Build a durable understanding of data preparation, features, models, pipelines, tuning, evaluation, and generalization with practical scikit-learn experiments.
Turn data into evidence
Machine learning is more than fitting an algorithm. It is a process of asking questions with data and testing whether the answers survive new examples. LearnMLFast connects models, experiments, and evaluation so you can produce results you can explain, compare, and defend.
Experience behind the lessons
Joe Chang brings more than 20 years of software engineering experience across startups, research, international banking and FinTech, entrepreneurship, and production systems. A physics graduate of New York University, he has taught more than 1,000 students across machine learning, deep learning, computer vision, reinforcement learning, and large language models. He has also worked with universities and government programs on AI engineering education.
Learning paths
Move from data and evaluation foundations into supervised learning, unsupervised learning, model selection, and practical scikit-learn workflows.
See what it really means to learn from data: models, objectives, training, optimization, generalization, and where machine learning fits within AI.
Go learning pathTurn raw tables into trustworthy inputs with sensible features, scaling, encoding, missing-data strategies, leakage prevention, and repeatable pipelines.
Go learning pathShape what a model can see by transforming, selecting, combining, and compressing features while preserving useful signal.
Go learning pathExplore how different models turn inputs into predictions, from linear and logistic methods to neighbors, SVMs, and probabilistic classifiers.
Go learning pathFollow how trees turn data into rules, then see how forests, bagging, boosting, and stacking change flexibility, stability, and predictive behavior.
Go learning pathLook for structure without labels using clustering and dimensionality reduction—then learn when the patterns you discover are actually meaningful.
Go learning pathChoose models without fooling yourself: build baselines, split data honestly, cross-validate, tune carefully, and keep the final test truly final.
Go learning pathGo beyond one score. Measure the errors that matter, inspect where models fail, and reason about imbalance, thresholds, calibration, drift, and uncertainty.
Go learning pathTake a fitted model beyond the notebook: preserve the full prediction workflow, validate new inputs, and produce repeatable inference.
Go learning pathFeatured tutorials
These tutorials connect the learning problem, data, model choice, and evaluation before you optimize algorithms.
You've seen the demos. A neural network identifies objects in photos, translates speech in real time, and writes fluent text. Meanwhile, someone keeps…
Read tutorial
AI, machine learning, and deep learning are not rivals competing for the same title. They are nested categories, each living inside the next.
Read tutorial
XGBoost has a reputation problem. It gets treated like a secret weapon—a mysterious algorithm that wins competitions and powers production systems while…
Read tutorialLatest
Fresh lessons on classical models, data preparation, evaluation, diagnostics, and scikit-learn workflows.
AdaBoost's algorithm listing looks like three separate rules stitched together: fit a weak learner, compute a coefficient, multiply the sample weights by…
Read tutorial
Anomaly detection is not "find the weird rows." It is a decision about what counts as normal, made before you ever run an algorithm. Get that decision…
Read tutorial
If you already understand decision trees, you know the dilemma: a deep tree memorizes the training data and fails on new data, while a shallow tree is too…
Read tutorial
You trained a model. You saved it. Now a new file of rows arrives, and each one needs a prediction. This is where a saved model proves its worth—or quietly…
Read tutorial
Retrain the same model on a different random split and the predictions move. Nobody changed the code. Nobody changed the hyperparameters. The wobble is not…
Read tutorial
Two models can post nearly identical test scores and still fail for opposite reasons. One misses because it never learned the pattern. The other misses…
Read tutorialBuild a machine-learning foundation you can actually use. The Machine Learning Starter Pack takes you from understanding how learning from data works through preparation, features, model families, scikit-learn workflows, validation, evaluation, diagnostics, unsupervised learning, and reusable predictions—so you can move from fitting your first model to understanding whether it works, why it works, and what to do next.
You’ll receive the bundle by email. You can unsubscribe anytime.
