EconPulse

Inspiration

The economy and its factors are a frequent point of conversation in our daily lives-- is the job market actually getting worse? Are we going into a recession? These questions are difficult to answer for the average student who doesn't have the time to understand all the data, news, and terminology behind the economy. EconPulse makes this information easily accessible and digestible with visualizations and models that illustrate how different factors affect the economy, so students can make informed financial choices.

What it does

EconPulse is a dashboard that measures the overall economic health and student cost pressure based on rent pressure, job market strength, inflation pressure, borrowing pressure, wage strength, and consumer sentiment. With different statistical and machine learning models(multiple linear regression, XGboost, LightGBM, MLP neural network), we can predict economic pressure for the next 3 months using measures like mean absolute error(MAE), root mean squared error (RMSE), and r^2(variation explained by the model).

How we built it

Hosted on Vercel. Frontend: React, TypeScript, Vite, Recharts, lucide-react, FigmaMake Backend: Python, pandas, NumPy, scikit-learn, XGBoost / LightGBM, JSON data pipeline

Challenges we ran into

Because this is the first hackathon for all the group members, consisting of 2 data science and 2 statistics majors, we have not had experience building a product using real data from back to front end. We had to use an API key for the first time to convert the data into something usable, and there was a lot of struggling to integrate the backend with the frontend, and thinking about the UI/UX of the website.

Accomplishments that we're proud of

The data was cleaned by group members from the official Federal Reserve Economic Data, instead of using pre-cleaned datasets from sites like Kaggle. EconPulse also has multiple graphs and statistical/ML models that the user can interact with, including clean visuals.

What we learned

Though we came into the hackathon after doing an ideate session prior, we completely pivoted our idea a few hours in. Next time, we’d like to know for sure what we’re doing by the time the hackathon starts. There was also a lot of delay in transferring work between group members, so it’d be more efficient if we could streamline the passing off of work, or making sure members can work based on placeholder data.

What's next for EconPulse

It would be nice if we could automate the website updating so it can keep up with real-time data. Additionally, we may want to fine-tune model accuracy or try adding improved models for economic prediction. We would also want to do user testing to improve the UI/UX of the website.