Reproduce “Enhancing User' s Income Estimation with Super-App Alternative Data” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Gabriel Suarez, Juan Raful, Maria A. Luque, Carlos F. Valencia, Alejandro Correa-Bahnsen
Researchers at Rappi developed an XGBoost model utilizing extensive behavioral and transactional data from a Super-App to estimate user income, achieving statistically superior accuracy compared to traditional financial bureau estimates. The alternative data proved sufficient, as combining it with bureau data did not yield a statistically significant performance improvement.
Key insights
- Super-App alternative data significantly enhances the statistical performance of income estimation models compared to models based solely on traditional financial system information.
- The specific industry-accepted bureau income estimates examined did not provide substantial incremental information beyond what was already captured by the Super-App alternative data.
- User behaviors such as frequent high-value consumption (e.g., expensive restaurants), engagement with financial services, and credit card preference are strong positive income indicators, while discount seeking suggests lower income.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.