• DATVF.ATLPHL
    1.638
    -0.014
    -0.8%
  • DATVF.CHIATL
    1.963
    0.087
    4.6%
  • DATVF.DALLAX
    0.897
    -0.106
    -10.6%
  • DATVF.LAXDAL
    1.549
    -0.024
    -1.5%
  • DATVF.SEALAX
    0.976
    0.052
    5.6%
  • DATVF.PHLCHI
    0.939
    0.039
    4.3%
  • DATVF.LAXSEA
    2.034
    -0.050
    -2.4%
  • DATVF.VEU
    1.513
    0.037
    2.5%
  • DATVF.VNU
    1.414
    -0.009
    -0.6%
  • DATVF.VSU
    1.223
    -0.065
    -5%
  • DATVF.VWU
    1.505
    0.001
    0.1%
  • ITVI.USA
    10,122.770
    72.300
    0.7%
  • OTRI.USA
    4.880
    -0.040
    -0.8%
  • OTVI.USA
    10,116.640
    68.200
    0.7%
  • TLT.USA
    2.420
    0.000
    0%
  • WAIT.USA
    150.000
    0.000
    0%
  • DATVF.ATLPHL
    1.638
    -0.014
    -0.8%
  • DATVF.CHIATL
    1.963
    0.087
    4.6%
  • DATVF.DALLAX
    0.897
    -0.106
    -10.6%
  • DATVF.LAXDAL
    1.549
    -0.024
    -1.5%
  • DATVF.SEALAX
    0.976
    0.052
    5.6%
  • DATVF.PHLCHI
    0.939
    0.039
    4.3%
  • DATVF.LAXSEA
    2.034
    -0.050
    -2.4%
  • DATVF.VEU
    1.513
    0.037
    2.5%
  • DATVF.VNU
    1.414
    -0.009
    -0.6%
  • DATVF.VSU
    1.223
    -0.065
    -5%
  • DATVF.VWU
    1.505
    0.001
    0.1%
  • ITVI.USA
    10,122.770
    72.300
    0.7%
  • OTRI.USA
    4.880
    -0.040
    -0.8%
  • OTVI.USA
    10,116.640
    68.200
    0.7%
  • TLT.USA
    2.420
    0.000
    0%
  • WAIT.USA
    150.000
    0.000
    0%
BusinessEconomicsInnovationInsightsMarket InsightNewsSupply ChainsTechnology

Commentary: Applying machine learning to improve the supply chain

FreightWaves features Market Voices – a forum for voices with unique knowledge of numerous transportation/logistics/supply chain sectors, as well as other critical expertise.

In “Tackling Climate Change with Machine Learning,” David Rolnick and his co-authors address how machine learning (ML) and artificial intelligence (AI) may be applied to the problem of climate change. They “identify high impact problems where existing gaps can be filled by machine learning, in collaboration with other fields.”

In this commentary, I focus on their recommendations for supply chain. In the paper they categorize the solutions they discuss as High Leverage, Long-term and High Risk. I will provide simplified and summarized descriptions of their High Leverage recommendations. This is a companion to How can machine learning be applied to improve transportation?

Supply chain problems are generalized dynamic assignment problems in which limited resources are deployed over a network to meet random demands occurring over time. The system is dynamic because each time a resource is allocated to meet a demand, the whole system changes in a way that cannot be predicted before the fact. This reflects the uncertainty of the real world in which supply chains exist. This class of problems is also referred to as dynamic resource allocation problems. 

The authors mention a few reasons why ML can be realistically applied in industrial settings to solve supply chain problems and reduce the emission of greenhouse gases (GHGs). First, the industrial sector collects large amounts of data. Second, affordable cloud storage and cloud computing makes it possible for ML to have a positive impact since it can be applied at scale.

They also issue words of caution. First, increased supply chain efficiency could lead to increased production, negating the positive impact the use of ML could have on GHG emissions. Second, industrial data is often proprietary, of poor quality, or otherwise unavailable to, and unusable by ML researchers and their algorithms.

