Understanding Machine Learning Algorithms

In an earlier post, we had enumerated the differences between data science vs machine learning. Simply put, machine learning is a concept that allows the machine to learn from experience without being explicitly programmed. It will be worthwhile to deep dive into what are machine learning algorithms and also deal with some commonly used algorithms.

Essentially machine learning algorithms are the engines of machine learning as they turn a data set into a model. While there are various types of machine learning models, their use depends upon factors such as the nature of data and the computing resources available.

Types of ML Algorithms

There are broadly three kinds of Machine Learning algorithms. These include:

  • Supervised Learning Algorithms

  • Unsupervised Learning Algorithms

  • Reinforced Learning Algorithms

1. Supervised Learning Algorithms

In these algorithms, the training dataset comprises of two aspects:

  • Input attributes

  • Expected Outcome

A great example of the supervised learning algorithm is to do with the Biometric Attendance system, where the machine is provided with inputs and outcomes and then can go on to validate your future input.

2. Unsupervised Learning Algorithms

Unlike supervised learning algorithms, the training dataset, in this case, does not have an expected output association. For example in clustering, the machine uses an unsupervised learning approach. A good example of unsupervised learning is to do with fraud detection.

3. Reinforcement learning algorithms

This refers to a method, which deals with learning via interaction and feedback. While the goal in unsupervised learning is to find similarities and differences between data points, in reinforcement learning the goal is to find a suitable action that would maximize the total cumulative reward of the agent. Essentially the agent is dealing with an exploration vs exploitation trade-off. While in the exploration stage it acts on trial and error, exploitation is when it performs actions based on the knowledge it has gleaned from the environment. Games such as Pacman, Atari, Mario are a great example of this algorithm.

5 Commonly used Machine Learning Algorithms

1. Linear Regression

In this process, a relationship is established between dependent and independent variables. This is known as a regression line and is represented by a linear equation Y= a *X + b.


  • Y – Dependent Variable

  • a – Slope

  • X – Independent variable

  • b – Intercept

You could use it to estimate values such as total sales or cost of houses, and more.

2. Logistic Regression

Logistic regression is a supervised learning algorithm, which is used to predict the probability of a target variable.

Logistic regression can be of the following types:

  • Binary- Where the dependent variable can have two possible classes, 0 or 1

  • Multinomial- Where the dependent variable can have three or more possible unordered types

  • Ordinal- Where the dependent variable can have three or more possible ordered types

3. Decision Tree

This is a very popular supervised learning algorithm that is in use to classify problems. The classifier, in this case, is structured like a tree with the following parts:

  • Internal nodes- These represent the dataset

  • Branches- These represent the rules

  • Leaf node- These represent the outcome

4. SVM (Support Vector Machine)

Once again this is a supervised machine learning algorithm. It is found to be of use in challenges related to classification as well as regression. In this method, you plot raw data as points in an n-dimensional space (where n is the number of features you have). The value of each feature is then associated with a particular coordinate, making it easy to classify the data.

5. Naive Bayes

Naïve Bayes algorithm is a supervised learning algorithm, which is based on Bayes theorem. It is essentially used to solve classification problems. Essentially a Naïve Bayes classifier assumes that a particular feature in a class is not related to any other feature. It is extensively used in calculating the probability of an outcome and is particularly helpful for large data sets.

There are many other models that are commonly used and that can be applied to data-related problems. If Machine Learning is something that intrigues you, you could begin to apply some of these machine learning models and witness the ease with which seemingly complex issues are solved.

Leave a Reply

Your email address will not be published. Required fields are marked *