fairLends

• 

  Home Screen

• 

  Processed Screen

Inspiration

Even today, discriminatory lending practices persist, leading to unequal access to financial opportunities and leaving certain communities at a disadvantage to home ownership and economic mobility. By shining a light on these inequities, we aim to build a future where financial access is fair, transparent, and inclusive for all.

What it does

fairLends analyzes a bank's lending history for potential discrimination using a logistic regression model. Leveraging GPT-3.5, it then generates a polished, well formatted report that details the results of the analysis and its interpretation, data visualizations, and actionable recommendations on how to improve.

How we built it

fairLends was built using Python and a suite of its libraries, including scikit-learn, statsmodel, and pandas for preprocessing and data analysis, matplotlib for data visualization, and GPT-3.5 for generating detailed reports. The frontend interface was developed with Python’s Tkinter library, enabling the creation a simple, user-friendly UI.

Challenges we ran into

We used a public dataset provided by the HMDA (Home Mortgage Disclosure Act), which had a lot of issues, including missing entries, as well as many irrelevant and confusing data fields. Because the dataset was so large and contained many unexpected problems, preprocessing the dataset into something useable caused quite the headache. Another major issue we faced was optimizing GPT -3.5's report generation, as it often times did not do what we specified of it. Lots of prompt tweaking was needed, and many credits were spent.

Accomplishments that we're proud of

Firstly, we are proud of creating an application that is fully functional, especially since this is the first hackathon for all of our team members. We’re particularly proud of overcoming the challenges of working with a massive and complex dataset, transforming it into actionable insights that address real-world issues in lending practices. Additionally, we are proud of how much we all grew as programmers through this experience.

What we learned

We learned how to work with a variety of data analysis libraries in Python, and learned how to preprocess a dataset into something useable for analysis.

What's next for fairLends

We hope to partner with banks to gain access to more comprehensive datasets, enriching our analysis and providing more thorough insights into lending practices.

Built With

• ai
• api
• gpt-3.5
• matplotlib
• python
• scikit-learn
• statsmodels