Nobody predicted a year ago that all segments of the bulk market would be at US $ 30.000 / day, that is, three or four times more than twelve months ago. No one predicted either that the four segments would trade with a minimal difference from each other, which is unusual, says Alphabulk's weekly report.

But these days of darkness are possibly over, since Big Data and Artificial Intelligence, it is assumed, would allow us to predict the future. Or at least that's what industry players who have "tapped" both tools to see through the mess claim.

According to the consultancy firm, the benefits are not obtained by always getting it right, but by getting it right more times than by being wrong and minimizing losses when things go wrong. That is, risk management.

A proper risk management system is cheap to implement and will give immediate results, while the search for a perfect prediction tool can be expensive and elusive.

 It is also pointed out that market participants must always be students of their markets, which implies collecting data, cleaning it and classifying it with any tool -mathematical or statistical- available and that, today, they are stupendous.

But what is Big Data?

One definition would be: “…extremely large data sets that can be analyzed computationally to reveal patterns, trends, and associations, especially related to human behavior and interactions.”

Big Data is usually characterized by: A large volume of data; a wide variety of data; and a high speed at which much of the data is generated, collected and processed.

From the foregoing it follows that Big Data does not currently exist in maritime transport purely at the ship level. Some suppliers claim that billions of data related to the world merchant fleet are collected daily, which today consists of just over 100.000 vessels: tankers, gas carriers, container ships, heavy cargo ships, etc.

Well, the limited data figures derived from 100.000 vessels can only help to detect the following patterns: Ships tend to move in the water between ports, they are empty part of the time, plus some other minor patterns.

For example, finding out within hours of its departure that a Capesize leaving Brazil is heading for the Cape of Good Hope or for Europe can be useful, but does not respond to a Big Data application. Counting the number of ships waiting anchored in various ports and detecting a sudden increase or decrease in that number is very useful and does come close to a Big Data application. It is also immensely useful to detect a change in the average speed of the fleet as an increase in available capacity, as well as its decrease.

Arguably, Big Data in its strictest form does not apply to shipping, as the world's merchant fleet is too small to generate "extremely large data sets...". However, maritime transport depends on many variables, such as global trade flows and the global economy, which do generate “extremely large data sets…”, meeting the strict definition of Big Data.

Artificial Intelligence 

According to IBM, "Artificial Intelligence uses computers and machines to imitate the problem-solving and decision-making abilities of the human mind." In fact, today there are two development paths for Artificial Intelligence: the human approach, that is, systems that think like humans and systems that act like humans; and the ideal approach, that is, systems that think rationally.

Simply put, artificial intelligence is about combining computer science with large data sets (Big Data) to enable problem solving. It does this by relying on subsets such as machine learning and deep learning, which is why these two subsets are often mentioned alongside artificial intelligence.

Anomaly detection

These subsets are made up of Artificial Intelligence algorithms that aim to create expert systems that make predictions and / or classifications from the input data. The idea is, first of all, to teach the machine to react like a human, including learning capabilities. But according to alphabulk, the defect of Artificial Intelligence is that it is not actually artificial at all, since it is programmed by humans and, therefore, with all its defects.

However, one of the applications that is considered to provide added value is the so-called parsing of email. Brokers and directors have to deal with incredible email flows and here Artificial Intelligence complementing ML and NLP can be very helpful. A parser system Mailbox allows data extraction from incoming emails, and can be configured to extract specific data fields from incoming messages. This makes it possible to convert unstructured text emails into structured data that can then be used for any purpose.

And there is a new field in shipping where Big Data and artificial intelligence could also be very useful: the field of predictive and preventive maintenance. With the help of an on-board server connected to a series of on-board sensors, a remote analysis platform can create a digital twin of any vessel.

This “digital” vessel can then be permanently monitored by collecting thousands of pieces of data daily. This includes elements such as temperature, pressure or vibrations. In each of them you can establish alarms based on thresholds.

After a certain period of data collection time, the platform can start to function as an automated anomaly detection system that effectively uses artificial intelligence and machine learning models to monitor on-board equipment for faults, triggering human intervention before a breakdown occurs. Here you can find the perfect intersection between shipping, artificial intelligence and Big Data.