ReliefX

Inspiration

During disasters like floods, cyclones, or earthquakes, delays in response often cost lives. Relief agencies struggle with identifying the most affected areas quickly and coordinating limited resources effectively. We wanted to create a simple, data-driven solution that helps responders prioritize and act faster, without needing a complex technical setup.

What it does

ReliefX is an AI-powered coordination tool that predicts high-risk zones during natural disasters and recommends priority areas for resource allocation. It uses open disaster datasets to forecast severity and visualize hotspots on an interactive dashboard. Relief teams can see where medical aid, food, or shelters are most needed, enabling quicker and more informed decision-making.

How we built it

We designed ReliefX with a minimal yet powerful tech stack to keep it lightweight and easy to deploy in real-world disaster scenarios.

1. Data Collection & Preprocessing: Collected historical disaster datasets (floods, cyclones, earthquakes) from Kaggle and government open-data portals. Used Pandas for data cleaning, normalization, and handling missing values.

2.Model Development: Implemented baseline predictive models using Scikit-learn (Logistic Regression and Random Forest). These models classify severity levels and predict potentially high-risk regions based on input features (e.g., rainfall, population density, past damage reports).

3.Visualization & Insights: Used Matplotlib and Seaborn to generate severity plots and risk heatmaps during experimentation. Risk zones are then converted into geo-mapped visualizations to make results more actionable. Dashboard & User Interface Built an interactive dashboard with Streamlit, allowing relief workers to easily input new data and view predictions. The dashboard presents results as heatmaps, tables, and priority alerts, ensuring usability even for non-technical users.

4.Deployment & Accessibility: Since Streamlit apps can be hosted easily, the solution can be run on a laptop or cloud with minimal setup. The lightweight Python-based stack ensures it works even in constrained environments.

Challenges we ran into

Limited availability of clean and consistent disaster datasets. Balancing predictive accuracy with the need for a simple, interpretable model. Designing a dashboard that remains usable even in low-resource or time-critical environments.

Accomplishments that we're proud of

Built a functional prototype with minimal tech stack that is still practical for real-world use. Created an intuitive interface that non-technical relief workers can use effectively. Demonstrated how open data + simple ML can support life-saving decisions.

What we learned

The importance of keeping solutions lightweight and interpretable for disaster relief contexts. How even small datasets can generate meaningful insights when combined with simple machine learning. Collaboration and iteration are crucial when building tools for sensitive, real-world applications.

What's next for ReliefX

Integrate real-time data feeds (satellite images, weather APIs, social media signals). Expand to multi-disaster predictions beyond floods/cyclones. Partner with NGOs and government agencies to pilot ReliefX in real disaster drills.

Built With

• cyclones
• earthquakes)
• kaggle
• matplot
• matplotlib
• open-government-data-platforms:-kaggle-(for-development-&-experimentation)
• pandas
• python
• scikit-learn
• seaborn
• streamlit-datasets:-kaggle-disaster-datasets-(floods