From Trainees to Excellent Business Data Science Consultants

Exciting to experience how eager our consultants are in expanding their knowledge beyond the textbook. They have truly entered the fast lane-mode accelerating their competences.


It takes a bit more than a degree in Data science to become a great business data scientist consultant. 

Rooftop’s first two trainees; Tobias Berg Egelund and Kristian Lorentzen have successfully completed the training period and are ready to take on client assignments together with our experienced Business Data Science team. 

During their training, they followed Rooftop’s Best Practice on how to disentangle a business idea, apply domain knowledge and deliver a great data science solution in our pilot project. Here, we aim for being able to predicting ‘duration until sentencing’, as well as the ‘length of a court sentence’ based on court sentences from the Danish Prosecution's Knowledge Base.

Parts of this data was already made available in HTML-format, but lacked sufficient quality for our ‘end purpose’. Hence, Tobias and Kristian developed OCR techniques, which improved the quality substantially. Then, they extracted and structured
the data, so we can proceed with developing the predictions by using machine learning.

The case is very exciting, and we continue to the next step, where we will be applying weakly supervised machine learning, e.g. Snorkel, as our data science technique to select and label relevant entities and entity-relations of importance for the prediction task.
In essence, it means that we, based on type of crime, specific case information and wordings of the charge from the court sentences aim to predict the length of a court sentence as well as the duration until sentencing. 

We expect to accomplish good results, which supports the Start-up company Ilas ApS’ further in their next development phase.