Data Science and Machine Learning at Mærsk

Hi everyone!

We'll be gathering at Mærsk to hear about how they are working with modern data science and machine learning techniques to extract value from the oceans of data they are collecting troughout the business.

The program is as follows:

17-17.30 Digital forecasting products in Mærsk

17.30-18.00 Break with food and drinks

18.00-18.30 Supply and demand in tanker shipping

As always, there'll be time for networking and hanging out afterwards.

Looking forward to seeing everyone there!

---- UPDATE: Full talk abstracts ----

Digital forecasting products in Mærsk

Maersk is currently undergoing a major digital transformation, a key component of which is the shift away from purely subject matter expert driven decision making, towards a higher degree of automation and data driven decisionmaking. A critical element in the success of a data driven decision making process is an accurate forecast,communicating the relevant expectations and risks associated with the future. There is a significant appetite for forecasts to support decision making in business units across Maersk T&L, to support processes in the realms of operations, finance, customer intelligence. In this talk we will discuss examples of specific forecasting applications which from those that are in the preliminary stages of development to those which are fully integrated into the daily operation of the business. We will delve into the methodological differences in the forecasting approaches employed, highlighting how the applied methods must evolve as the business becomes more mature in its adoption of data driven processes, and demand shifts from simple point forecasts to more complex probabilistic information to support a more sophisticated decision making process.

Supply and demand in tanker shipping

The tanker industry is a very conservative industry, and Maersk Tankers is the oldest still active business unit in the Maersk Group. This conservatism means that there is ample opportunity to revolutionize the industry by incorporating advanced analytics into the daily workflow of the frontline units. Maersk Tankers is determined to continue to pioneer this area by a devoted focus on making data driven daily business decisions on all levels in the organization.

In this presentation Christian Schultz, Quantitative Analyst at Maersk Tankers, will give a brief presentation of the existing forecasting techniques employed in Maersk Tankers. The forecasting is primarily done by an ensemble model consisting of Gaussian Processes, Elastic Nets and Support Vector Regression. Furthermore some of the interesting projects in the pipeline will be discussed, among others the effort to build an Automated Cargo Identification System and the Automatic Voyage Predictor Model based on highly detailed historical vessel positions from the global tanker fleet.