🕯️ HydroSpark: Physics-Driven Burn Rate Prediction

Overview

This project was developed for the HydroSpark ML Quest, a challenge centered on predicting the exact time a burning candle will extinguish. Unlike standard "black-box" ML approaches, this repository utilizes Physical Logic—specifically fluid dynamics and thermodynamics—to engineer features that make predictions more stable and explainable.

🧠 The Methodology: Why Volume Matters

A common pitfall in candle modeling is focusing solely on the vertical height drop. However, height change is dependent on the candle's diameter. To create a robust model, we shifted the focus to Volumetric Burn Rate ($dV/dt$).

1. Feature Engineering: We calculate the cross-sectional area ($A$) using the diameter ($d$): $$A = \pi \cdot \left(\frac{d}{2}\right)^2$$
2. Universal Metric: We convert height measurements into Volume Consumed ($V_c$): $$V_c = A \cdot (H_{initial} - H_{current})$$
3. Modeling: We apply Linear and 2nd-Degree Polynomial Regression to the pooled volumetric data. This allows the model to generalize across different candle shapes and sizes, even if the specific wax type is unknown.

✨ Key Features

• Multi-Wax Support: Individual profiles for Paraffin, Soy, and Beeswax.
• Generalized Model: A pooled dataset approach that calculates a "Universal Volumetric Burn Rate" for mystery candles.
• Physics-Informed: Respects the material density and cross-sectional constraints of the physical setup.
• Visual Analysis: Includes matplotlib visualizations to track prediction curves vs. actual burn data.

🛠️ Tech Stack

• Language: Python 3.x
• ML Framework: Scikit-Learn (Linear & Polynomial Regression)
• Data Handling: NumPy
• Visualization: Matplotlib

📊 Results & Performance

The model successfully predicts extinction times for variable diameters (e.g., transitioning from a 20mm training candle to a 9mm test candle) by normalizing for the volumetric drop rate, significantly outperforming simple time-series height models.

---

📁 Repository Structure

├── src/
│   ├── main_prediction.py      # The core modeling and prediction script
│   ├── data_visualizer.py      # Script for plotting burn curves
├── data/
│   ├── sample_measurements.csv  # Training data for Paraffin, Soy, and Beeswax
├── README.md                   # Project documentation
└── requirements.txt            # Necessary Python libraries