Let’s start this project with a simple question, do you trust all the news from social media? How can we detect fake news from real news? It’s a tough question. Luckily, we can detect fake news using a supervised machine learning method.

Fake news is a piece of news that is not true and deliberately designed to mislead people. It is usually spread via social media or other online platforms. Fake news is usually politically driven to give advantages or disadvantages to a political party. Such news items may contain false and exaggerated claims and because of certain algorithms, trap users in a filter bubble.

In this project, we’ll use two different datasets:

News dataset available on Kaggle.
The second dataset we’ll create ourselves using the News API. We will use this API to load some data and then append that data to the other dataset.

In the end, we will use a passive-aggressive classifier to classify and differentiate the fake news from the real ones. The passive-aggressive classifier is a classification algorithm in machine learning that changes the model whenever there is a wrong prediction. If there is no wrong prediction, the model will stay the same.