NexAtlas

Home Page
Syria Data Visualization
Ukraine Data Visualization
Wiki Page (w/ Example Response)
Forecast Page
Forecast Example: Flood Prone Areas (Vietnam)

Inspiration

The United Nations distributes over $1 billion annually through Country-Based Pooled Funds (CBPFs) to support developing nations facing humanitarian crises such as healthcare shortages, food insecurity, displacement, and lack of shelter. However, historical data reveals significant inequalities in how these funds are allocated, with some of the most vulnerable countries consistently receiving less support than their humanitarian needs demand. This misallocation leads to critical gaps in aid delivery, leaving millions of people without adequate access to humanitarian resources.

To address this challenge, we developed NexAtlas, a data-driven platform that analyzes years of funding patterns alongside real-world country indicators, including poverty levels, health outcomes, conflict intensity, and food security. By identifying nations experiencing the most severe humanitarian crises yet systematically underfunded, NexAtlas provides actionable insights to support more equitable, transparent, and effective funding decisions. Our goal is to empower the UN Office for the Coordination of Humanitarian Affairs to optimize the allocation of CBPFs, ensuring that limited resources reach the communities that need them most and maximizing humanitarian impact worldwide.

What it does

NexAtlas is a data-driven analytics platform that helps the United Nations identify, understand, and address inequalities in global humanitarian funding. By combining historical UN funding data with real-world humanitarian indicators such as poverty, health outcomes, conflict intensity, displacement, and food security, NexAtlas highlights countries experiencing severe crises that are consistently underfunded.

The platform provides interactive visualizations, geospatial mapping, and predictive analytics designed specifically to support the United Nations in making more informed, data-driven funding decisions. By enabling users to explore funding patterns, compare aid distribution across regions, and uncover systemic disparities, NexAtlas helps identify where current allocation strategies fall short.

In addition, machine learning models forecast future humanitarian risks and funding outcomes, allowing the UN to act proactively rather than reactively. Together, these tools empower the United Nations to reduce funding disparities, optimize resource allocation, and ensure that humanitarian aid reaches the populations that need it most, ultimately maximizing global humanitarian impact.

How we built it

To begin with, we found numerous datasets ranging from the UN’s Office for the Coordination of Humanitarian Affairs (OCHA), United States Geological Survey, EM-DAT, and more regarding previous funding to nations, the demographics of those suffering, specific humanitarian projects funded, and yearly budgets for respective nations. We loaded these datasets into a Volume in Databricks’ Unity Catalog. After that, we cleaned the data and transformed it using Python Notebooks in Databricks. We trained some classification models (e.g. HDBSCAN, KNN) and forecasting models (LightGBM) to predict future outcomes if we continue with current funding circumstances. We leveraged ML Experiments on Databricks to validate our models.

After completing our work in Databricks, we built an interactive web application that enables users to explore, analyze, and understand global humanitarian funding patterns. The platform provides visual analytics, geospatial mapping, and data-driven insights that highlight disparities in funding allocation across countries and regions. Users can compare funding levels against key humanitarian indicators such as population in need, malnutrition rates, displacement, conflict intensity, and healthcare access. This makes it easy to identify countries that are consistently underfunded despite facing severe humanitarian conditions.

Additionally, we developed a Wiki feature powered by vector search over a vectorized database in Databricks, enabling users to efficiently explore historical projects, case studies, and intervention outcomes. This intelligent search system allows users to quickly learn what has been done to address specific issues both within a given country and across other regions worldwide. By surfacing relevant knowledge and best practices, NexAtlas supports data-driven decision-making, knowledge sharing, and evidence-based humanitarian planning.

Challenges we ran into

Implementing Databricks was a major challenge for our team, as none of us had prior experience using the platform. As a result, we had a slow start after uploading our datasets because we were unfamiliar with how to effectively work within the Databricks environment. Fortunately, we attended a Databricks workshop where we learned the fundamentals of bronze, silver, and gold Delta tables, as well as how to use AutoML to train and validate our machine learning models. This training helped us build momentum and significantly improved our progress with the data.

Following this, we spent many hours attempting to implement vector search functionality in Databricks. We encountered numerous roadblocks, particularly when converting Delta tables into an embedded format, which required extensive troubleshooting and experimentation.

On the visualization side, developing a Django server capable of accurately querying SQL tables in Databricks was also time-consuming. We struggled to establish a reliable connection between Django and our Delta tables. Since this functionality was a critical component of our application, we dedicated significant effort to debugging and refining the integration until it worked correctly.

Accomplishments that we're proud of

To begin with, none of us had prior experience using Databricks before this hackathon. In just 36 hours, we progressed from complete beginners to confidently setting up Unity Catalogs and running SQL queries. We are incredibly proud of how quickly we were able to learn and apply such a powerful platform.

Beyond the technical achievement, we are especially proud of the insights our application delivers. Rather than presenting surface-level observations that can be easily derived from publicly available UN datasets, our visualizations uncover deeper, systemic issues within UN OCHA’s funding strategies. Specifically, our analysis reveals how factors such as gender, healthcare access, and population dynamics influence which nations are consistently denied the funding they urgently need. We are proud of the meaningful insights our application provides and the value it can offer to UN OCHA’s decision-making process.

What we learned

Moreover, through learning Databricks and Django, we gained valuable insight into the role of the United Nations in addressing humanitarian issues, including its long-term goals and the impact of its funding efforts. We explored the reasons behind global funding disparities and examined the socio-economic factors that contribute to them, which allowed us to view these challenges from a broader, more nuanced perspective. On the technical side, all team members now feel confident in loading, transforming, and querying data on Databricks, as well as training machine learning models using Databricks.

What's next for NexAtlas

Next for NexAtlas is the development of a smarter forecasting model that integrates live seismic data, tectonic geography, and other real-time environmental signals to support the United Nations’ efforts in natural disaster relief and crisis preparedness. By enabling earlier and more accurate predictions, we aim to help identify high-risk regions before disasters escalate into full-scale humanitarian emergencies. In parallel, we will build a comprehensive supply chain and logistics analysis layer to track how critical resources move through the humanitarian system, pinpoint bottlenecks, and uncover inefficiencies beyond funding constraints. Because not all humanitarian challenges can be solved with financial resources alone, NexAtlas will focus on analyzing the root structural, logistical, and systemic factors that drive these crises, providing deeper insight into the true causes of humanitarian issues and enabling more effective, data-driven interventions.

Databricks Raffle Challenge Video: https://youtu.be/KWwt85HVZDQ