Probability-Calibration-3-CC adjusts a Gradient Boosting classifier's overly confident predictions for a 3-class problem using **Sigmoid calibration**. Visualized on a **2-Simplex**, the project demonstrates how calibration significantly reduces **Log Loss**, making probabilities more reliable.