Predicted impact of a $10 Loan to someone in need in Brazil
We wanted to leverage machine learning to help lenders to Kiva learn where their money could make the greatest impact. Kiva is an international nonprofit, founded in 2005 and based in San Francisco, that works with microfinance institutions on five continents to provide loans to people without access to traditional banking systems. Lenders invest money in small businesses and fundraisers in underprivileged parts of the world.
Our goal was to quantify the impact of people's donation and show people how even tiny amounts of money can help multiple families across the globe start businesses that are self sustainable.
What it does
Kinvest is trained on Kiva's large datasets. We used these to train a predictive model that can score the value of a dollar amount in a certain country and accurately predict the number of families that are directly impacted by a donors donation. This encourages people to donate more.
How we built it
Kinvest is built in
python. We used data and machine learning libraries like
sklearn. We also integrated Flask, Beaker Notebook, and Firebase.
Challenges we ran into
With the data we had, it was very difficult to define what a successful loan was, or who a successful lender was. It was also difficult to learn how to leverage all of our large tech stack in just 24 hours.
Accomplishments that we're proud of
- The potential impact of Kinvest is huge. We are excited to see where Kinvest will impact those in need!
- Finishing our hack in 24 hours
- Going from unfamiliar with these technologies to being able to properly implement them into an application was no small feat
What we learned
- The satisfaction of hacking for good
- How to work with a large dataset in a limited amount of time
- The importance of picking up languages and technologies on the fly
What's next for Kinvest
We want to study Kiva's dataset more deeply so that we can better predict what a good choice for a loan is. The possibilities for linking up with other datasets (World Bank, Census, economic data) is nearly limitless. We want to see where the societal impact of Kinvest can go.