It doesn’t matter where you are or what you’re shipping—ETA is critical. Some say it’s the single most important data point in logistics. So why is it so difficult to get right?
Service days, routes, weather, train types, and more all have the ability to affect a shipment’s ETA. That means if your visibility platform is just taking an average of time passed between origin and destination from previous trips, it’s likely wrong more often than it’s right.
At TransmetriQ ‒ a Railinc brand, we’re studying the decades of data gathered by Railinc to determine how we can provide a better ETA. Railinc is the industry’s repository for rail data, and the only team with access to the millions of shipment lifecycles needed to study ETAs.
Ultimately, creating a more accurate ETA comes down to understanding what scenarios are likely to impact the shipment and when.
So how can technology predict events that haven’t even happened? The answer: Artificial intelligence (AI) sequence modeling. Sequence models predict future events in a time sequence, meaning they can iteratively predict events using historical data and make adjustments in near-real time.
In layman’s terms, sequence modeling can predict how say, a route change, will impact your ETA. Or, if your shipment moves through Dallas twice as fast as usual.
Looking at this data across millions of shipment lifecycles, AI is able to pick up on the patterns that matter while ignoring irrelevant irregularities.
While ETAs aren’t yet perfect, AI is the future of solving the challenge. The current TransmetriQ Advanced ETA offering is the third iteration of ETA from Railinc. There’s no way to improve ETA accuracy without studying every trip that has ever happened, and there’s no way to ingest and analyze that much data at scale without AI. That’s why Railinc’s network-wide visibility and historical data will be critical in continuing to solve the challenge of ETA.
