EASLT 0.0.5

Enhanced Automatic Shifted Log Transformer with Monte Carlo optimization

Enhanced Automatic Shifted Log Transformer (EASLT)

Automatically transform skewed data into more normal distributions using a Monte Carlo–optimized shifted log transformation.

---

🚀 Features

• Works with negative, zero, and positive values
• Reduces skewness and kurtosis
• Adaptive, data-driven transformation
• Multi-test normality assessment
• Scikit-learn compatible
• Numba-accelerated for speed

---

📦 Installation

pip install EASLT

---

⚡ Quick Start

from EASLT import AutomaticShiftedLogTransformer
import numpy as np

# Example skewed data
np.random.seed(42)
data = np.random.exponential(2, (1000, 3))

# Transform
transformer = AutomaticShiftedLogTransformer(mc_iterations=1000, random_state=42)
data_transformed = transformer.fit_transform(data)

---

📚 How It Works

EASLT:

1. Detects data complexity (normal, mild issues, needs transformation)
2. Applies robust shifting for negative/zero values
3. Detects outliers and optionally winsorizes
4. Optimizes parameters via Monte Carlo with Dirichlet sampling
5. Applies Feng's shifted log transformation

---

📝 License

MIT License © 2025 Muhammad Akmal Husain

🔗 Links

• GitHub Repository
• Documentation