Understanding supervised machine learning

Machine learning focuses on using data-driven approaches to create autonomous systems that can help us to make decisions with or without human supervision. In order to create these autonomous systems, machine learning uses a group of algorithms and methodologies to discover and formulate repeatable patterns in data. One of the most popular and powerful methodologies used in machine learning is the supervised machine learning approach. In supervised machine learning, an algorithm is given a set of inputs, called features, and their corresponding outputs, called labels. These features often comprise structured data like user profiles, historical sales figures, or sensor measurements, while the labels usually represent specific outcomes we want to predict, such as customer purchasing habits or product quality ratings. Using a given dataset, a supervised machine learning algorithm is used to train a model that captures the complex relationship between the features and labels represented by a mathematical formula. This trained model is the basic vehicle that is used for predictions.

The ability to learn from existing data in supervised learning is similar to the ability of the human brain to learn from experience. This learning ability in supervised learning uses one of the attributes of the human brain and is a fundamental way of opening the gates to bring decision-making power and intelligence to machines.

Let’s consider an example where we want to use supervised machine learning techniques to train a model that can categorize a set of emails into legitimate ones (called legit) and unwanted ones (called spam). In order to get started, we need examples from the past so that the machine can learn what sort of content of emails should be classified as spam.

This content-based learning task using text data is a complex process and is achieved through one of the supervised machine learning algorithms. Some examples of supervised machine learning algorithms that can be used to train the model in this example include decision trees and Naive Bayes classifiers, which we will discuss later in this chapter.

For now, we will focus on how we can formulate supervised machine learning problems.