OWL NIGHTERS

The .csv submission file can be found in our github repo with the name: submission.csv

Inspiration

Fuel demand has always been a tricky yet vital problem. Accurately understanding and predicting it isn’t just interesting—it’s impactful. With that in mind, we set out to create something that could make fuel demand forecasting simpler and more accessible for everyone.

What it does

Our app turns massive, messy datasets into clean, actionable insights. Users can upload their dataset, and our app delivers accurate predictions in .csv format. For non-technical users, we’ve made it even easier—intuitive visualizations allow anyone to understand trends, patterns, and predictions without needing to dive into the raw data. It’s simple, fast, and effective.

How we built it

We started by building and training several machine learning models, experimenting with techniques to find the most accurate. Once we had the best models, we stacked them to further improve performance. Finally, we used Streamlit to create a user-friendly app that brings everything together in a clean, interactive interface.

Challenges we ran into

With 3/4 of our team being datathon first-timers, it took time to figure out the workflow. Feature engineering was especially tough—at first, the dataset categories felt completely unrelated. We also had to grasp and implement complex machine learning techniques while ensuring the app worked seamlessly.

Accomplishments that we're proud of

Despite the lack of experience, we decided to challenge ourselves to an interesting problem. After countless hours without sleep, we managed to get a high-performing model and applied some creative methods to feature engineering and data visualization. Our final model and app reflect this result. We are also proud to have made this app accessible to every user, whether they have a technical background or not, and help inspire their love of Data Science, just as this contest inspired us.

What we learned

We learned about the nuances of predicting fuel demand and how to apply machine learning to real-world problems.

What's next for OWL NIGHTERS

With further feature engineering and fine-tuning, this project can evolve into a powerful tool for businesses and policymakers. Being able to engineer more features for this model can help us improve our accuracy by a lot.