Supervised learning by definition is machine learning using data that is labeled. But what does that mean? First you must understand that machine learning is a type of artificial intelligence that takes in historical data and allows programs to predict outcomes without being explicitly programmed to do so. For instance you can use a linear regression model to predict how well a movie will perform at the box office by inputting data from similar movies released in the past. Similarly, you can use a logistic regression model to predict the likeliness a person has a heart attack. These are powerful tools that are able to help solve a number of real world problems.
Unsupervised learning uses similar machine learning algorithms to analyze and cluster data. These algorithms just find hidden patterns in data without the intervention of humans. These methods are used a lot in recommendation systems such as "Customers also bought this:".
The goals for each type are different as well. For supervised we can usually expect what type of results we will get and is used for predictions. While unsupervised focuses more on figuring out what is different and important in the data set.
Supervised learning can easily be done with languages like R and Python. Unsupervised problems tend to take powerful machines and need complex computations due to the size of data sets needed in order to generate good results.
So the question comes "Which learning should I use?". And while the answer can get complicated fast, it's best to first take a look at the data you are using. Is it labeled? Or do you or a colleague have the capabilities to label it? What problem are you trying to solve with the data you are using? And what algorithms would support your data?
Top comments (0)