Model Card for mrob937/2024-24679-tabular-autogluon-predictor

Model Description

This is an AutoGluon Tabular AutoML implementation on a tabular dataset recording NFl reciever performance statistics. The model predicts whether a given reciever is "Elite" (1) or Not (0)

• Model developed by: Myles Robinson
• Model type: AutoGluon TabularPredictor

Model Sources [optional]

• Dataset: james-kramer/receiverstats

Direct Use

• This model was intended to practice automl implementation on a tabular dataset

Bias, Risks, and Limitations

• Small data size.
• Personal preference of the dataset creator in classification.

Training Data:

The model was trained on the augmented split of the "james-kramer/receiverstats" dataset. The data includes features such as target, receptions, recieving yards, and more and classifies the aggregation of these features as being elite or not.

Evaluation Data:

The model achieved an accuracy of 0.9333 and a weighted F1 score of 0.9343 on the original dataset.

Model Card Contact

Myles Robinson (Carnegie Mellon Univerity)- [email protected]