What is Supervised Learning?
All machine learning problems in general can be classified into two categories:
Supervised Learning and Unsupervised Learning.
Supervised learning problems are essentially ones where the right answer is known or expected. The data set provided to the learning algorithm includes a relationship between the input and the correct output.
Supervised learning problems in turn are categorized into regression and classification problems. A regression problem is one where the input is mapped to a continuous output function. Meaning the correct output is a distinct value but it could be ‘any’ distinct value. An example of a regression problem is the price of houses based on the size. Given a particular data set of house sizes, the prices of those houses is a continuous function.
A classification problem is one where the input is mapped to a discrete set of output values. In other words, we are trying to classify the output into distinct categories. Determining whether a tumor is benign or malignant is an example of a classification problem. Another example is, determining whether a house will sell for more or less than the asking price. Here the input is mapped to a specific correct output (characteristic of a supervised learning problem) but the range of what those output values could be is defined, and limited.
Next week we will explore Unsupervised Learning, a different form of machine learning where there is no correct output and we’re looking to identify patterns.