Inspiration came from
HKO provides a vague estimation of rain forecast
What social problems do you solve?
- Not much people use it
- UI is quite bad
- In theory, it can be to-the-minute with estimations (sacrificing some accuracy)
How do we start from 0 to 1
- Parse data from HKO
- No official API seen, probably need to scrap from somewhere
- Create relay database (for further calculation and reduce gov server load)
- map out rain probability of locations
- If the data is in map-gird -> topography-like estimation of raincloud location
- If data is only available in single point per request -> queuing, sampling may be needed
- Prepare to send data out : API
- Input: coordinates, +- time (if historical data is kept)
- Output: Raining probability (0/1 or in percentage, depend on result)
- Front end app
- Presumably in single function app, for quick launch and notification
- Possibly provide rain probability for static locations Eg. display board of buildings;
The challenge we encountered
- I have a hard time retrieving data from HKO, with no background knowledge nor traning, this is the bottleneck
The resources we needed
- Help in getting data from HKO
- Data handling (depends on data received) (might be the most time and resource consuming area)
- Front end app sample (May require less help)
- Tester for accuracy
who are we
Self learnt coder, learning new programming languages by shipping small projects.
This is the biggest (potential) project so far, so I got a lot to learn
Links
GitHub workspace
https://github.com/thomassth/hkoRainfallUnofficial