Machine Learning: An Introduction

Hello guys, this is a new series in which I'll be talking about "Machine Learning" as it is one the most booming fields in the IT industry and as a student of Information Technology, I should try to share my knowledge on this topic.

Here I'll be discussing all the major topics and their implementation and some theoretical concepts behind them. Also, I'll be providing links to various topics and documentations from where you can learn about them as there are already many excellent explanations scatter throughout the internet. So, without further ado, LET'S GET STARTED...

What is Machine Learning?

It is an act of making computers learn something without being explicitly programmed. They learn from the data that we provide them. The data might be labeled or non-labeled.

There are several steps involved in Machine Learning:

1. Collecting Data: In this step, we collect the necessary data that will be required to build the model. The variety, density, and volume of data are essential for a better and more generalized model.
2. Preparing: After we have got the desired amount of data at our disposal we have got to prepare that data i.e. to eliminate unnecessary information, fix missing values, modify its structure, etc.
3. Training the model: After we are finished with the pre-processing part we can start training the model using that data. We can use various algorithms for different purposes. Each algorithm has its own advantages and disadvantages. We split our data into two parts, training data on which we train the model and test data on which the model is tested.
4. Evaluating the model: After we are done with the model, we need evaluate its performance on a set of data which it was not trained on. There are various metrics that we can use to evaluate the performance of the model.
5. Improving the model: Last but not the least we need to check if we can improve the existing model and get accuracy. This can be achieved by tuning the existing algorithm or by using a completely different algorithm. We might also need to choose better features to train the data on.

What are the types of  Machine Learning?

Supervised Learning

This model tried to predict the outcome of a given input using historical data. This type of model is given clear instruction during the training phase about the data that it is being trained on generally known as Labeled Data. It learns from that labeled dataset and gives out the prediction.

It can be used to predict house prices, customer choices, future sales, etc.

Examples of supervised learning algorithms: Linear Regression, Multiple Regression, Naive Bayes, Decision Trees, etc.   

Unsupervised Learning

It is used to train models on data that has no labels or target. For example if you want to find out some groups in a set of data like a combination of products that customers tend to buy.

Examples of Unsupervised Learning Algorithms: K-Means Clustering, Apriori.

Reinforcement Learning

It is a model in which machine takes decisions for some particular situation and then evaluates itself to improve its performance and decisions. The idea behind this is that the machine trains itself continuously on the basis of the environment it is exposed to and applies its knowledge to solve various problems. One of the examples of this type of learning is self-driving cars that improve continuously over time.

The application of these algorithms are widespread and I'll be discussing some algorithms in subsequent posts. Hope you got some idea about machine learning from this post.

THANK YOU FOR READING :)