What is Machine learning?
There is no universal definition for machine learning, but in simple terms, it is defined as a method of teaching computers to make a prediction based on some data.
In this learning approach, we apply some process or algorithms on data by using the computer so that the computer can help to provide the insight.
Two Categories of Machine Learning are:
1)Supervised Learning: Making a prediction of specific outcomes using data.It is also called as predictive modelling.
Example 1: if we have series of email messages as a data then, our supervised task will be to predict the new appeared email as spam or non-spam (ham). This task is called supervised because the algorithm has specific outcomes to predicts i.e. possible labels could be assigned to unknown data based on its features.
Example 2: Asking child, given fruit is apple or mango. Here child will learn about two fruits based on its features such as its colour, shape etc. Now child knows that for a given fruit its outcomes would be either apple or mango and nothing else based on its features.
Based on above two examples we can say that in supervised learning class labels (in example 1: spam or ham and in example 2: apple or mango)are known beforehand that we are trying to predict.
2)Unsupervised Learning: Extracting structure from data or can say learning how to best represent the data.
Example 1: In given store shoppers data we need to do Segmentation of store shoppers into groups of clusters which exhibit similar behaviour. Here we might find college students in one cluster, parents are in other & so on based on their behaviour but specific outcomes are not known beforehand.
Example 2: If you are given some animals, the task is to categorise animals into the group of cluster based on their similar features (eye, nose, ear, height etc) where it is unknown that given animal is dog, cat or etc. That means we make the group of cluster based on its similarity features where initially class labels are not known.