In Machine Learning

While a recent Forbes article on machine learning in supply chain focuses mainly on a particular vendor’s supply chain solution, it serves to highlight some of the ways that machine learning has more broadly been used to address persistent logistics challenges in business. (Given some companies’ reluctance to adopt “unproven” technologies, articles demonstrating that machine learning has been helping optimize operations for a long time are most welcome.)

A notable example from the article describes how machine learning can forecast across different time horizons to solve production or distribution problems, such as needing to predict a two week production schedule or a one week projection of inventory levels in a given region.

Machine learning continues to advance, and now operations leaders can also apply a machine learning model for simultaneously optimizing logistics operations spanning multiple time horizons. To illustrate, let’s look at a logistics hub, where goods come into the hub from one transportation mode and are sorted to continue on their designated routes. The machine learning model must optimize use of the sorting hub, optimize the transportation resources, and meet all levels of service guaranteed to the shippers (the customers).

The delivery time horizon could range from overnight to days or weeks later. But the sorting itself, within the logistics hub, needs to be optimized right now with all packages completely routed through the hub within a few short hours. The machine learning model needs to optimize for both time horizons.

Machine Learning Models

Using advanced techniques such as stochastic optimization, advanced machine learning models can account for a wide assortment of both known and unpredictable variables, such as shipping origin and destination, the varied service commitments for each package, resources available (staff, trucks/rail/air/ship, etc.), weather and weather-related delays, traffic and traffic-related delays, and so on.

The model can optimize for more variables than the human mind can handle across various time horizons, so the hub to meet their business requirements and ultimately process more shipments in the same window of time, reducing costs and improving efficiencies.

Discover how advanced machine learning can optimize operations in this white paper (registration required).

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.

economics of energy storageMachine Learning-Optimized Distribution Hubs