Machine Learning in Python with scikit-learn

Lesson 1: What is machine learning, and how does it work?

This lesson gives a plain definition of machine learning, shows examples, and explains at a high level how learning from data works. It sets the vocabulary for the rest of the series. A notebook is available for download.

Play lesson

Lesson 2: Setting up Python for machine learning: scikit-learn and Jupyter Notebook

The lesson covers the pros and cons of scikit-learn, how to install a preferred Python distribution, and the basics of Jupyter Notebook. It also points to resources for learning Python if you are new to it. This prepares your environment for the rest of the path.

Play lesson

Lesson 3: Getting started in scikit-learn with the famous iris dataset

You load the iris dataset into scikit-learn and explore it. The lesson introduces key machine learning terminology and the four requirements for working with data in scikit-learn. This is the first hands-on step with a real dataset.

Play lesson

Lesson 4: Training a machine learning model with scikit-learn

This lesson uses a classification model to predict iris species. It explains how K-nearest neighbors works and walks through the four steps for model training and prediction in scikit-learn. You see a full training cycle from start to finish.

Play lesson

Lesson 5: Comparing machine learning models in scikit-learn

The lesson addresses how to choose which model is best. It covers the train/test split process for evaluating models and explains how this helps avoid overfitting. You learn to estimate how well a model is likely to perform on new data.

Play lesson

Lesson 6: Data science in Python: pandas, seaborn, scikit-learn

This lesson follows the data science pipeline from data ingestion with pandas to visualization with seaborn to machine learning with scikit-learn. It trains and interprets a linear regression model and compares three evaluation metrics. It shows how the tools fit together in one workflow.

Play lesson

Lesson 7: Selecting the best model in scikit-learn using cross-validation

The lesson explains K-fold cross-validation and how it is used for selecting tuning parameters, choosing between models, and selecting features. It compares cross-validation with the train/test split procedure and discusses some variations. This is a core technique for reliable model selection.

Play lesson

Lesson 8: How to find the best model parameters in scikit-learn

You learn how to search for optimal tuning parameters, also called hyperparameters, to improve model performance. The lesson starts with an exhaustive grid search using scikit-learn's GridSearchCV. It then shows more efficient search approaches.

Play lesson

Lesson 9: How to evaluate a classifier in scikit-learn

This lesson covers how to properly evaluate a classification model with common tools and metrics. It starts by showing the weaknesses of classification accuracy, then demonstrates other metrics. It also covers adjusting a classifier to match business objectives.

Play lesson

Lesson 10: How do I encode categorical features using scikit-learn?

Categorical features must be encoded numerically before use in a model. This lesson shows how to do that correctly with dummy or one-hot encoding. It covers OneHotEncoder and ColumnTransformer in scikit-learn.

Play lesson

Lesson 11: Machine Learning with Text in scikit-learn (PyCon 2016)

This is a long PyCon 2016 tutorial on working with text data in scikit-learn. It explains how to transform raw, unstructured text into data usable by machine learning models. The goal is to expand the amount of data your models can learn from.

Play lesson