Implementation of Naive Bayes from scratch, and using it for sentiment analysis on IMDB comments.
The dataset is known as 'Large Movie Review Dataset'. It contains 50000 .txt files of IDMB comments, which are labeled as positive or negative. We keep the n most frequent words, skipping the first m. We calculate the learning curves and accuracy. The diagrams are located at the xlsx file.
The data used can be found here