Automated HDB Trends Reporting

Problem

Monthly housing reporting is easy to delay and hard to standardize.

Even when the underlying HDB resale data is available, analysts still need to clean the data, compare the latest month against prior periods, add geographic context, and translate the results into something readable. That reporting burden compounds quickly.

Approach

I built an automated reporting workflow focused on Singapore’s HDB resale market.

• SQL was used to pull the latest HDB resale transactions across a rolling time window.
• Python handled the processing layer, including town-name standardization, monthly median calculations, and joins to geographic reference data.
• Town-level spatial data was merged into the workflow so the reporting layer could compare market movement across locations instead of only at the national level.
• A language-generation step was then used to convert the processed outputs into concise narrative summaries.

The point was not to automate opinion. It was to automate the repetitive descriptive work around trend reporting so the analytical layer could scale more reliably.

What the Workflow Produced

The reporting pipeline focused on two questions:

• how the latest month’s median resale prices compared with earlier periods
• which HDB towns were showing stronger or weaker price patterns

By combining transaction history with spatial context, the workflow could produce both numerical summaries and location-aware commentary. That made the final output more useful than a simple monthly table export.

Why It Mattered

This project sits close to the kind of work I want to keep building: analytical infrastructure for housing research.

The value was not only in the dashboard or the generated text. It was in creating a repeatable system that could keep translating raw HDB transaction data into readable market updates with less manual effort, more consistency, and clearer town-level context.