SAFARI Framework: Simulated IMU Fatigue Dataset and Pipeline Code for Real-Time Neuromuscular and Metabolic Fatigue Classification

Dataset and pipeline code for the SAFARI (Stochastic Adaptive Fitness-Aware
Real-time Inference) framework for real-time neuromuscular and metabolic
fatigue classification in sprint and jump athletes.

Contents:
- generate_imu_data_v2.py: Simulated IMU dataset generator (9 subjects,
  540,000 samples, 6 channels at 100 Hz) with temporal fatigue injection
  calibrated to published biomechanical effect sizes
- safari_diagnostics.py: Full data diagnostics pipeline
- safari_full_v3.py: Complete SAFARI pipeline including entropy feature
  extraction, OC-SVM anomaly detection, Banister adaptive thresholding,
  and four-state fatigue classification
- imu_prepared_v3.csv: Prepared dataset (winsorised, labelled)
- safari_results_v3.csv: Primary results (AUC-ROC 0.9820, F1 0.8835)

Related publication: Moroke, K.M.; Moroke, N.D. Real-Time Neuromuscular and
Metabolic Fatigue Classification in Sprint and Jump Athletes: The SAFARI
Framework. Sports (MDPI), 2026 (submitted).