How Big Data Can Optimize Global Shipping
The shipping industry is the lifeblood of the world economy, moving 90% of merchandise valued at over $19 trillion in 2014, according to the World Trade Organization. However, the forecasts relied on by many shipping companies leave something to be desired. Now, with big data analytics settling into the wheelhouse, the industry is set to make a sizable leap in efficiency.
It’s easy to forget how quickly the field of big data analytics has arisen, and what steps a given organization must take to be in a position to take advantage of it. For industries like financial services that digitized their process long ago, the move up to using predictive analytics was not a traumatic one. In healthcare, the passage of the American Recovery and Reinvestment Act in 2009 carried the mandate to shift to electronic medical records by 2014, which greased the wheels for the adoption of big data analytics in medicine.
But some industries are far behind the digital curve. The shipping industry, in particular, has lagged behind other businesses in moving beyond paper-based processes and using centralized computers to automated key processes. To be sure, there are big, customized ERP systems in place at many of the biggest shippers. However, many of the smaller ones have only dipped their toes into the digital waters. The challenge is formidable, to be sure, but that makes the pay-off potentially much bigger.
Industry leaders are stepping up to confront the challenges and opportunities that digitization and big data present to shipping. The advent of GPS navigation, real-time weather data feeds, and RFID-equipped smart containers promise to transform the shipping business, optimizing the delivery of goods and the distribution of containers, according to Dieter Berg, the head of business development for Munich Re Global Marine Partnership.
“I’m convinced that the entire field of traffic and port logistics as well as just-in-time shipping will change,” Berg writes in a story on the impact of big data and the Internet of Things (IoT) on his company’s website.
Multi-stop trip routing in is a notoriously difficult problem to solve, and it’s no different in the shipping business. There are more than 100,000 ships in the maritime shipping fleet, ferrying merchandise-filled containers and tanks full of bulk cargo to ports thousands of miles across the ocean. In 2012, it was estimated there were 20.5 million standard containers, around the world, accounting for 60% of the world’s seaborne trade.
One company that’s aiming to optimize the flow of containers around the world is the Silicon Valley startup ClearMetal. Founded two years ago by data scientists and engineers from Stanford University and Google, ClearMetal takes a math-based approach to predicting what containers should be where, and when.
“We’re a data science and artificial intelligence company,” says Adam Compain, ClearMetal CEO and co-founder. “We predict all events in the logistics cycle with precision. By doing that, we’re able to provide different parties in the supply chain – whether it’s a shipper, retailers, or freight forwarders or terminals – with actionable insights into how they can better manage their operations.”
Forecast: Not So Excel-lent
Among the various parties in the global supply chain, much of the planning for container placement is done in off-the-shelf forecasting tools, Excel, or worse–between the ears of an experienced manager.
“Excel and intuition are not that sophisticated in managing the kinds of variability there are in all the elements that go into operations,” Compain says. “It’s simply not really a sharp knife. They’re top-down approaches and static models. They look at overall averages, pump data through a static model, and come out with a forecast.”
Those forecasts give the average shipper less-than-stunning results when it comes to positioning a fleet of 100,000 to a million shipping containers in the optimal position. The shippers need to decide whether to keep a container sitting in a certain location, in hopes that a retailer will fill it up and move it, or to re-position it to a location where it’s more likely to be utilized.
Getting the right mix of containers in space and time is not easy, especially considering the vagaries of global commerce. “There’s massive complexity and uncertainty around how customers are going to behave, how markets are going to move, and how the operations are going to perform.”
Machine Learning to the Rescue
Rather than take an aggregate approach to predicting container demand, ClearMetal uses machine learning and data science techniques to forecast demand at a much more granular level, right down to the individual shipping container.
It all starts, as every data science project does, with the data. ClearMetal feeds its random forest models with data around booking information, vessel voyage information, and container movement data. The company uses these models to power Monte Carlo simulations that try to predict exactly where and when containers will be needed at each leg of their journeys. The models, which run on Amazon’s AWS cloud, provide up to an eight-week window into the future.
“Our system is actually looking at every piece of the shipping cycle in a way that’s tailored to the shipment industry to come up with much more accurate and dynamic real-time model and forecast of what will happen,” Compain says.
ClearMetal seeks to provide specific answers to specific questions. “As a carrier we’d provide them an application that helps them underrated exactly where they have containers, where they need them, and what their imbalance is,” he continues. “For freight-forwarders, it’s an aggregation helping them understand, of the bookings they get from shipper customers, what will actually dispatch and ship and precisely where. And for a terminal, it might be predictions and applications around exactly when a vessel will arrive and berth, or how containers might flow in and out of their gate.”
