Data-Driven Heritage Preservation: Interactive Educational Suite

Overview

This repository contains interactive educational materials developed for the NCPTT grant: "Data-Driven Heritage Preservation: Leveraging Machine Learning for Informed Adobe Conservation Strategies".

Designed for All Skill Levels: These notebooks use a "Scaffolded Learning" approach:

1. No-Code: Use interactive sliders and buttons to explore concepts.
2. Low-Code: Read explanations of how the logic works.
3. Pro-Code: Inspect the underlying Python code to learn implementation.

---

🚀 Getting Started

No installation required! You can run these tutorials directly in your browser using Google Colab.

1. Launch the Tutorials

Lesson	Topic	Link
Tutorial 1	Data Exploration & Pattern Analysis
Tutorial 2	Expert-Statistical Intervention Framework

2. Load the Data (Important!)

When you open the notebooks in Colab, you will need to upload the data file:

1. Download synthetic_adobe_data.csv from this repository.
2. In Colab, click the Folder Icon 📁 on the left sidebar.
3. Upload the CSV file.

---

📚 The Tutorials

Tutorial 1: Data Exploration & Pattern Analysis

What you'll do:

• Data Explorer (No-Code): Use a dropdown menu to visualize different damage types.
• Factor Analysis (Low-Code): Use a slider to group damage types into "Co-occurrence Patterns." See the heatmap update in real-time.
• Weight Recovery (Low-Code): Use Elastic Net to recover the implicit weights of the scoring system and identify which conditions drive the Total Score.

Tutorial 2: Expert-Statistical Intervention Framework

What you'll do:

• Matrix Builder: Interactively build an intervention priority matrix by combining statistical importance (from Tutorial 1) with preservation ethics.
• The Budget Game: You set the budget! Define your total funds and unit costs, then use sliders to allocate repairs. Try to maximize your preservation score without going over your limit.

---

📁 Repository Structure

FOUN_ncptt/
├── README.md                           # This guide
├── generate_synthetic_data.py          # Creates privacy-preserving dataset
├── synthetic_adobe_data.csv            # The data (generated by script above)
├── data_exploration_and_pattern_analysis.ipynb # Tutorial 1 (Diagnostics)
├── intervention_matrix_notebook.ipynb  # Tutorial 2 (Decisions)
├── generate_tex_figures.py             # Script for publication figures
└── texfigures/                         # Output folder for images

---

🎓 For Instructors

Learning Objectives:

1. Conceptual: Understand Factor Analysis and Elastic Net / Weight Recovery without getting bogged down in syntax.
2. Applied: Translate statistical findings into preservation decisions.
3. Technical: (Optional) Learn the Python implementation of these methods.

Customization: You can modify the generate_synthetic_data.py script to simulate different types of heritage structures (e.g., masonry bridges, timber frames) to adapt the tutorials for other domains.

---

Author: Dr. Rebecca Napolitano, Penn State University
Grant: National Center for Preservation Technology and Training (NCPTT)