Intervestment.tech

Empowering policymakers and citizens to understand the global impact of U.S. foreign aid, viewed through the lens of financial investment, in an era of rising debate over its allocation.

Comment

Inspiration

Intervestment.tech was inspired by the growing need for a more objective and effective way to analyze U.S. foreign aid. Policymakers often lack the tools to truly assess the impact and ROI of foreign aid, relying on data that is sometimes fragmented or subject to political bias. In a climate where aid funding, particularly from USAID, is being defunded and scrutinized, there’s an urgent need for a data-driven approach that transcends political agendas. By analyzing key indicators like GDP, HDI, unemployment rates, and energy consumption and their relationship to US foreign aid spending, our project empowers decision-makers with a clear, unbiased view of how aid is being spent and which countries will use it most effectively. This allows for more informed decisions, fostering a discussion grounded in data rather than politics, and helps ensure aid reaches those who need it most.

What it does

Our project is a comprehensive data visualization tool designed to solve the problem of fragmented and biased foreign aid data. It provides an intuitive interface that details the distribution of global statistics and U.S. foreign aid spending patterns over time. The application analyzes the latest national data, generating a Return on Investment (ROI) score to reflect a nation's improvement relative to aid received, a Fit score to measure alignment with historical spending patterns, and an aggregate score to determine the nation’s priority for additional funding. By presenting this data clearly, the tool helps policymakers make informed, unbiased decisions about where aid can be most effectively allocated.

How we built it

We gathered public data from U.S. government sources and online archives, then preprocessed it using Pandas before converting it to JSON format and storing it in MongoDB. The data visualization was implemented with D3.js within a React application, which communicates with a RESTful Express backend. To calculate the ROI score, we employed a model based on linear regression and R-squared values to assess correlations. For the Fit score, we utilized deep learning through a sequential neural network built with TensorFlow to evaluate alignment with historical spending patterns. This combination of tools and techniques allows for a powerful, data-driven analysis of U.S. foreign aid.

Challenges we ran into

Compiling and consolidating the vast amount of data proved challenging, as it required significant effort to bring together multiple datasets. Fortunately, MongoDB's flexibility as a non-relational database made storing and managing our JSON data more efficient. Given the complexity of international relations, isolating factors to determine a pure correlation between aid spending and predictors was difficult. However, we improved the accuracy of our analysis by segmenting countries by income group (as reported by the U.S. government), normalizing the data, removing outliers caused by war or public crises and accounting for global trends, such as the natural increase in life expectancy worldwide. This approach allowed us to refine our model and make the analysis as precise as possible.

Accomplishments that we're proud of

I’m proud of successfully compiling and preprocessing all the data we collected, despite the challenges it posed. It was incredibly satisfying to see it all come together and function within our application. Additionally, I’m proud of our ability to build a fully-fledged tool that performs meaningful analyses to address a real-world problem, demonstrating the potential of data science and technology to drive impactful solutions.

What we learned

As software engineers with limited experience in data science, this project was my introduction to the data science workflow, including the often-grueling process of searching for and preprocessing data. I had never used the Pandas library before, but I now feel comfortable with it, as well as with finding and curating relevant datasets online. This experience has sparked my excitement to dive deeper into data science and continue exploring its potential as I advance in my career.

What's next for Intervestment.tech

By collaborating with the U.S. government and other international organizations, Intervestment.tech can access more comprehensive and consistent data, further improving the accuracy of its predictions. Currently, the data is fragmented and contains inconsistencies, such as how it handles contentious territories like Taiwan and Kosovo. The deep learning model can also be enhanced by incorporating additional factors, such as global conflicts or crises, to provide even more accurate insights. Ultimately, Intervestment.tech highlights the transformative potential of deep learning, demonstrating how it can be leveraged to drive better, data-driven decisions that benefit people worldwide and contribute to more effective resource allocation.

Built With

Share this project:

Updates