DroPTC

This repository accompanies the research paper:

“Interpretable root cause analysis of drone flight logs”, submitted to [DFRWS EU, 2026].

It contains:

• Source code (src/) for model training, interpretability, and evaluation.
• CLI tool (src/cli/) to run the proposed forensic method on new evidence.
• Datasets (dataset/, src/evidence/) experimental and case study datasets.
• Notebooks (notebooks/) for statistical analysis, feature attribution, and visualization.
• Experiment outputs (experiments/) where all the experimental results are.

---

Installation

Clone the repository and install dependencies:

git clone https://github.com/yourusername/project-name.git
cd project-name
pip install -e .

The -e flag installs the package in editable mode, allowing changes without reinstalling. Dependencies are listed in requirements.txt To install:

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Reproducing Experimental Results

We provide a single bash script to reproduce all experimental scenarios:

bash train_classifier.sh

This will:

1. Train classifiers under multiple settings.
2. Store results in the experiments/ folder.
3. Generate trained models that can be reused by the CLI tool.

Analysis of results is available in the notebooks/ folder:

1. Performance aggregation (mean ± std)
2. Best-performing model analysis
3. Feature attribution heatmaps

Running the CLI Tool

The forensic tool is provided under the src/cli/ folder. It supports per-evidence log analysis and outputs timelines + reports. See the dedicated CLI README for usage instructions.

Repository Structure

dropt/
│
├── dataset/            # training and testing datasets
├── experiments/        # experimental results
├── notebooks/          # analysis & visualization notebooks
├── src/                # source code (models, utils, interpretability, CLI, evidence)
│   ├── cli/            # forensic CLI tool
│   ├── evidence/       # forensic case study data (see evidence/README.md)
│   ├── model.py        # model definitions
│   ├── train_classifier.py
│   ├── utils.py
│   └── interpretability.py
│
├── requirements.txt
├── train_classifier.sh
├── README.md           # this file