning Blog Article

How Daft Punk Inspires us on Supply Chain Planning

In their 2001 hit “Harder, Better, Faster, Stronger”, Daft Punk demonstrated how using electronic music did not mark the end of human creativity but instead opened a new playground on which numerous artists have blossomed, to the benefit of all. Synthetic music made creativity available to a whole new range of artists that would not have had the chance to perform, should they have had to buy traditional instruments: a computer and a microphone were enough to make a demonstration record and get produced.

Almost twenty years later, the title still echoes as the debate that opposes human and machine is louder than ever and these words are blamed for being a source of anxiety: Harder – Better – Faster – Stronger. What if we could read these words as a solution, rather than a problem?

Decision making in the context of Supply Chain Planning has become Harder:

  • supply chains are more fragmented, speculative and the geopolitical context recently became much more uncertain than it used to be;
  • consumers expect products to be more tuned to their needs, which results in shorter product life cycles: this increases the business complexity as it multiplies the number of references, increases stock, and reduces production efficiency;
  • as supply chain awareness rise, organizations expect more from supply chain planners than just a balanced plan between demand and supply: through S&OP or IBP processes, supply chain planning aims at being the sole reconciliation process between the company’s strategy, tactic decisions, and operations; this approach requires to consider more information such as financial aspects and rely on numerous “what-if” analysis with countless scenarios;
  • the turnover of supply chain planners makes it harder to capitalize on internal knowledge while knowing the market specificities and the supply constraints is more critical than ever;
  • supply chain planners are bombarded with information, both quantitative and qualitative, that they do not have time to assess or take decisions from; relevant information can get lost in the middle the noise and this can lead to decision paralysis.

In this context, an effective supply chain planning process and underlying systems are no longer a luxury but a vital condition for the operability and the competitiveness of one’s organization.

New technologies are part of the answer to obtain Better quality outputs from lower quality inputs in a Faster way. The most promising is Machine Learning (ML): a branch of Artificial Intelligence already used in a wide variety of decision-making problems such as suggesting relevant content on Netflix or the early detection of skin cancer. Machine Learning is a set of theories and techniques in which systems learn and improve themselves based on past performance and/or on correlations within the data provided. The reason this technology produces rich results in complex environments is that it is not explicitly programmed, hence it can unveil relations that no one had thought about before. It can also refute links that planners thought existed or human-biased decisions.

The Value Proposition of Machine Learning

QAD DynaSys provides two types of usages in its latest version of DSCP:

  • automation: using the machine to perform simple and recurring tasks based on prior resolutions and/or pure data analysis. Concrete applications in the 2019 versions of DSCP are the data cleansing and the clustering.
  • augmentation: having the machine suggest courses of actions or “what-if” analysis to the planner based on the data available; in this case, the planner will be told something he did not know before and may investigate further to decide to follow-up on this lead: the main goal is to obtain actionable insight. The 2019 version of DSCP includes Machine Learning-based forecast algorithms in which the user can include any data that may be of interest; the algorithm will benefit from it or disregard it depending on its true relevance.

The value proposition of Machine Learning for an organization is still under discussion because its highlighted applications often implies big data, organizational changes and in-house knowledge of data-science; all of which are challenging today. At QAD DynaSys, we believe in bringing value and high Return On Investment (ROI) from this technology without gathering more data nor hiring rare and expensive resources. Contrary to the common belief, Machine Learning is a flexible tool that already gives better results on small traditional datasets and that can grow much richer and Stronger as the users integrate additional information: this is why investing soon in this technology can (1) bring immediate value while preparing the future and (2) demystify this buzzword. The algorithms used at QAD DynaSys are designed so the planners can use it without the technical background.

To summarize, though the context of supply chain planning is harder, Machine Learning solutions exist to obtain better results by treating more data faster; these solutions will get stronger as the planners identify relevant data to incorporate. Harder – Better – Faster – Stronger.