What is machine learning?

This is the introductory post in the series of articles on machine learning techniques that I am going the share in coming days. My objective is to help you understand the domain of machine learning and the different tools that are available for us to solve real life problems.

Lets enter into the worlds of Machine learning!!

If you are already familiar with some basics of machine learning then you may skip this article and look at other technical articles. However, I would suggest that you quickly scan through this article just to be clear about the direction we are heading towards.

Machine learning can be understood as a broad set of data driven tools and algorithms that are used to make prediction, classification, or segmentation based on the data. Unlike conventional algorithms, their output is data driven.

For example, you can build a house price prediction model which can be trained using the past trends and prices. Now, you can predict the price of a house based on information available about the house. Similarly, you can build spam classification system to classify emails as spam and not spam.

In simple words, you can train a machine with huge set of data and then get results (something that you don’t know) based on the input that you give to it.

One of the questions that you may have now is, how to decide about the correctness of the model (or machine learning algorithm)?

Well, you can generate the accuracy score of the model and determine how well the model is able to solve your problem. Accuracy score is the ratio of the total cases that are correctly predicted to the total number of cases.

When we develop a machine learning algorithm, we divide the data into two parts, train and test data. The train data is used to train the model and the test data is used to test the accuracy of the model. The higher the accuracy the more reliable is the model.

There are two broad categories of machine learning algorithms: Supervised and Unsupervised learning. Initially, in this series we will look at the supervised learning and then we will explore the unsupervised learning.

Supervised learning algorithms are the one where we provide both the input variables as well as the output variables. Here, the objective of machine learning algorithm is to predict the output value for the set of input values. On the contrary, unsupervised learning requires only input values. Here, the algorithm itself figures out all the relations, features and behavior.

In the next article we will discuss about Naive Bayes Classifier.

Comments are closed.