Open mpitts23 opened 3 years ago
katcorr
commented
3 years ago What an interesting idea! Great motivation for the topic, and cool integration of text analysis. I'm looking forward to your blog post!
I'd recommend adding more detail to your schedule -- when you do plan to meet as a group / check-in with each other?
Update 1: 10/10
jackdove23
commented
3 years ago Update #2
Our group has solidified our wrangling process: for 25 NFL quarterbacks, we will be counting their social media (twitter + instagram) posts by week, pairing those counts with their game performances, and investigating their links both for individual players and for the entire sample.
We have completed our first QB together (Ben Roethlisberger) and have divided the rest up to be completely collected wrangled by Monday, November 9th. We are two days behind schedule (simply due to academic conflicts), but our analysis will be much less time-consuming than the data collection process.
Schedule:
We are on a great track to write a compelling, thorough post about the relationship between 2020 NFL QB social media activity and on-field performance.
katcorr
commented
3 years ago Thorough update. Good plan going forward.
Just a quick clarifying question -- original plan said "for the 32 NFL quarterbacks with the most total snaps taken". Were there 7 quarterbacks that don't tweet, or why is there 25 now?
Update 2: 5/5
jackdove23
commented
3 years ago Update 3
We are right on schedule, as we have finished our wrangling process. Data for 20 QBs (those who:
Our next steps are to configure the text storage of the API key, complete our analysis, and begin designing our blog post. Schedule: Analysis Completed: Nov. 12 API Key Issue Completed: Nov. 13 Blog Post Completed: Nov. 15 Presentation Recorded: Nov. 17 Presentation Revised: Nov. 18 Presentation in Class: Nov. 19 Blog post submitted: Nov. 20
katcorr
commented
3 years ago Great!
Update 3: 5/5
Our group plans to move in a different direction from our shiny app; instead of pursuing historical team results, we will be viewing player performance from a unique angle. A common complaint from NFL fans is that players have “twitter fingers” and post thoughts on social media that sometimes get them in trouble. We have decided to investigate the link between twitter use and weekly performance. Using the rtwitter API, we will scrape tweets since September 1st for the 32 NFL quarterbacks with the most total snaps taken (effectively the 32 QBs with the most playing time). To measure performance, we will download ESPN data on quarterbacks, which includes game by game fantasy point totals (a fair performance barometer), as well as football statistics and win/loss outcomes.
We want to address a few key questions: Does the number of tweets tweeted before a game (from the end of the last game to right before the next game) of a player impact his performance? Does the sentiment of tweets tweeted before a game of a player impact his performance?
We will create a blog post that includes data in the form of graphs that determines the relationship between the number tweets of quarterbacks and their performance, measured in different variables such as fantasy points, yards and more. The main visual cue will be line graphs and other comparison tools that will lead to a conclusion to our question of ‘Does the number of tweets affect performance?’. The line graphs will be available to the user in the form of the shiny app where the user gets to choose which quarterback statistics they would like to look at. In addition, a user will be able to view information by quarterback, which will include their last ten tweets, their 2020 game logs, and their individual tweet sentiment analysis.
Upcoming Project Schedule: Load/Wrangle data (already started) - due by 11/7 Wrangle NFL week-by-week player performance data by 11/4 Wrangle quarterback tweets by 11/6 Merge data by 11/7 Write the Blog Post - due by 11/14 Create Plots and Analyses that flow with the blog - due by 11/14