Traditional approaches of software development may not always provide solutions for today’s applications that require Artificial intelligence. Programming mainly focuses upon rules and conditions with which we first implement the algorithm for a problem. Machine learning is the kind of AI which enables systems to learn and find out patterns by feeding huge amounts of data. It is all about making predictions and finding patterns. With more data being generated alongside more computing power, researchers have realized the potential of building applications which are not possible by traditional approaches.
There are mainly 2 types of machine learning models: Supervised and Unsupervised learning
1.) Supervised Learning: In this we already have the data set for a give problem along with the labels for the right predictions. This data can be structured like tables in a database or unstructured like image. But the important thing is that the right predictions are also available. So this is generally known as the training set. So once we train our ML model using this set, the new test set which will be new unseen examples will not have the labels and our model still predicts the right result. The predictions i.e. label values can be either be continuous or discrete. A graph for continuous predicted value is shown below, where we already have the prices for some of the house sizes. This is our training set. Now for our test the, the best fit line will predict the price value.