Machine Learning Foundations
What learning from data really means, without the hype
6 lessons · 1.2 hours of reading
· Intermediate · 3 free
A first ML course built on honest explanations: regression from scratch, classification, overfitting, and the metrics that stop you fooling yourself. Includes when machine learning is the wrong tool.
Lessons
1
What learning from data means, and when ML is the wrong tool
The one idea behind every ML algorithm, and the cases where you should not use one.
2
Supervised vs unsupervised learning
The split that decides which algorithm you can even consider.
3
Linear regression from scratch, then with scikit-learn
Derive the line yourself in ten lines of NumPy, then get the same answer in three lines.
4
Classification, logistic regression and the confusion matrix
Predict a category, then look at the four numbers that describe every mistake.
5
Overfitting, train/test split and cross-validation
Why a model that scores 100 percent on your data is usually the worst one.
6
Why accuracy is a bad metric on imbalanced data
A 99 percent model that catches nothing, and what to report instead.
Lessons 4 to 6 need a Pro pass
A pass also opens every other course's paid lessons, all 50 mock test papers and the company-wise test series. It ends on its own date — nothing renews by itself and nothing is charged without you pressing a button.
See passes — from ₹29