Machine Learning : Supervised Learning

A teacher in class give 2 kind of animals pictures, Dogs and Cats. Students will learn how to differentiate it, the ears, the eyes, the size, etc. At the end of class teacher give some test with another picture and students have ability to differentiate dogs and cats.

Yes, analogy above shows how Supervised Learning looks like. Machine learn from data that already labeled, data that already have classes or categories.

In formal way we can say that Supervised Learning created a function mapping model y = f(x). This is a mapping of input vector variables (x) to output variables (y). The model is made as accurate as possible so that when the machine is given new input data x', it can accurately predict the output variables y' for that data.

Supervised Learning is used for solve two kinds of problems, that is regression and classification. Classification problem is a problem if output/label in training set data looks like category, like price of home (expensive, middle, or cheap), body fat (fat, ideal or thin) etc. Example of classification algorithm is

• Decision Tree,
• K-Nearest Neighbor,
• Logistic Regression,
• Random Forest.

In the other hand Regression problem is a problem if the output is a number, like stock(USD), body weight (kg), and price of something (USD). Example of regression algorithm is

• Linear Regression
• Generalized Linier model
• Desicion Tree

Decision tree can be used for classifications and regression problem. And i will create the explanations for those algorithm in next article.