In this project, various machine learning techniques are used to analyse the sentiment of tweets from the Sentiment140 dataset, which includes the use of three basic classifiers: Logistic Regression, Bernoulli Naive Bayes (BNB), and Support Vector Machine (SVM), as well as the TF-IDFand CountVectorizer method to analyse the frequency of terms in the data. The accuracy score containing precision, recall and F1 Scores are then used to evaluate the performance of machine learning classifiers.
We may deduce the following conclusion after assessing all of the models:
When it comes to model accuracy, Logistic Regression turned out be the one with the highest accuracy of 83 percent. However, there was not a significant difference in the accuracies of the model but Support Vector Machine outperformed Bernoulli Naive Bayes method.
Bernoulli Naive Bayes:
Logistic Regression:
Support Vector Machine:
