# Context The dataset is a csv file compiled using a python scrapper developed using Reddit's PRAW API. The raw data is a list of 3-tuples of [username,subreddit,utc timestamp]. Each row represents a single comment made by the user, representing about 5 days worth of Reddit data. Note that the actual comment text is not included, only the user, subreddit and comment timestamp of the users comment. The goal of the dataset is to provide a lens in discovering user patterns from reddit meta-data alone. The original use case was to compile a dataset suitable for training a neural network in developing a subreddit recommender system. That final system can be found [here][1] A very unpolished EDA for the dataset can be found [here][2]. Note the published dataset is only half of the one used in the EDA and recommender system, to meet kaggle's 500MB size limitation. # Content user - The username of the person submitting the comment subreddit - The title of the subreddit the user made the comment in utc_stamp - the utc timestamp of when the user made the comment # Acknowledgements The dataset was compiled as part of a school project. The final project report, with my collaborators, can be found [here][3] # Inspiration We were able to build a pretty cool subreddit recommender with the dataset. A blog post for it can be found [here][4], and the stand alone jupyter notebook for it [here][5]. Our final model is very undertuned, so there's definitely improvements to be made there, but I think there are many other cool data projects and visualizations that could be built from this dataset. One example would be to analyze the spread of users through the Reddit ecosystem, whether the average user clusters in close communities, or traverses wide and far to different corners. If you do end up building something on this, please share! And have fun! Released under [Reddit's API licence][6] [1]: http://ponderinghydrogen.pythonanywhere.com/ [2]: https://github.com/cole-maclean/MAI-CI/blob/master/SubRecommender/EDA%20Notebook.ipynb [3]: http://cole-maclean.github.io/blog/files/subreddit-recommender.pdf [4]: http://cole-maclean.github.io/blog/RNN-Based-Subreddit-Recommender-System/ [5]: https://github.com/cole-maclean/MAI-CI/blob/master/notebooks/blog%20post.ipynb [6]: https://www.reddit.com/r/%20reddit.com/wiki/api-terms