The exponential rate at which textual information is being generated over the internet makes extracting useful knowledge from these vast volumes of information essential and increasingly important. Analysis of sentiments or opinion engineering plays a vital role in retrieving actionable knowledge from users or customer web reviews. Sentiment analysis of movie reviews help users to quickly determine which movie to purchase or watch. Also, it helps movie producers to get customers feedback on their movies. This paper presents a movie reviews sentiment classification model using Naïve Bayes (NB), Support Vector Machine (SVM), and K-Nearest Neighbors (KNN). Prior to building sentiment classification models, data pre-processing techniques were applied on the labelled polarity movie reviews dataset. The most important features (unigram and the mixed-unibigram) were extracted from the dataset using Term Frequency (TF) and Term Inverse Document Frequency (TF-IDF) feature extraction techniques. The extracted features were classified using three (NB, SVM, and KNN) supervised machine learning algorithms. The result of the implementation shows that KNN had 95.9% accuracy with TF and mixed-unibigram features, NB and SVM had an accuracy of 90.6% and 92.22%, respectively. Therefore, the result shows that KNN gives the best performance.