GeekCoding101 make your way to geek

Supervised Machine Learning: Course Notes, Day 1 to 12

9 posts on supervised-machine-learning, newest first.

9posts
2024first post

These are my day-by-day notes from working through Andrew Ng’s Supervised Machine Learning course. They run from the first lectures to overfitting and regularization. If you’re taking the course too, start at Day 1 and read forward. If you need one topic, use the roadmap below to jump straight to the right day.

I started the course while building an AI-powered knowledge base chatbot on Django, LangChain, OpenAI, Chroma and Gradio. I wanted to understand the layer underneath the frameworks I was using. Some days were long and some got 15 minutes before bed. The notes are honest about both.

Course roadmap: what each day covers

DayMain topicsPost
1Why I started the course, coming from the AI application layerSupervised Machine Learning – Day 1
2–3First course videos, plus 3Blue1Brown’s “But what is a neural network?”Supervised Machine Learning – Day 2 & 3
4–5Multiple features, vectorization (NumPy)Mastering Multiple Features & Vectorization
6Reviewing earlier notesSupervised Machine Learning – Day 6
7Feature scaling, checking that gradient descent convergesMaster Feature Scaling & Gradient Descent
8Choosing the learning rate, the Jupyter lab, feature engineeringMaster Learning Rate and Feature Engineering
9Gradient descent revisited, binary classificationMaster Gradient Descent and Binary Classification
10Decision boundary, logistic regression cost function, logistic loss, gradient descent for logistic regressionGrinding Through Logistic Regression
11–12Overfitting vs underfitting; fixing it with more data, simpler models or regularizationOverfitting! Unlocking the Last Key Concept

Where to start

  • New to the course: read Day 1 to Day 5 in order. They set up the vocabulary and the vectorized way of thinking the rest relies on.
  • Stuck on gradient descent: go to Days 7–9 for feature scaling, the learning rate and how to tell whether training is converging.
  • On the classification part: Day 9 introduces binary classification. Day 10 works through logistic regression end to end.
  • Model fits training data but fails elsewhere: Days 11–12 cover overfitting, underfitting and regularization.

FAQ

Which course are these notes from? Andrew Ng’s Supervised Machine Learning course. The notes follow it day by day, from the first lectures through overfitting and regularization.

Do I need to take the course to follow the notes? They make the most sense next to the videos and labs. Each post also summarizes the key ideas, such as feature scaling or logistic loss, so you can use one as a quick refresher on a single topic.

In what order should I read them? Oldest first, starting at Day 1. The list above is newest first, so use the roadmap table if you want course order.

Is this free to read? Yes. Every post on GeekCoding101 is free to read and there is no sign-up.