Supervised Learning

Welcome to the third chapter in this course! Assuming you have finished the first two chapters, including the Titanic lesson, you should have some confidence in loading and visualizing data and training scikit-learn's models. However, you have already seen some words that didn't make a lot of sense, including "f1 score" and "overfitting".

What you will learn, among other things, are:

• How to measure performance of classifiers.
• Learn internal workings of the training progress, such as the definition of cost function.
• What is a sigmoid activation function
• How to perform Stochastic Gradient Descent (the workhorse of both traditional ML and deep learning)
• How to optimize hyperparameters using cross validation and grid search
• How to visualize various learning processes (learning curves)
• How to increase model complexity using polynomial features
• …and how to avoid adding too much.

After this chapter, you will have a rough understanding of bias, overfitting and estimator selection. You should be able to select and compare fitting estimators (algorithms) for any classification or regression problem you might face.