A homework assignment for George Washington University's CSCI 4907.82 Natural Language Understanding class
Naive Bayes and Logistic Regression classifiers were compared after developing, training, and testing on dataset of 5,000 tweets. The two classifiers were assessed based on their accuracy and F1 score for sentiment analysis from the true labels: positive, negative, or neutral. sk-learn, spacy, and numpy were used. The text was pre-processed using common techniques such lower-casing, lemmatization, and removal of stop words.