ConfirmFare

Inspiration

HDB resale data is public but scattered. Prices are up about 55% since 2020, and many buyers (couples above the BTO income ceiling, same-sex couples, singles below the eligible age) have no other option. We want to create a tool that tells them if a flat is fairly priced and affordable for them.

What it does

Price Fairness Checker: a fair-price range for a specific flat, based on its features, nearby amenities and recent comparable sales. Personal Affordability: validates the price against your income, cash, CPF and debt, and flags red flags like MSR and loan-to-value limits. Market Explorer: trends and a 6-month town forecast that factors in BTO flats reaching their MOP, explorable through AI/BI Dashboards and Genie.

How we plan to build it

A medallion pipeline on Databricks: data.gov.sg and amenity data land in Unity Catalog, Spark Declarative Pipeline builds Bronze & Silver, Silver is a clean star schema with quality checks, and Gold holds flat features and market trends. MLflow tracks the fair-value and forecast models, which are served through a Databricks App with Lakebase for saved user profiles. Lakeflow Jobs keep it all refreshed.

Challenges

Stitching together six sources with different formats
The late-2024 Standard/Plus/Prime rule change, which compromises older comparisons
Accounting for non-linear price decay as lease shortens
Fitting a full pipeline, model and app into two weeks

Group Members

Lee Jia Kang - Y4 NUS Data Science & Analytics Quek Chui Qing - Y4 NUS Data Science & Economics Ang Kok Chun - Y4 NUS Data Science & Economics Aninda Metta Citta - Y4 NUS Data Science & Economics