Life as a Data Scientist at Signal
by Danai Margariti, Associate Data Scientist
I chose to start my career at Signal, as I was impressed by the fact that it is one of the first technology companies in Greece that fuses data science with shipping data to improve the efficiency of one of the most robust and traditional industries in the world: commercial shipping.
When I was studying Applied Mathematics & Physical Sciences, my favorite subjects were those that contain real datasets analysis. I always wanted to apply the theory that I was learning, so I knew that I was going to choose this field as it combines powerful mathematical tools with the ability to end up in a meaningful result. Data science, especially in a technology company, is fascinating because of the vast amount of information and the fast, iterative, pace of change.
Today, I am an Associate Data Scientist, currently working in a team that develops products for the container market. What is a typical day at Signal you might wonder? Well, there is no such thing as a typical day when you are working in an exciting industry that continuously evolves with real-time data. Of course, some aspects of the day remain the same, such as working with data science models, working with people, and always trying to keep up with the field.
Signal: A Data Scientist’s Paradise
For every new feature that is released on The Signal Ocean Platform, there is a lot of work that needs to be done in advance. I will focus on my part, the Data.
At Signal, we are using various databases to store our data. Specifically, in my team, we are working with both SQL (relational database management system) and NoSQL (document-based database) databases, like MongoDB, since the product deals with a plethora of different data sources that are essential to reach our objectives.
As a company, every day we receive and process a vast amount of AIS data (AIS is an automatic tracking system that uses transponders on ships) to track every vessel across the globe along with commercial private data from our clients that are combined with state-of-the-art algorithms to turn them into powerful knowledge for commercial shipping. As one could guess, in several cases, the data can be characterized as geo-spatiotemporal data, where one needs to develop special skills to handle them.
Signal is a big fan of open source technologies where we use and extend open-source geographic information systems to effectively manipulate data to meet our needs and efficiently address the business challenges that come up. Signal has a lot of structured and unstructured data from different kinds of sources and it is a data scientist’s little paradise.
Working With Data, Data Everywhere
I am embedded in a cross-functional product team, which means I work most closely with the product manager, engineers, and data scientists. Every day begins with our daily stand-up meeting, where we share what we did the previous day, what we are up to today, and if there are any blocking issues.
Then, I check the logs of the application I am working on and if there is nothing wrong, either I continue with a task I have or start a new one, always using Jira, a proprietary issue tracking software, and begin my day.
As a Data Scientist, one first step is to try to get an idea of what the characteristics of the market are and extract your data based on them. The process of data science starts by having the right direction to find what the problem is. Then, to analyze the problem, you need to start asking a lot of questions, but of course, before that, you have to get an idea of the business by getting — really — involved with your data. This includes data curation and, in general, data manipulation, cleaning and eliminating errors from the data. It is imperative to ensure data accuracy and completeness, as we want the data to be as good as possible because these are going to compose the input for our models.
The goal is to end up in a conclusion using an algorithm that is adaptive and robust in every market condition, with low variance in its decisions.
By interacting with different cases every day, you can gain an understanding of the market, propose your ideas, and of course take initiatives to end up in the desired result. From a Data Scientist’s view, the desired result is the modeling of the market, analyzing the data to uncover hidden patterns and trends and define algorithms to answer specific questions and challenges.
What is great about working with shipping data is that the more understanding you have of the market, the more you can appreciate the existing data and the results you can produce with them. This requires very good scientific and technical knowledge as well as good business understanding. It is a unique combination, out of which powerful products can be created.
Working at Signal made me realize how much of an impact a Data Scientist can have in the shipping industry!
###
For more tech related articles, click here to follow Signal’s Tech Blog on Medium.
