Shell.ai Hackathon 2025: Fuel Blends

This repository contains the solution for the Shell.ai Hackathon 2025, focusing on predictive modeling for fuel blending properties.

Website : Predictive Fuel Blend Analysis

Project Structure

• model.py: The main script containing the Stacking Regressor model (LightGBM, CatBoost, XGBoost with RidgeCV meta-learner).
• corr_heatmap.py: Utility script for generating correlation heatmaps of the data/predictions.
• final.csv: The generated output file matches the target blend properties with high accuracy.

Setup & Usage

1. Ensure you have the required Python libraries installed (pandas, numpy, sklearn, lightgbm, catboost, xgboost).

2. Place train.csv and test.csv in the root directory.

3. Run the model:

   python model.py

   This will process the data, train the stacked models, and generate final.csv.

Methodology

The solution employs a stacking ensemble approach:

1. Feature Engineering: Weighted averages of component properties, outlier removal, and quantile transformation.
2. Base Models: LightGBM, CatBoost, and XGBoost regressing on 10 target blend properties.
3. Meta Learner: RidgeCV (Ridge Regression with cross-validation) to combine base model predictions.