I was fortunate to be in the organizing team as well as an attendee of a unique facet of Data Science Problem Formulation through LEGO Serious Play workshop by Women in Data Science (WiDS), Pune 2019 as a pre-conference event.
The learnings through this workshop were invaluable. Here are my takeaways-
1. Why most of the Data Science projects fail, is because the data scientists sometimes fail to put themselves in their customer’s shoes. Looking at the problem from the customers’ point of view can literally be the key here because previously unnoticed aspects might then be realized, and this could save up a lot of time as a whole.
2. Communication is the key. A Data Scientist must always be curious and ask the question ‘why?’ a lot many times. Speak with your team, experts, people with domain knowledge, market knowledge, business knowledge of the problem we are trying to understand and get a better understanding of the problem at hand.
3. Lego serious play is a method of fostering creative problem-solving approaches. It is a practical process for building confidence, commitment, and insight. The workshop used a kinaesthetic based approach to learning Data Science, where one learns by doing and learns fast.
4. Empathy is the link between Data Science and Lego Serious Play. Building visual three-dimensional Lego constructions bring the imaginary scenarios into reality, that gives one the ability to build rapport with the customer. When we see the situation from the customer’s perspective and gain new insight about it, we unlock the ability to empathize about it.
5. Lego Serious Play enables one’s thoughts to be seen, touched and heard, so we can comprehend it much better than we would otherwise. This can have tremendous impacts on the actions we take thereby. I experienced LSP as a Strategic Play, wherein one uses the constructs of play as a tutoring tool around a specific problem.
6. Formulating a Data Science (DS) problem is one of the most important parts of a DS pipeline. For this, it is necessary to ask questions from multiple perspectives. After thoroughly understanding the problem at hand, one should define the scope of the DS team along with the objectives and quantifiable targets. This was carried out through exercises of building from the Lego bricks and multiple stages of asking questions. After clustering into appropriate categories, a detailed data science problem was formulated.