Guillaume's Blog - Data Science

In many situations, a business requires to simplify the customer base by excerpt some typical representatives [i.e. segments or clusters] that, put together, describe approximately the main trends within a customer base.  More than any other, Marketing departments need it to understand: Trends (what segments are growing/shrinking) Customer LifeCycle (Do I recruit the same customers?) What are the shopping/behavioral/spending/channel habits of my customers? Who is the most profitable group of customers (is it a function of my Marketing Spend?) Where are the Marketing dollars spend? Is it a good investment In brief, at a high-level, understanding the interactions between customers and the Brand The Analytics to tackle the issue are pretty simple conceptually. Yet, it is easy to get lost between all the techniques available. Let's make a rundown on what's out there. Overview: Family Name Description Use Drawback Upside Partitional Clustering K-mean

As it is customary to introduce oneself in a first post, I will tell you more about myself. Born and raised in rural Normandy, France, I am a 26yo Frenchie working in the United States. Growing up, I did my Engineering Undergrad in Lyon (2nd largest city in France) before transferring to Duke University for a dual Master Program in Engineering Management. My first full-time job was at Opera Solutions (winner of the Netflix Prize hosted on Kaggle) before joining AlixPartners 1.5 year later, a Turnaround and Restructuring boutique consultancy where I helped companies fix their Analytics. At AlixPartners, I specialize in Advanced Analytics applied to businesses. My work essentially consists in identifying algorithms that can be leveraged in business situations, in the vast majority, in the Marketing and Healthcare field. After that, I joined BCG's GAMMA team in March 2017 with a focus on Healthcare and personalization with the ambitious goal to fix Healthcare using Analytics f