Quantitative Biology Student Solves Churn - Propheto Case Study
Published: January 12, 2021
Nolan was in the final year of his Quantitative Biology PhD program. In between his days TAing, he spent most of his time swimming through unstructured data building machine learning models in Tensorflow. While he has spent years this way immersed in his research, he always wondered how perhaps he could apply these same skills in the real world, outside the walls of academia.
So when he came across Propheto, he felt it was the perfect opportunity to see what industry had to offer. As he observed from his first project experience:
"In my own research, I am very disconnected from experimental application and therefore the results of my work are not uncovered until months after I put it forward. It is very refreshing to work directly with a client and see them ingest the conclusions of my project and turn them into action within the span of a few weeks if not days."
Upon signing up, Nolan was soon sent an alert about a new project. At first, it seemed like a bizarre case for data science; an oral care subscription ecommerce business? A company that sells electronic toothbrushes with heads on subscription - what kind of data science needs would they have?
The company needed help identifying and understanding the signals that would indicate that a subscriber may churn. Nolan realized it was a classic statistics problem - how to predict if something is going to happen based on certain descriptive variables.
Within a week of getting matched on the project, Nolan was in the weeds exploring a new dataset, gaining familiarity with the client’s business, and making high profile presentations to executive leadership. While it was challenging, Nolan learned quickly and became an integral part of their team. There was a particular moment that Nolan recalled when he felt it all came together:
"On one occasion, a conclusion I reached through a quantitative analysis sparked an idea from a marketing officer that they claimed was a ‘million dollar idea’. To see their excitement and understand that it was rooted in my analysis, made my day."
Working with Propheto helped Nolan in a variety of ways. He improved his communication skills by distilling complex analyses for non-technical stakeholders. He sharpened his technical skills in SQL and Python. Above all, he’s risen to the level of technical and professional competence required to be an effective data scientist in an industry environment.
"I am learning a lot about how to communicate ideas in a non-academic fashion. I've learned that industry expects a more compact presentation of the concept, leading with the conclusion and then addressing the reasoning in a succinct manner. This is very different to what I am used to in academia... I've also picked up on a bunch of skills, in database systems, that I would not have had otherwise."
Nolan also recognized how unique the Propheto experience is. For a student that wants to explore work opportunities in industry, the PhD process is restricting. Propheto offered Nolan a way to overcome this obstacle:
"[Propheto] offers supreme flexibility especially for grad students who are juggling responsibilities in their own research and lab. They understand that the time commitments research requires changes and that the amount of time one can commit to projects wanes and waxes. I don't think I could get this level of understanding elsewhere."