Social media provides a low-tech means of communication for people to express their thoughts, feelings, and attraction. The papers goal is to extract different sentimental behaviors that will be utilized to help decide on a course of action and classify peoples feelings and affections as neutral, contradictory, or clear. Noise removal was used to pre-process the data in order to eliminate noise. The research project used a variety of methodologies. The popular classification techniques were used to extract the sentiment once the noise was removed. The data was classified using multi-layer perceptions (MLP) and convolutional neural networks (CNN). These two classification results were compared against other techniques such as support vector machines (SVM), random forests, decision trees, Naïve Bayes, etc. using the sentiment classification from Twitter data and the consumer affairs website. In the proposed work, convolutional neural networks and multi-layer perceptions exceed other machine learning classifiers.