Final Peer Review

[jha62]

The goal of this project is to predict whether or not a hitter with swing, take, or swing and miss at a pitch. They are using pitch by pitch data from the 2019 season. This data includes characteristics about each pitch and the outcome. Using these characteristics they try and predict the outcome of the pitch.

Likes:

1. I really like the visualizations. They are clear and helped me understand a somewhat complicated dataset.
2. I think the choice of data was very good. I am a big baseball fan and this data is very well suited to the analysis performed.
3. Using the XGBoost model and achieving such good results was impressive. I was unfamiliar with this type of model and the parameters were explained well.

Improvements:

1. The paper mentions a model where they shrink the output into only two classes (swing vs. take). I would have liked to see more of this model since it sounds interesting.
2. An explanation of why they chose certain models would have been helpful, especially since I am unfamiliar with XGBoost.
3. Perhaps if the goal is to predict whiffs then fitting a model that classifies whiffs and combines contact and takes into one class would work.

Overall this project was very good. The dataset was well suited to the goal and the analysis performed was advanced and interesting.