Data is mainly unstructured (more than 80%) and the size of them is increasing. They are documents, graphs, images, videos. Finding the data pattern, time spent to do this process is getting complex and more via humans.

At this stage machine learning comes into action. This mainly helps the data handlers to learn the data or train the data in the minimum time. There are three major types of machine learning.

Supervised

Unsupervised and 

Reinforcement

Supervised- The first type in machine learning follows the task-driven approach where the labeled data is used to train the algorithms. Initially, learning algorithms accept the marked data as the inputs. 

Features denoted by X and the outputs Y supervised algorithm starts learning by comparison of actual production with correct outputs to find errors. Based on the error value, the whole model has been modified accordingly. 

Unsupervised- A second type in the machine learning algorithms where the unlabeled data are used to train the algorithm. This means the data has no historical variables. The main purpose is to explore the data and find the structure with that.  Since the data is unlabeled and without preprocessing, data division into train or test is not possible. 

Reinforcement Learning- The third stage in machine learning where there is no raw data is given as input. This reinforcement learning is frequently used in the streams like robotics, gaming, etc. This training consists of 3 main components such as agents, environment, and actions. 

The applications of machine learning algorithms are reflected in various areas such as medical, defense, technology, finance, security, etc.