SentinelMap

Inspiration

I wanted to create a tool that helps people quickly understand global social and political trends, so NGOs, journalists, and policymakers can respond to crises faster.

What it does

SentinelMap analyzes global event data in real time, predicts trends using AI, and visualizes hotspots of political unrest, humanitarian issues, or social tension on an interactive map.

How I built it

I used the GDELT Event Database to gather global event data, trained an XGBoost model to predict sentiment and risk, and built a frontend with Streamlit that displays predictions and heatmaps interactively.

Challenges I ran into

Handling missing or inconsistent data from GDELT was tricky. Encoding categorical features like country codes and event types for machine learning required careful preprocessing. Scaling the data efficiently for real-time visualization was also a challenge.

Accomplishments that I'm proud of

I successfully trained a predictive model on massive, messy real-world data and created an interactive map that clearly communicates complex global trends. The project turned abstract data into actionable insights.

What I learned

I learned how to preprocess large-scale event datasets, engineer features for machine learning, and deploy an AI-driven visualization. I also gained experience bridging backend ML models with frontend interactive dashboards.

What's next for SentinelMap

I plan to integrate real-time GDELT updates, improve predictive accuracy, add geospatial clustering, and allow users to filter by themes like human rights, climate, or political unrest for more actionable insights.

Built With

• gdelt
• geodata
• python
• streamlit