Archaeological Site Prediction Model for Arsacid Period Research

Overview

This repository contains two versions of a machine learning model for predicting archaeological site locations:

1. Original Research Model (original_model.py)

• The exact code used to produce results for the paper "Predictive Modeling for Targeted Archaeological Survey of Arsacid Period Sites in the Iranian Borderland Region of the Araxes River Valley"
• Contains the complete workflow with specific data paths and parameters
• Preserves the exact methodology and results from the published research

2. Modular Framework (archaeological_prediction_framework.py)

• A cleaned, reusable version designed for reproducibility and ease of use
• Removes hard-coded paths and makes parameters configurable
• Allows researchers to adapt the model for different study areas and datasets
• Maintains the same core methodology while improving usability

Key Features

Both models use:

• Principal Component Analysis (PCA) for feature reduction
• Multiple machine learning algorithms (Random Forest, Logistic Regression, SVM, etc.)
• Cross-validation for model selection
• Environmental variables including elevation, slope, walking time, river proximity, and soil data
• Probability mapping for archaeological site prediction

The framework version adds:

• Flexible raster configuration
• Customizable data transformations
• Easy parameter adjustment
• Better code organization and documentation

Use Cases

• Researchers wanting to reproduce our exact results: Use original_model.py
• Researchers adapting the method for new areas: Use archaeological_prediction_framework.py