Spread of fake news on social media has been a rising concern ever since the exponential growth of the internet. Propagation of fake news is rapid and can manipulate people’perception of reality effortlessly. Due to these deleterious effects of fake news, fake news detection has become essential and is gaining a lot of attention these days. Moreover, fake news is like real news in terms of structure and context. It becomes exceedingly difficult to detect fake news articles. Here, Machine Learning (ML) can play a significant role in classifying fake news. This study covers the classification of the fake news using machine learning models like Neural Network - Keras, Support Vector Machine (SVM), RandomForestClassifier and Logistic Regression. This study also covers the analysis of fake and real news articles that have been made to understand how these articles are structured. Next, there is a comparative analysis of the hate content in these articles with the publicly available datasets to understand the extent of hate content in fake news. Afterwards, there is the implementation of 4 ML models and the effect of performance for each model is measured with respect to the changes in the type of feature vector extraction techniques and the size of the dataset. Each of the ML models is evaluated in terms of its accuracy and there is further hyperparameter tuning performed to optimize the accuracy of the models. This study helps us in understanding the features of fake news articles and produces an optimal way to build a model to detect fake news that is propagating around us in social media.