The code for our final project for STOR 565 Spring 2019 class at UNC

Project Report
Kaggle Link

The dataset contains over 1 million data points each with a label indicating sarcasm, the comment, the author, the subreddit, the score, the ups (likes), the downs (dislikes), date of comment, created in utc time, and the parent comment. The link also provides two other datasets, but we (and other people) could not figure out what they mean. However, the main dataset should be enough for our purposes.

Our goal with the data set is to see if we can use machine learning techniques to accurately classify sarcastic comments from samples on reddit. Our group as a whole thought this is an interesting topic and it will help us learn new machine learning techniques outside of the class. We hope to see that this project is challenging but also feasible with the amount of time we have.

We are hoping to use a variety of machine learning techniques and compare how well each does against each other. This includes logistic regression, decision trees, random forests, neural networks, and others that we may see fit (we hope to do as much as we can with the amount of time we have). Another technique we need to tackle for this project is trying to use the comment data, which is just a string of words. This would require us to decide on how to treat this data (maybe add a score or a checklist of sarcastic words/capitalization/punctuation) and also learn some natural language processing techniques. We hope that these techniques are not too technical for our group to learn and apply it to our dataset.