Searching for a solution
The state government in Melbourne, Australia contracted the machine learning (ML) team at GHD, a global consulting company, to improve container supply chain processes through the collection and understanding of large datasets from industry, government, and transport software service providers.
The team, led by Nikita Atkins, Data Science Global Leader, GHD, used Alteryx to narrow 100 million shipping container and commodity records down to 1.9 million with a 99.9965% match accuracy rate.
Additionally, using Intelligence Suite, they were able to forecast and build predictive models to better estimate where containers were going, what commodities they held, capacity level of each container, and how long it would take for them to return from their destination to the origination point to better assess government spending and infrastructure investment.
Every five years, the Port of Melbourne (PoM) in Australia is required to track all shipping containers that enter and exit. Understanding where freight moves is critical to ensuring the correct infrastructure, industrial land, planning controls, and policy settings are in place to support efficient supply chains.
The PoM was utilizing more than 57 independent groups to track the data in over 60 different formats. This process typically takes hundreds of hours of manual time and resources, and historically, the forecasting rate fell below 30%. Additionally, they were unable to complete a successful match analysis – until recently.
In 2019, Nikita Atkins and the machine learning team at GHD were able to develop a predictive modeling process and champion the commodities and container survey. They used Alteryx to gather data from September to October 2019 (over 250,000 container trips) standardize it, combine it, and de-dupe 100 million records provided and include over 200 business rules before making the data consumable. They were able to compare their final data set of 1.9 million records to the PoM and found that their data cleansing with Alteryx yielded a match of 99.9965%.
They used Alteryx Intelligence Suite to build 10 predictive models and compare their effectiveness. They narrowed their selection and were able to estimate a container’s location — including its origination and destination point, the commodities held, the capacity level of each container, and provide insight into what a container’s return trip ending point and timeline was.
After completing this process, Nikita’s team gathered the exact data from the PoM and compared it to what their Alteryx Intelligence Suite model had forecasted. The results showed their work had a 77% accuracy rate in tracking the trip cycle of commodities and shipping containers. This was undoubtably the highest predictive yield PoM had ever expected.
More to Discover
McLaren Racing fast tracks data analytics in the race to accelerate
McLaren Formula 1 team consolidates 11.8 billion data-points to maximize race performance.
Growing a data-culture, naturally, at Roquette
A global leader in plant-based ingredients implemented Alteryx to enable strategic decision-making at scale.
7-Eleven Brings Key Insights In-House with Alteryx
From small size to big gulp, 7-Eleven's tax team increased efficiency by 60% with Alteryx. See how they did it.