DataSwish

Inspiration

HDB flats are among Singaporeans’ biggest financial decisions, yet buyers and sellers often lack clear, property-specific insight into whether a price is fair or when to transact. Planners also need clearer evidence on affordability and price movements across towns.

What it does

DataSwish is an HDB intelligence platform that visualises price trends, forecasts resale prices, assesses affordability, and explains fair value through transparent price drivers.

It helps buyers assess whether to buy now, sellers benchmark asking prices and offers, and planners identify local factors affecting HDB prices.

How we built it

We built an end-to-end Databricks pipeline using HDB resale transactions, household income, housing supply, MRT access, school distribution, policy releases, and search-interest data.

We combine hedonic regression, SARIMAX forecasting, affordability indicators, and symbolic regression with genetic programming to produce interpretable fair-value estimates.

Challenges we ran into

Our key challenge was balancing prediction accuracy with explainability. A price prediction is not enough if users cannot understand the factors behind it.

We also had to reconcile datasets with different update frequencies, geographic definitions, and levels of detail.

Accomplishments that we're proud of

We created a single framework that serves buyers, sellers, and planners rather than only one group of market participants.

We also designed the platform to explain why a flat is priced a certain way, not simply predict its price.

What we learned

HDB prices are shaped by more than flat type and size. Remaining lease, accessibility, nearby amenities, local housing supply, affordability, and market conditions all interact.

We learned that transparent and interpretable insights are essential for high-stakes housing decisions.

What's next for DataSwish

We will test our models on 2025 to 2026 transactions against nearby-sales and town-median benchmarks.

We also plan to beta test whether buyers and sellers estimate fair value more accurately, and whether housing experts can identify local price drivers and interventions faster.