Kopi C Kosong

Inspiration

Singapore keeps getting hotter. We had 29 high heat-stress days in 2025, up from 21 the year before. We notice in our neighbourhood seniors sitting in malls, coffee shops and void decks for periods to escape the heat. It made us think what about the seniors who can't easily get to a cool place, or who live alone and have no one checking on them on the hottest days?

How we'll build it

We're building everything on Databricks Free Edition. Every hour, a scheduled job pulls the latest temperature and heat stress readings from data.gov.sg. We clean those readings and match them up with census data on where seniors live, and with the locations of eldercare centres. A model then learns from past weather to predict tomorrow's heat in each town, and we use an AI model to turn those numbers into a briefing anyone can read. People see the results as a map, a chat box they can ask questions in, and a small app for finding a cool place nearby.

What we learned so far

We were surprised by how much the open data already showed us. On the same afternoons, Sembawang was 2.6°C hotter than East Coast Park, so heat really isn't the same across the island. We also found that cool places are spread very unevenly. In the latest open data, Hougang had one eldercare centre for every 12,300 seniors. Bukit Merah had one for every 1,600.

Challenges

Nobody publishes open data on how many people end up in hospital because of the heat, so we can't check our predictions against real health outcomes. Instead, we'll check them against the official heat stress readings.

The other problem is that the only open map of eldercare centres is from 2016, which is clearly out of date. We'll add community clubs and libraries so the picture is closer to what's actually out there today.