EligiBot

Inspiration

We recognize that many people, especially those of lower socioeconomic status, struggle with understanding and comparing loan terms—especially factors like interest rates, repayment timelines, and hidden fees—we saw an opportunity to use linear optimization and AI to make this process more transparent and tailored. This application also comes in handy as an every day tool for college students or people who are new to careers and saving money.

What it does

By analyzing users' financial profiles (annual income, loan amount needed, term length, and credit score), EligiBot suggests the most financially sound option and educates users on what makes a particular loan a good fit for them. This would empower users to make informed decisions, avoid common pitfalls in borrowing, and ultimately support their financial well-being.

How we built it

We developed the front end of our application using React Native, with Flask handling backend operations and Axios for seamless API integration. We used Kaggle to take data, cleaned it, and fine-tuned a ML model. Our process began with an initial Figma design, which guided our UI/UX development before we moved on to backend functionality.

To determine loan eligibility, we trained a model using a custom-generated dataset containing financial profiles. Leveraging the train_test_split function from the Scikit-Learn (SKLearn) library, we split the data for training and testing, allowing the model to accurately assess a person’s eligibility based on their financial information.

Challenges we ran into

We encountered challenges with cleaning the data from Kaggle, so Ananya generated sample data specifically for determining loan eligibility. However, we still used the Kaggle dataset to train the models for optimization purposes.

Accomplishments that we're proud of

End-to-End Development: Successfully building and connecting both the front end (in React Native) and backend (with Flask and Axios) to create a functional application. Loan Eligibility and Optimization Models: Overcoming challenges in data cleaning and training effective models to assess loan eligibility and optimize loan recommendations based on user profiles. Custom Data Generation: Generating sample data for loan eligibility assessments when standard datasets proved challenging to clean, enabling us to move forward with model training. UI/UX Foundation: Designing and implementing a user-friendly UI/UX from an initial Figma prototype, laying the groundwork for a more refined version in future iterations. Data Privacy Focus: Ensuring user privacy by keeping personal identities anonymous as we expand database capabilities and aim for real-time global data integration.

What we learned

Data Cleaning/Management: Tackling challenges with large datasets and understanding the importance of data quality in training reliable models. Full-Stack Development: Gaining hands-on experience with the complete development cycle, from initial design in Figma through front-end and back-end integration. Machine Learning Implementation: Applying machine learning techniques, such as train_test_split from Scikit-Learn, to real-world financial data and understanding how to interpret model results for practical applications. User Privacy Practices: The importance of maintaining user privacy, especially when handling sensitive financial data, and the various ways to implement anonymization. Iterative Design and Development: Learning to adapt and improve through each stage of development, recognizing the need for refined UI/UX and robust authentication for a seamless user experience.

What's next for EligiBot

We plan to develop a more refined UI/UX design and implement an authentication system. Currently, we've completed the calculators and trained our models. Moving forward, we aim to expand our databases and incorporate real-time data from global sources, ensuring user privacy by keeping all identities anonymous.

Built With

• axios
• figma
• flask
• kaggle
• python
• react
• react-native
• sklearn