In this repository for my sentiment analysis project, I made two iterations: one that analyzes the sentiment based on the polarity of a cleaned review and another that tokenizes a review and finds the sentiment based on the average polarity in all the tokens. Though at the end, you can see that there is a "26.0%" accuracy, this is because the dataset that I analyzed was ChatGPT generated, so the sentiment that ChatGPT originally generated was slightly wrong, which you can see if you read the reviews and sentiments in the dataset. If you were to utilize the code, modify the variable names, and use a credible dataset with accurate sentiments, you would see that the accuracy of the algorithm would greatly improve. The algorithm analyzes sentiment very accurately using polarity with TextBlob but just utilizes a dataset that isn't very reliable.