Social media is a useful platform that facilitates the sharing of information and builds a virtual network among communities. This platform is sometimes useful to connect people and share useful information among friends and relatives. Nowadays, this platform is also used to share skills through short videos, transfer payments through the Unified Payments Interface (UPI), advertise and promote products to enhance businesses. However, social media platforms are connected through public networks, so a lot of hackers are connected to the network. The hackers want to steal personal information or manipulate the views of the users. Therefore, it applies various social engineering activities to gather personal information and spread a lot of fake news through various channels, apps, and social media pages. Therefore, it is a great challenge to detect fake news. Currently, many research communities are working to implement an algorithm that can detect fake news automatically based on an analysis of data. In an analysis of data, machine learning approaches such as regression, classification, and clustering methods may play an important role in detecting fake news from various datasets obtained from social media sites or the internet. In this paper, a deep analysis of combined datasets for fake news detection has been performed and this analysis is based on machine learning approaches. In addition, it compares the performance of machine learning models such as logistic regression, support vector machines, random forests, passive-aggressive classifiers, and decision trees.