receiverstats-autolguon-predictor

datasets:

• other library_name: autogluon license: mit metrics:
• name: accuracy type: accuracy value: 1.0
• name: weighted f1 type: f1 value: 1.0 tags:
• autogluon
• tabular
• automl
• binary-classification

AutoGluon Tabular Predictor for Receiver Stats

This model was trained using AutoGluon's TabularPredictor on the "Receiver Stats" dataset from the Hugging Face Hub. This model works to classify popular NFL wide-recievers as "Elite" or not based on an augmented set of season statistics. This was was validates on a set of real-world wide-reciever statistics.

Training Details

• Problem Type: binary
• Trained Models: AutoGluon automatically trained and ensembled a stack of models.
• Training Time: 222.63 seconds

Best Model: WeightedEnsemble_L2

WeightedEnsemble_L2 is a meta-model that combines predictions from other base models. The goal is to create a more accurate and robust final prediction than any single model alone.

AutoGluon builds a hierarchy of models called "stacking levels."

• L1 Models (Base Models): This is the first layer of models. It includes a variety of individual machine learning models, like LightGBM, CatBoost, and others, each trained directly on the original dataset's features. The predictions of these L1 models are saved and used as the "features" for the next level.

• L2 Models (Meta-Models): The WeightedEnsemble_L2 is the meta-model at this second stacking level. It doesn't train on the original data features. Instead, it uses the out-of-fold predictions from all the L1 models as its input. Its job is to find the optimal way to combine these predictions to create a final, superior prediction.

Performance on Original Data

The following are the performance metrics on the original test dataset (df_orig). This represents the model's performance on unseen, real-world data.

• Accuracy: 1.0000
• Weighted F1: 1.0000

Best Model Information

• Problem Type: binary
• Evaluation Metric: accuracy
• Validation Score: 0.9968
• Fit Time: 0.0756 seconds

Best Model Hyperparameters:

use_orig_features: False
valid_stacker: True
max_base_models: 0
max_base_models_per_type: auto
save_bag_folds: True
stratify: auto
bin: auto
n_bins: None

Child Model Information (if applicable):

• S1F1:
  • Model Type: GreedyWeightedEnsembleModel
  • Fit Time: 0.0756 seconds
  • Fit Hyperparameters:
    • ensemble_size: 1

Model Leaderboard

This table shows the performance of the individual models and ensembles AutoGluon trained on the augmented data test set.

How to Use

You can load this model and make predictions using AutoGluon Do not use this model to draw real-world conclusions, this model is for educational and instructional purposes only.