WESAD Visualizer App

A Python-based interactive GUI application for exploring physiological sensor data and classifying emotional states using Machine Learning.

📚 Dataset: WESAD (Wearable Stress and Affect Detection)

This project uses the WESAD dataset, a multimodal dataset designed for stress and affect recognition using wearable physiological signals.

• Collected from 15 participants using chest-worn and wrist-worn devices.
• Includes physiological data such as EDA, EMG, RESP, Temperature, ECG, etc.
• Each record contains time-synchronized sensor data labeled with emotional states: Rest, Amusement, Stress.
• Dataset is provided in .pkl format.

📖 Citation:
Philip Schmidt, Attila Reiss, Robert Duerichen, Claus Marberger and Kristof Van Laerhoven,
"Introducing WESAD, a multimodal dataset for Wearable Stress and Affect Detection",
ICMI 2018, Boulder, USA, 2018.
Read the paper

---

🔍 What It Does

• Loads physiological sensor data from the WESAD dataset
• Performs feature extraction on signals such as EDA, EMG, RESP, and Temperature
• Trains a Decision Tree model to classify emotional states (Rest, Amusement, Stress)
• Presents key plots and ML predictions inside a user-friendly Tkinter GUI

---

🧵 What I Practiced & Built in This Project

From raw sensor data to predictions: I worked with .pkl files from the WESAD dataset and turned raw physiological signals (like EDA, EMG, TEMP, and RESP) into a format suitable for machine learning.

Feature extraction made simple: I applied basic statistical operations (mean, std, etc.) to slices of time-series data, creating feature vectors for classification tasks.

Basic machine learning integration: I trained a DecisionTreeClassifier model with scikit-learn, saved it with pickle, and used it to classify emotional states inside the app.

A GUI that actually works: Built an interactive interface using Tkinter where users can load data, see plots, and get predictions — all in one place.

Modular code structure: Kept everything clean and organized by splitting the code into folders like src/, gui/, model/, and data/.

Running it is easy: The project includes a requirements.txt and handles model loading automatically — just drop a WESAD file into the right folder and you're good to go.

---

📁 Folder Structure

wesad-visualizer-app/
│
├── gui/
│   └── main_gui.py
├── src/
│   ├── main.py
│   ├── preprocessing.py
│   ├── predict.py
│   └── plot.py
├── model/                  
├── data/                   
│   └── (not included in repo)
├── requirements.txt
└── README.md

---

⚙️ How to Run

1. Download a .pkl file (e.g., S2.pkl) from the WESAD Dataset
2. Place it in the data/ folder
3. Install the requirements:

pip install -r requirements.txt

4. Run the GUI:

python main.py

---

⚠️ Notes

• This project was designed for academic purposes. I aimed for demonstrates a basic level end to end project.
• GUI layout may not be fully responsive (first complete GUI attempt :)).
• Model performance is not the focus — the goal is demonstrating the full ML workflow.

---