This project scrapes every NFL team's subreddit - beginning five minutes before each game's end time and ending twenty-four hours after the game's start - and uses sentiment analysis to determine a "positivity" and "negativity" score for every individual comment. We then aggregate sentiment results on a per-team, per-post, per-comment, and per-word level, and rank each NFL team by its subreddit's net positivity. We also identify the specific posts which elicited the most positive and negative comments, as well as the most positive and negative comments themselves. Finally, we identify the words most commonly appearing in each NFL subreddit's positive and negative comments.
A full description of the project can be found at saisenberg.com. To use the Shiny app of weekly fanbase sentiment, visit my shinyapps.io page.
- Python (suggested install through Anaconda)
- Python:
- praw (
!pip install praw) - matplotlib, nltk, numpy, pandas, re, sys, unicodedata (
all installed with Anaconda)
- praw (
This code comprises the entirety of the project summary above.
A few important notes:
-
Change the week variable in the appropriate kernel to change the week to scrape.
-
Change the mins_scrape variable to change the number of minutes before the end of each game to scrape.
-
Change the hours_scrape variable to change the number of hours after the start of each game to scrape.
-
Keep in mind that PRAW can only scrape the one thousand most recent posts in a given subreddit.
Additionally:
-
If this is your first time scraping Reddit, make sure to set up an app.
-
Be sure to update your Reddit app client ID, client secret, and user agent, as well as your personal Reddit username and password in the appropriate kernel.
- Sam Isenberg - saisenberg.com | github.com/saisenberg
This project is licensed under the MIT License - see the LICENSE.md file for details.
StackOverflow user Yann Dubois