In proof of concepts, ClearMetal has demonstrated that it can deliver container demand predictions that are 30% to 60% more accurate than the predictions shippers are currently using. For a large shipping company with hundreds of thousands of containers, that can translate into a savings well into the millions of dollars. All told, the company estimates more than $2 billion could be saved by optimizing the shipping supply chain.
ClearMetal is tackling just one aspect of the massive shipping industry. But as more sensors are installed and companies become more comfortable using the Internet of Things (IoT), there are a number of other applications that are possible to boost the efficiency of logistics operations, especially in the areas of security, safety, just-in-time shipping, and food spoilage.
August 7, 2020
- Sumo Logic Expands its Observability Suite with Added Solutions
- Google Cloud Delivers Enhancements to Looker that Optimize Performance, Accelerate Application Development
- Terbium Labs and DarkOwl Announce Partnership
- Mode Analytics Raises $33M in Series D Funding, Led by H.I.G. Growth Partners
August 6, 2020
- Online Applied Data Analytics Program Focuses on Data Decision-Making for Working Professionals
- Informatica and Google Cloud Expand Strategic Partnership with Deeper Integrations
- Swarm64 Announces Strategic Partnership with Command Prompt
- Confluent Launches Confluent Cloud in All Three Major Cloud Marketplaces
- HPE, SAP Partner to Deliver SAP Hana Enterprise Cloud with HPE Greenlake Cloud Services
- Zencity Raises $13.5M in Funding
- NIH Harnesses AI for COVID-19 Diagnosis, Treatment, and Monitoring
August 5, 2020
- Isima Introduces Self-Service, Hyper-Converged Data Platform
- Brussels Hospital Manages COVID-19 Outbreak with Qlik
- Jupiter Announces Launch of ClimateScore Global
- Research: 83% of IT Leaders are Not Fully Satisfied with their Data Warehousing Initiatives
- The NLP Summit 2020 Program Announced
August 4, 2020
- Yellowbrick Data and Emtec Enter Partnership
- cnvrg.io AI OS Delivers Accelerated ML Workloads with Support of NVIDIA A100 Multi-Instance GPU
- Domo Enhances its COVID-19 Global Tracker with Google and Apple Mobility Trends
- Quantum ActiveScale Software Verified as Veeam Ready Object Solution
Most Read Features
- Big Data File Formats Demystified
- Big Data Apps Wasting Billions in the Cloud
- How to Build a Better Machine Learning Pipeline
- What’s the Difference Between AI, ML, Deep Learning, and Active Learning?
- Is Python Strangling R to Death?
- To Centralize or Not to Centralize Your Data–That Is the Question
- How COVID-19 Is Impacting the Market for Data Jobs
- Is Hadoop Officially Dead?
- Hacking AI: Exposing Vulnerabilities in Machine Learning
- Tracking the Spread of Coronavirus with Graph Databases
- More Features…
Most Read News In Brief
- Researchers Explore Link Between American Individualism and Poor COVID-19 Response
- Left for Dead, R Surges Again
- Data Prep Still Dominates Data Scientists’ Time, Survey Finds
- Why Gartner Dropped Big Data Off the Hype Curve
- HPE Acquires MapR
- Global DataSphere to Hit 175 Zettabytes by 2025, IDC Says
- Spark 3.0 Brings Big SQL Speed-Up, Better Python Hooks
- Kepler AutoML Targets Next-Gen Business Analysts
- IBM Brings Back a Netezza, Attacks Yellowbrick
- War Unfolding for Control of Elasticsearch
- More News In Brief…
Most Read This Just In
- FortressIQ Launches Adaptive Computer Vision-Based Firewall for Data Privacy
- UBS Launches Big Data Shareholder Activism Tool
- Cloudera Foundation Announces Grant Partnership with Urban Institute
- Orange and Google Cloud to Form Partnership in Data, AI and Edge Computing Services
- Syniti Acquires Virtyx Technologies
- KNIME Analytics Platform 4.2 is Now Available
- Hazelcast, Sorint Expand Partnership to Address In-Memory Computing Adoption
- Privacera Raises $13.5M in Series A Funding
- MariaDB Platform X5 Adds New Distributed SQL
- TileDB Closes $15M Series A to Expand its First Universal Data Engine
- More This Just In…