Making a dent in waste

Waste is one area in which ML can make a difference. Much of the waste that occurs in supply chains is the result of inadequate demand forecasting. Poor demand forecasting leads to over-production and overstocking. ML can mitigate these problems by improving the accuracy of demand forecasts for manufactured goods. By combining improved demand-forecasting with just-in-time manufacturing on a regional basis, suppliers can lower the amount of excess inventory that they hold as well as lower the overall cost of producing goods for sale.

For example, the fashion industry discards billions of dollars of unsold clothing each year. This unsold clothing would never have been manufactured in the first place if demand forecasts were more accurate. Moreover, the manufacture of excess clothing leads to unnecessary textile waste; waste created from sold and unsold clothing.

There’s a growing movement towards circular and regenerative supply chains instead of the linear and extractive supply chains we are accustomed to. In the latter, once a product has been used it is discarded as waste. In the former, once a product has been used it becomes raw material for a new product. The new product may fall within the same category, or it may belong to a completely different category. Examples are processing old clothing items made from cotton into new cotton fiber used to make new clothing, and using discarded plastic bottles as raw materials for new athletic shoes.

Materials and construction is another area in which ML can be applied to transform traditional methods of producing cement and steel so that GHG emissions are reduced substantially. ML can be used to create new building materials that match the performance characteristics of cement and steel but that do much less harm to the climate. Progress in this area will depend on a combination of ML with advances in materials science and simulation.

Industrial production and energy management are areas in which optimization can be combined with ML and applied to managing energy consumption for factories, warehouses, cold chain storage facilities and other types of industrial facilities.

ML could be applied in:

  • Fertilizer production – to reduce the amount of energy consumed.
  • The management of industrial HVAC systems – to improve energy consumption, reduce GHG emissions and lower operating costs.
  • The management of industrial air-conditioning systems – to improve energy consumption, reduce GHG emissions and lower operating costs.
  • The ongoing analysis and monitoring of machinery and equipment – to improve predictive maintenance and prevent costly breakdowns.

Conclusion – transportation and supply chain present a rich opportunity for ML

While ML offers great potential for applications in transportation and supply chain, there are a few preliminary questions that must be answered before going down the path of implementation:

  • Who has access to the data required to train ML models?
  • Given the specific context, will ML provide significantly superior results over the status quo or other techniques that are easier to implement?
  • How likely is it that an ML implementation will become integral to how the business operates on an ongoing basis?
  • What are the potential unintended consequences of incorporating ML into transportation and supply chain operations as it pertains to GHG emissions?

The authors conclude by stating that ML is worth pursuing when:

  • There are large quantities of high quality and readily available data about transportation routes, supply chain networks and industrial processes.
  • When different firms that form part of production and distribution network are willing to collaborate by sharing data and access to granular data about relevant routes, networks and processes.
  • When the benefit of implementing new systems outweighs the cost of maintaining old systems and processes.
  • When there’s an alignment of goals and incentives to reduce GHG emissions.

The authors’ emphasize the role ML can play in combating climate change, an area of paramount concern. Several of my previous commentaries have highlighted the impact that climate change will have on supply chains overall. However, while companies work to develop a concerted response to climate change in collaboration with their partners and competitors, there are many potential benefits for individual companies that adopt and implement ML on an individual basis.

Tags
Show More

Brian Aoaeh

Brian Laung Aoaeh writes about the reinvention of global supply chains, from the perspective of an early-stage technology venture capitalist. He is the co-founder of REFASHIOND Ventures, an early stage venture capital fund that is being built to invest in startups creating innovations to refashion global supply chain networks. He is also the co-founder of The Worldwide Supply Chain Federation (The New York Supply Chain Meetup). His background covers the gamut from scientific research, data and statistical analysis, corporate development and investing for a single-family office, and then building an early stage venture fund from scratch - immediately prior to REFASHIOND. Brian holds an MBA in General Management, with a specialization in Financial Instruments and Markets, from NYU’s Stern School of Business. He also holds a Bachelor’s Degree in Mathematics & Physics from Connecticut College. Brian is a charter holding member of the CFA Institute.

Leave a Reply

Your email address will not be published. Required fields are marked *

Close