Fake news has turned out to be a menace. Distinguishing Fake news is a critical advance towards safeguarding the uprightness and prosperity of the society. With the increasing popularity of social media, there’s an upsurge in the propagation of counterfeiting news. There is a lack of proper frameworks for dealing with fake news. The proposed work aims at exploring the various machine learning techniques for detection and analysis of fake news. In this way, the accompanying task goes for proposing a worldview for ordering counterfeit news and utilizing learning systems such as Naïve Bayes, Support Vector Machines, Stochastic Gradient Descent and Neural Networks and draws an act comparison between the same. The analysis is based on a dataset of 20000 news samples collected from various sources including social media, news websites, online gossips etc. pre-processed using TF-IDF and count vectorizer. Similar ideas are utilized for incorporating an assessment investigation application for a sample 2014 American Presidential Election social media dataset which depicts user behavior with relevant subjects dependent on pertinence, freshness and criticalness. The model results in an underlying accuracy of 93% with further improvements to be expected based on cross-referencing, dynamism and tracking history of reputation of news sources. The research in the area of fake news detection has been vastly inhibited by lack of quantity and quality of existing datasets along with algorithms to model the given problems. To counter this issue, we thoroughly assemble and outline trademark machine learning algorithms and a context-independent dataset for analysis. © 2019, Research Trend. All rights reserved.