Torch: Season 1, Episode 5
From Witchcraft To Science – How Data Science Is Done
Can data scientists predict the future? As a data scientist, I know that most predictions are actually wrong. However, there is still some truth to this myth. In my Torch talk, I will show you how wrong predictions can still be useful for statistical forecasts.
In this torch episode, I will show you:
- How you can use historical data to calculate the probability of a customer to stay or leave.
- Why this probability – although not a certainty – is still a valuable basis for the prediction of customer churn.
- How this probability helps a company target customers purposefully.
Hi there, my name is Sacha Mourier and I am a data science senior analyst at Trivadis – Part of Accenture. I come from the realms of statistics – that is the subject I got my Master’s degree in. It comes as no surprise therefore that I am also interested in inference, optimization methods and business analytics. My heart does not only beat for data science, however: for 9 years, I have played professional tennis, first in France and then in the US. Originally from Paris, I am now based at Lausanne. Did you enjoy my Torch talk? Then let’s get in touch – I would love to pass the torch and discuss the world of data science further!
Takeaway
Using «wrong» assumptions to make predictions about the future – that is what my team and I did a while ago for a client from the public sector. How we estimated their future customer churn using machine learning and implemented this into a business intelligence solution? That is what you will find concrete answers to in my takeaway – based on our biggest learnings from this real use case.
Download now
MORE EPISODES
