Chapter 1 Introduction to Statistical Learning

Statistical Learning is a field at the intersection of statistics, computer science, and data science that focuses on developing and understanding models that can learn from data. While statistics has long dealt with model building, validation, and prediction, the dramatic growth in the volume, complexity, and dimensionality of data has strained many classical techniques. The introduction of sophisticated algorithms, data-management tools, and optimization methods from computer science has enabled more powerful approaches suited to modern data regimes.

In the supervised setting, learning from data typically means that, given an outcome (quantitative or categorical) and a (usually large) set of features, we construct a prediction model (also called a learner) that can accurately forecast the outcome for new, previously unseen observations.

 

We can separate statistical learning problems in two types:

  • Supervised Learning: when an outcome variable is present to guide the learning process. The two primary examples are:
    • Regression: Response is quantitative (continuous or discrete numeric).
    • Classification: Response is categorical (binary or multi-class).
  • Unsupervised Learning:when an outcome variable is not present to guide the learning process. The goal is to discover latent structure or patterns in the data (e.g., clustering, association rules, hidden Markov models, etc.).