In machine learning, there are two important categories- Supervised and Unsupervised learning. As you can see in ML, the machine learns from the data that is available to it. In a way, we are training the machine to achieve desired results. Now, what is the difference between traditional mathematical models and machine learning models? Suppose, you have a regression problem; Y equal to A plus BX. When you solve it using a mathematical model; with the given input data, you will get the same output or result. There is no learning because Y equal to A is the intercept and B is your slope, Y should be the dependent variable; so, everytime you give the data it gives the same result over and over again. This is not an intuitive model, unlike the machine learning model. In machine learning, we train the model with the available data till the desired output is achieved.
Supervised learning, an algorithm learns from a training dataset. We know the correct answers or desired output, the algorithm makes predictions using the given dataset and is corrected by the “supervisor”. The learning stops as and when the algorithm achieves a level of performance which is acceptable. There are two types of supervised learning- regression and classification.
Regression models are applied to solve various problems; e.g. predicting stock price.
Now, what is unsupervised learning? In unsupervised learning there is no trainer or “supervisor” as the name suggests. We do not show the output, or the specific input required to achieve specific output. The machine learns based on its own capability, it decides what should be the input and output- clustering is a method of unsupervised learning, where the machine learns on its own. A well known algorithm in clustering is k-means clustering.