Amazon Forecast Proof of Concept
About Duty Free Americas
Duty Free Americas (DFA) is a leading travel retailer in the Western Hemisphere with more than 180 retail locations. A large portion of their retail footprint encompasses major airports in the United States, as well as numerous locations on the northern and southern borders with Canada and Mexico.
Amazon has been making use of Machine Learning for their retail business for well over a decade. When leveraging Amazon Forecast, businesses can tap into amassed ML experience and success of Amazon.com, and do so without the same investment in infrastructure, training, and talent.
Duty Free Americas has a process of calculating demand and forecasting inventory levels for inventory planning. While this process has worked, they were hoping to measure their current processes and mechanisms against other options and identify areas that may need improvement. They were also considering ways to potentially increase the accuracy of the forecasts and their ability to use them to adequately maintain inventory levels of some products.
Through the use of Amazon Forecast, Duty Free Americas would be able to increase forecast accuracy by leveraging Forecast’s DeepAR+ algorithm to validate their inventory planning forecast models. In addition, DFA would be able to improve forecasting accuracy through inference by combining the standard target time series dataset with a related time series dataset and an item metadata dataset.
Improved accuracy of inventory forecasting would save time and money by improving their ability to maintain adequate inventory on all products and reducing their losses associated with excess inventory, reduced sales prices, and shipping excess inventory between distribution centers and retail locations to move it.
The initial steps of the project involved 1Strategy working with DFA to build and test the Forecast service with the target time series dataset. With the initial target time series dataset, 1Strategy was able to test the training of a predictor and generation of a forecast. After the initial test run, 1Strategy worked with DFA to identify other metrics that could be included in the related time series, as well as important metadata related to products to be included in the item metadata dataset. After a second round of testing to train a new predictor using all three datasets, the entire process was automated and documented using a series of Jupyter notebooks, and the forecasting process was tested against historical data to get an idea of the forecast accuracy.
Through this project, Duty Free Americas was able to assess their current processes and identify areas that needed improvement. “Working with 1Strategy was very helpful in measuring and validating our current systems and gave us confidence to move forward. We are evaluating other ways we can expand our inventory management capabilities leveraging this technology,” said Josh Adler of DFA.
About the Partner
1Strategy is an Amazon Partner Network (APN) Premier Consulting Partner, focusing exclusively on Amazon Web Services (AWS). 1Strategy helps businesses architect, migrate, and optimize their workloads on AWS, creating scalable, cost-effective, secure, and reliable solutions. 1Strategy also helps customers get real value from their data using comprehensive machine learning models and artificial intelligence. 1Strategy holds the AWS DevOps, Migration, Data & Analytics, Well Architected, and Machine Learning Competencies, and is a partner of the AWS Public Sector Program. 1Strategy was one of the initial ten AWS Partners globally who was qualified and authorized by AWS to conduct a Well-Architected Review and is among the top Well Architected partners in the AWS eco-system. With experts having deployed AWS solutions since 2007, 1Strategy is a leader in custom training—providing customers with the knowledge, tools, and best practices to manage those solutions over time. 1Strategy is a TEKsystems Global Services company with teams in Seattle and Salt Lake City, supporting customers throughout the US and across every vertical.