One area of interest in today’s end-to-end supply chain is machine learning. And this is certainly a topic that we have written about quite often. Over the last few months, Steve Banker, Clint Reiser, and I have written about artificial intelligence and machine learning in a number of contexts and how it impacts the supply chain. These topics have included transportation management, warehouse management, and supply chain planning, among others. This technology continues to be a hot topic for companies as is evident by how often the Logistics Viewpoints team and others are writing about it. In fact, we were just featured in a CSCMP Supply Chain Quarterly article that looks at the rise of machine learning within the supply chain. You can read the full article here.
Machine learning is a branch of artificial intelligence. “Learning” occurs when a machine takes an existing data set, observes the accuracy of the output, and updates its own model so that better outputs will occur. Any machine that does this is using machine learning. The end result is a smarter machine that creates efficiencies and, in most cases, cost savings.
As we have written about, there are many supply chain applications that machine learning can impact. Below are some highlights for a few areas.
Transportation management systems have a proven ROI. The primary reason companies buy a transportation management system is for freight savings. These freight savings can be attributed to simulation and network design, load consolidation and lower cost mode selections, and multi-stop route optimization. An ARC Advisory Group strategic report on the ROI of TMS found that respondents indicated freight savings of approximately 8% with the use of a TMS application. But that does not mean that there is not room for improvement. Machine learning enables a TMS to better handle competing objectives and discover non-obvious impacts on performance.
Machine learning gives companies the ability to maintain high service levels while achieving these savings. Shippers can learn which carriers meet on-time service levels and which do not, which lanes typically carry more chance for delays, and whether there is an optimal number of stops before shipments become late. Machine learning can aid shippers in better understanding how to drive efficiencies without sacrificing service levels.
For example, in last mile routing the time a job takes to complete is dependent not just on the miles that need to be driven, but on the congestion, the type of product being delivered, the type of residence, and whether the value-added services are provided at the destination. Machine learning can be used to “learn” these constraints rather than having to do time studies and hard code these constraints into the solution.
Real-time visibility solutions are raising the prospect that machine learning can be used to improve ETAs as well. But unless a TMS provider has access to network transportation data (public cloud providers do), the visibility provider will be in a better position to use artificial intelligence to provide these enhanced ETAs than the TMS provider.
IoT and connected trucks and containers are gaining a lot of traction in the market. When assets are on the move, data must be used to optimize functional operations. This includes both fleet safety and route optimization. Often, customers have a difficult time understanding the data they are collecting from their assets. However, when combined with artificial intelligence, the right IoT application can help make sense of data to improve asset performance. Sensors in trucks can help provide better visibility and make predicted ETAs more accurate. Additionally, connected trucks can provide information on specific driver’s driving patterns and habits, leading to improved safety as well as visibility.
Part of what makes warehousing a suitable application for machine learning is the fact that a warehouse operating environment is constantly in flux, especially in today’s direct-to-consumer facilities. These facilities must constantly balance the competing priorities of efficiency and responsiveness. At the same time, there are numerous potential constraints on warehouse operations, and it is difficult to predict under which circumstances a given function or resource may become a constraint on throughput. Predictability becomes especially difficult when a facility dynamically introduces orders into an existing workload. Machine learning’s ability to adapt to changing conditions in complex environments means that it can produce insights that would not be possible with traditional software.
Steve Banker published an article on that discussed three qualities that distinguish machine learning from predictive analytics . They are self-modification, high automation, and embedded within a process or workflow. These qualities provide a useful lens to view machine learning, its potential in WMS applications, and how it compares to other algorithms in use.
The quality of self-modification is what we consider to be a defining characteristic of machine learning for warehouse applications. It is especially useful for its ability to adapt to changing situations. Essentially, it is good for dynamic data sets, where the relationship between the dependent variable and the independent variables is fluid. When the predictive power of an algorithm changes, machine learning can recognize this degradation and create a new input-output relation that offers more robust predictive power.
Machine learning’s ability to adapt to changing conditions makes it especially well-aligned with the nature of today’s e-commerce warehouses that dynamically introduce orders into the existing work load. Waveless warehouse operations are a strong fit for the application of machine learning. Machine learning is used to predict the time required to complete work. An optimization algorithm then uses those results to balance competing requirements while optimally utilizing available capacity.
Supply Chain Planning
Demand planning is a good application for machine learning because the measure of success – the forecast accuracy – is clear. To learn, an application needs a clear measure of success. Having a clear measure of success sounds easy.
But often, defining success in not easy. Consider a situation where a manufacturer learns of a shortage of a key component. Customers have already been promised products that depend upon that key production input. The supply chain planning engine needs to be rerun to generate a new supply plan.
Some suppliers of supply chain planning are looking to “solve” this problem by using pattern recognition to see how planners resolved similar problems in the past and then suggest a similar resolution the next time the problem arises. The idea is that the solution will suggest a resolution, but people make the choices. But this type of solution may be difficult to achieve. The more at bats a system has, the faster it gets smarter. If you are doing daily demand plans, and every day you have new data on how well yesterday’s forecast did, you have a system that can improve quickly.
The above examples are just a small portion of the supply chain applications that are impacted by machine learning. Other areas such as warehouse automation, transportation execution systems, global trade management, order management, e-commerce fulfillment, and a myriad of other applications rely on machine learning. Machine learning will continue to help organization evolve and refine their supply chain processes.