skip to main content
skip to main content
Global Search Configuration

How data analytics and machine learning helps the shipping sector thrive

More Data Storytelling

Although raw AIS data is the default go-to for information, it’s difficult to plan ahead when about 32% of vessels sail without a destination logged, 36% of vessels’ AIS transmissions do not include ETAs (Estimated Time of Arrival). Even when those transmissions do include ETAs, 27% of vessels arrive late. Information inaccuracies lead to wrong judgments and cause significant logistical, competitive, and financial challenges — including spoiled cargo, wasted fuel, excess pollution, or additional demurrage and detention (D&D) charges.

The value of the AIS data, or any information for that matter, is in its accuracy. The more accurate and reliable the information, the better equipped service providers will be, to target prospects and make the best possible business decisions.

Achieve greater prediction accuracy through an improved model
Predictive Fleet Analytics combines Lloyd’s List Intelligence’s foundational data with machine learning to achieve the highest level of certainty about future vessel activity.

Most probable destination predictions show at least a 35-percentage-point increase in accuracy over standard AIS data. Taking a snapshot of key vessel types, Predictive Fleet Analytics’ predictions were 7 times more accurate for bulk carriers and oil tankers, 5 times more accurate for chemical tankers and 6 times more accurate for general cargo, when compared to raw AIS data input by the crew. 

Vessel Type

Predictive Fleet Analytics (predicted by the model)

Automatic identification systems (AIS)

Added value by Predictive Fleet Analytics

Predicted Destination1

Predicted ETA2

AIS Destination3


Predicted Destination is __ times more accurate than AIS

Predicted ETA is __ times more accurate than AIS

Accuracy 5 days out

Error (+/-) 5 days out (in hours)

AIS Accuracy

Error (+/-) 5 days out (in hours)

Bulk Carrier







General cargo














Chemical tanker







Oil tanker







Liquefied gas tanker








1: Predicted destination accuracy derived by comparing the place predicted by the model with the port call of the vessel for a random sample of vessels.

2: Predicted ETA accuracy derived by comparing the prediction with the date and time the call started.

3: AIS destination accuracy derived by calculating the number of times the AIS Destination was matched to the vessel's actual destination on a year's worth of data.

4: AIS ETA accuracy derived by comparing the first ETA from AIS messages on the 5th day before the port call, and the arrival time at port.


Examples of real vessels tracked using Predictive Fleet Analytics

Vessel 1 – Erietta IMO 9713959

This Liberian flagged bulk carrier was tracked having left La Planta Anchorage, Argentina, for an unknown destination. Although the AIS transmission did not show her intended final port or ETA, Predictive Fleet Analytics was able to accurately predict this ship was headed for Singapore, 9 days out to port.

Predicted by the model: (9 days out) Singapore arrival on 11 September, 12:22 GMT 
Result: Erietta arrived at Singapore East Anchorage, on 10 September, 11:01 GMT

A bunker trader who pinpoints Erietta as a potential new trade is able to see where this ship is likely to arrive, at what time, as well as previous ports of call and draught changes, which indicate whether fuel is required. Predictive Fleet Analytics’ ETA accuracy combined with useful ETB (Estimated Time to Berth) and ETD (Estimated Time of Departure) predictions, means the trader can be confident when making an approach that they know the timescale in which this delivery will need to be made and completed. 

Vessel 2 – Winning Nature IMO 9659866

This Singapore-flagged bulk carrier was tracked having left Port-Kamsar Anchorage, Guinea, without a destination shown through AIS. The Predictive Fleet Analytics model was able to provide more clarity by suggesting the most likely destination was Yantai Port, China, while the vessel was still 25 days out.  The model also predicted a one-day wait to enter port and a four-day stay with a predicted departure date of 11 September.

Predicted by the model: (25 days out) Yantai Port, China on 8 September, 04:35 GMT 
Result: Winning Nature arrived at Yantai Port, China, on 8 September, 18:59 GMT

A fleet operations manager who tracks the Winning Nature will have a clear picture of its arrival time more than three weeks before it arrives in port. They will know that a wait time of four days is predicted, so his team can request that the captain slow down in advance to save fuel and limit emissions. Congestion levels at specific ports can also be closely monitored continuously with operations decisions made to maximise efficiency and prevent demurrage and detention fees.

Vessel 3 – Julius Caesar IMO 9912244

Departing Galveston Anchorage, USA, on 30 July 2022, this Marshall Island-flagged crude oil tanker did not display a clear destination via AIS data and its ETA was shown to be 20 September. Almost one month before her arrival, Predictive Fleet Analytics was able to accurately predict the destination port as Ulsan, South Korea, and offer an arrival date accurate to the same day. The predicted wait time was one day.

Predicted by the model: (24 days out) Ulsan, South Korea arrival on 26 September, 23:47 GMT 
Result: Julius Caesar arrived at Ulsan, South Korea, on 26 September, 23:33 GMT

A logistics operator who tracks Julius Caesar using Predictive Fleet Analytics would have clear foresight about when this delivery of crude oil will be at Ulsan, one month before it arrives. This enables them to keep refineries and chemical companies waiting for this product informed, and aids onward transportation planning. The ETA shown through AIS was six days out, compared to the prediction by the model which was only minutes different to the actual arrival time. The additional ETB prediction also gives this logistics operator an idea as to whether the vessel will be delayed getting into port.

In summary, accurately predicting vessel arrival, berthing and departure times aids operations, helps boost sales, and increases supply chain efficiency. Predictive Fleet Analytics delivers insights you cannot find elsewhere:

  • Destinations predicted with 70% accuracy — at least a 35-percentage-point increase in accuracy over standard AIS data. 
  • One-stop platform and easy-to-use interface provides accurate data in a format which allows situations to be swiftly assessed
  • Reliable foresight enables efficient planning and boosts new business potential


Watch our new maritime analytics solution in action, specific to your workflows: 

Watch product overview videos

Thanks for getting in touch

We have received your request and we will get back to you soon.

Terms and Conditions

Privacy Policy    Terms & Conditions

When submitting a form, we request some personal details, including a work email address. In addition to supporting your query, this data is used for marketing purposes by Lloyd’s List Intelligence. This processing is carried out based on our legitimate business interests. Being able to promote our events, products, and services to interested parties allows us to grow and develop our business.

You can unsubscribe at any time by following the links in any marketing emails we send to you or by emailing For full details on how Lloyd’s List Intelligence processes your personal data please see our Privacy Policy.