Skip to content

Latest commit

 

History

History
50 lines (39 loc) · 2.37 KB

group_proposal.md

File metadata and controls

50 lines (39 loc) · 2.37 KB
Raw

Group Proposal

Paul Nguyen & David Herrero Quevedo

Proposal 1

For this topic, my group will do a combination of inference and prediction. The idea is to describe how the NBA goes from era to era. For instance, how recently the big man dominance has shifted to star small ball guards. We would look at how players stats have changed over time, such as 3 point attempts, or stats that aren't necessarily connected to a single player, such as points per possession or pace. Finally, my group plans to do some sort of prediction to look and find out what the next star player's "type" could be... a long wing? Could use data from could use data from https://www.basketball-reference.com/ using this package: http://asbcllc.com/nbastatR/

Proposal 1b

One topic of interest that I think my group would be able to study is the NBA. A potential question that we could look at is whether a player will be an all star for the upcoming season. This would be a predictive classification model. We could use data from https://www.basketball-reference.com/ using this package: http://asbcllc.com/nbastatR/ We could train our model from looking at some statistics of previous all star players for the past ~50 years. Important stats can be: Usage Rating, Points, Blocks, Rebounds, Assists, Value over Replacement Player, Field Goal Percentage, 3 point percentage etc. I think that this problem is pretty cool, as we would be able to see how the NBA has transitioned from era to era and see what is valued, and incorporate these different emphasises in our model. In addition to predicting all stars for the upcoming season, we can predict who will win the other awards: MVP, All NBA teams, MIP, sixth man of the year, all defensive team, DPOY, etc.

Proposal 2

Another potential topic could use the NHANES dataset. We could use models to predict whether a person or not will have diabetes (predictive, classification). I think that it could also be interesting to develop a model to determine if an individual smokes or not, as these individuals participate in unhealthy behavior when they usually know that smoking isn't good for them.

Proposal 3

  • Question: Can we predict who will be the next Ferrari driver using previous season performance with variables: comparison to teammate, years in Formula 1, performance of current Ferrari drivers...?

  • Type: Prediction

  • Dataset: http://ergast.com/mrd/db/