Supply Chain Prescriptive Analytics
Predictive analytics identify patterns of potential issues or even causes of certain outcomes. On the other hand, Prescriptive analytics prescribe solutions to issues that are found or appear to be in the supply chain. For example, high inventory cost or too many late orders. They prescribe what to do in order to avoid or improve the underlying issue.
By its nature, planning problem is an interactable problem that falls in the class of NP-Hard. Adexa has a unique unified data model from network optimization all the way down to a piece of equipment that enables use of a number of prescriptive techniques to come up with solutions in the fastest possible time. Many of these techniques now fall under the umbrella of AI search techniques. These include Divide & Conquer, Tree search and Constraint Propagation search (CSP) that we use in conjunction with mathematical techniques such as Mixed Integer Linear Programming (MILP) for sourcing optimization and resource allocation. Adexa solutions also use some of the most innovative techniques for inventory optimization and postponement strategies such as Gradient Descent (GD). This is a search technique that optimizes the balance between cost of inventory and leadtime to deliver.
Depending on the level of planning and availability of data different search techniques are deployed in order to deliver the solution when needed. At the supply chain level producing a plan for the entire global supply chain may be done in minutes; but an Available-To-Promise request must be done in sub-seconds. Similarly, when an operator is waiting for instructions as to what to do next it must be done in real-time. Adexa deploys a combination of optimization and heuristic search algorithms, as described above, in order to deliver what is needed at different levels of supply chain in a timely manner.