Machine Learning
Classic ML the way practitioners use it: regression, trees, boosting, evaluation, and scikit-learn.
Before this: Python & Math for AI.
-
7 min
Step 1Bias and variance
The trade-off behind every modeling decision.
Machine Learning Fundamentals: Bias and VarianceStatQuest with Josh Starmer
microsoft/ML-For-Beginners★ 92.3k -
27 min
Step 2Linear regression
The simplest model, and the one everything else gets compared to.
Linear Regression, Clearly Explained!!!StatQuest with Josh Starmer
-
24 min
Step 3Gradient descent
How models actually learn, step by step.
Gradient Descent, Step-by-StepStatQuest with Josh Starmer
-
9 min
Step 4Logistic regression
Your first classifier, and still a strong baseline.
StatQuest: Logistic RegressionStatQuest with Josh Starmer
-
6 min
Step 5Cross-validation
How to know if your model works on data it hasn't seen.
Machine Learning Fundamentals: Cross ValidationStatQuest with Josh Starmer
-
7 min
Step 6The confusion matrix
Accuracy lies. Precision and recall don't.
Machine Learning Fundamentals: The Confusion MatrixStatQuest with Josh Starmer
-
16 min
Step 7ROC and AUC
Compare classifiers across every threshold at once.
ROC and AUC, Clearly Explained!StatQuest with Josh Starmer
-
18 min
Step 8Decision trees
Readable models that split data into simple rules.
Decision and Classification Trees, Clearly Explained!!!StatQuest with Josh Starmer
-
10 min
Step 9Random forests
Many trees beat one tree. Still a top choice for tabular data.
StatQuest: Random Forests Part 1 - Building, Using and EvaluatingStatQuest with Josh Starmer
-
16 min
Step 10Gradient boosting
The idea behind the models that win Kaggle tabular competitions.
Gradient Boost Part 1 (of 4): Regression Main IdeasStatQuest with Josh Starmer
-
26 min
Step 11XGBoost
The library that made boosting fast and famous.
XGBoost Part 1 (of 4): RegressionStatQuest with Josh Starmer
-
8 min
Step 12K-means clustering
Find groups in data with no labels.
StatQuest: K-means clusteringStatQuest with Josh Starmer
-
6 min
Step 13PCA
Squash many features into a few you can actually look at.
StatQuest: PCA main ideas in only 5 minutes!!!StatQuest with Josh Starmer
-
2h 09m
Step 14Scikit-learn in practice
Put it all together with the library everyone uses.
Scikit-learn Crash Course - Machine Learning Library for PythonfreeCodeCamp.org
ageron/handson-ml3★ 14.5k
Build it
Watching isn't learning. Pick one and put it on GitHub.
- Enter Kaggle's Titanic or House Prices competition with a cross-validated scikit-learn pipeline.
- On an imbalanced dataset, plot precision and recall as you move the threshold, and pick one for a stated cost.
- Compare logistic regression, a random forest, and XGBoost on the same splits and explain the gap.