feat(mini-nn): add baseline neural network for PlayerStatistics class…

[LalitYemireddy]

What I changed

• Added new file: src/mini_nn/lalit_model.py

• Built a small PyTorch neural network (MiniNet) for binary classification

• Model predicts whether a player scores 20+ points in a game

• Uses data from PlayerStatistics.csv

• Removed scoring-related columns (FG made, 3PT made, FT made, etc.) to prevent target leakage

• Includes:

  • Train/validation/test split
  • Feature scaling (StandardScaler)
  • Early stopping
  • F1 score and AUC evaluation

---

Why I changed it

This introduces a baseline neural network model to evaluate learning performance on NBA player game statistics. It helps:

• Validate the data pipeline end-to-end
• Establish a baseline before using more complex models
• Detect and prevent feature leakage
• Provide a foundation for future feature engineering

---

How did I test it

Ran locally inside the project virtual environment:

python src/mini_nn/lalit_model.py

Dataset

• Rows: 1,655,736
• Features used: 12
• Positive rate (20+ pts): 12.9%

Final Test Results
Test Accuracy: 0.7249
Test F1: 0.4319
Test AUC: 0.8374

Model trains successfully and produces realistic (non-leaky) performance.

---

Related Issues

N/A – baseline model addition

---

PR Title (Conventional Commit)

feat(mini-nn): add baseline neural network for PlayerStatistics classification

[JonathanPLev]

looks great, no questions. good job!