Two ends of the same road

Data science and machine learning aren't separate worlds — they're points on one continuum. Classical analysis asks "what happened and why?"; machine learning asks "what will happen next?" Both start from the same place: a clear question and clean, well-understood data. Everything you've learned here — asking sharp questions, EDA, cleaning, honest stats — is exactly the groundwork a model needs.

Description vs. prediction

• Descriptive / explanatory analysis (this course) explains the past and present: which factors relate to churn, whether a change moved a metric. The output is understanding.
• Predictive analysis (machine learning) uses patterns in past data to forecast new cases: will this customer churn? The output is a prediction.

You usually do the first before the second. EDA tells you which features matter; understanding the data tells you what a model can and can't learn.

Why your skills transfer directly

A model is only as good as the data and questions behind it:

• Clean data — a model trained on messy data learns the mess.
• Good features — the variables you understand from EDA become the inputs that make or break a model.
• Honest evaluation — your skepticism about samples and noise becomes train/test discipline and avoiding leakage.
• Clear questions — a vague goal yields a useless model just as it yields a useless analysis.

Your next step

When you're ready to move from explaining to predicting, the Machine Learning Foundations course picks up exactly here.

Machine learning isn't a different skill set bolted on top — it's data science pointed at the future. Master the question, the data, and the skepticism, and you've already built most of what good modeling requires.

Try this: Take a question you'd answer descriptively ("which customers churned, and what did they have in common?") and rewrite it as a prediction ("which current customers will churn?"). That small reframing is the exact bridge from data science into machine learning.