Social media has become a significant platform not only for communication, but also for business and marketing of content and services. These activities involve large volumes of data with different patterns that have to be harnessed for businesses or users to maximize their meaning and make relevant brand decisions. Machine learning algorithms and tools like Natural language Processing (NLP) facilitate the system’s ability to recognize, extract cloud data patterns, and classify them in a more realistic way. The previous method only made mere literal guesses to aid automated data tracking and decision-making. The aim of this study was to carry out social media data analytics using machine learning tools. In this study, we proposed novel Naïve Bayes and Maximum Entropy classifiers. These classifiers were used to conduct a sentiment analysis to categorize and analyze messages collected from Twitter in order to determine whether they are positive, negative or non-partisan, and ultimately get the targeted output. The experimental results were successful and the machine learning tools were accurate in determining the nature of the Twitter messages. The study demonstrates the capacity of machine learning algorithms in conducting actionable social media data mining and analytics.