Day 7 – Covariance & Correlation in Machine Learning

Hey everyone 👋 Dhairya here.

Before diving in — a small personal update: I had to take a 4-day break because I was feeling sick 🤒. I didn’t want to force myself and burn out, so I allowed my body to rest. I’m back now and ready to continue the challenge with full energy. 💪

---

🔢 What I Learned Today

Today’s focus was on Covariance and Correlation — two fundamental statistical tools that measure the relationship between variables.

• Covariance

  • Measures how two variables vary together.

  • Positive → variables increase together

  • Negative → one increases while the other decreases

  • Limitation: depends on scale (not standardized).

• Correlation

  • Standardized version of covariance (values between -1 and 1).

  • +1 → strong positive relationship

  • -1 → strong negative relationship

  • 0 → no linear relationship

  • Used a lot in ML for feature selection, multicollinearity detection.

---

🌱 Reflections

I always thought correlation was just "relationship strength," but now I see it’s directly tied to covariance — just normalized. This makes correlation much more useful in ML since it’s scale-independent.

Visualizing scatter plots and comparing covariance/correlation values made things click.

---

💻 Notebook

My Day 7 notebook (with examples & plots) is here 👉 GitHub Link – Day 7 Notebook

---

📚 Resources

🎥 YouTube

• StatQuest – Covariance and Correlation

• Khan Academy – Correlation and Causality

• 3Blue1Brown – Essence of Linear Algebra (Covariance Intuition)

🌐 Websites

• GeeksforGeeks – Covariance vs Correlation

• Towards Data Science – Correlation in ML

• Machine Learning Mastery – Statistics for ML

---

🎯 What’s Next?

For Day 8, I’ll explore Probability Distributions in ML Applications (how they’re used in actual algorithms).

See you tomorrow 👋
— Dhairya