Predicting the NBA
The task: Using the stats from 2012 through 2017, predict how the 2018 season will finish out. Any and all modeling methods can be used, but the final solutions must perform the best on the 2018 season.
Data is provided from the website, https://basketball.realgm.com/nba/stats and is broken into three areas…
- Team Standings
- Team Stats
- Player Stats
Those three tables are included, as well as a JSL script to download the data yourself. Additionally, there is an FAQ list included in the Word Document.
The questions to answer are the following…
- Closest Order of Final Team Standings, as ranked by percentage of wins, by the end of the 2018 regular season, with a tie breaker being the closest to having the correct winning percentages.
- Closest Order of Final Player Standings, as ranked by average points per game, by the end of 2018 regular season, with a tie breaker being the closest to having the correct PPG values
- Best Visualization created with the data, voted on by the User Group community at the meeting. If you aren’t a data junky, this is your chance to shine. What is the neatest, coolest, most interesting graph/chart/etc. that you can come up with?
The Scope and Rules
- Teams of 1-3 people. To sign up, go to this link, https://www.surveymonkey.com/r/HYRGQM9.
- Build 1 or 2 models to predict the above questions. Models must be able to run live, during a JMP User Group meeting.
- Provide a visualization that can be created during the meeting, and will be voted on during the meeting.
- At least one member of each team must be present to win.
- Register your team by April 28th. Solutions will be presented and winners will be announced at the Summer Upstate JMP User Group (typically in June)
- Solutions can be in the form of a JMP script, a data table, a JMP Journal, or other JMP data type that will be available to run during the meeting.
If you have any questions, feel free to post them here, or message me directly.
Brad Foulkes
