Over 100 data science students followed the lecture at the TU/e, hosted by Kevin Schaul and Lars van Geet, on best practices and lessons learned about real world use cases using traditional machine learning and deep learning techniques. Explaining that although data science is called 'the most sexy job', most of the work is data wrangling (data is cheap, labeling is expensive). It means hard work to get it to work on a day to day basis. Did you know that data scientist spend around 60% of their time on cleaning and organizing data?

It all starts with the domain expert

For every project we start with defining the business problem of the domain expert. When the current situation is summarized, it is important that our perception of the business situation is correct and validated. The problem is then translated into solutions (decomposed into sub tasks) getting results fast in order to fail fast. 

This workflow - starting small, working agile, but always with the end in mind - focuses on an operationalized, working model that is used on a day-to-day basis. Learn more about real world data science use cases and download the handout of our lecture at the TU/e. 

If you are also interested in neural networks, read the blog from last year's lecture.