nathan/go-weatherstation
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

work on model training

Browse Source
This commit is contained in:
Nathan Coad
2026-03-05 11:03:20 +11:00
parent 96e72d7c43
commit c8e38cd597
10 changed files with 534 additions and 30 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
README.md
View File

@@ -3,6 +3,7 @@
Starter weather-station data pipeline:
- MQTT ingest of WS90 payloads -> TimescaleDB
- ECMWF (Open-Meteo) forecast polling -> TimescaleDB
- Python rain-model worker (periodic training + inference writes) -> TimescaleDB
- Web UI with live metrics, comparisons, and charts
## Quick start
@@ -73,6 +74,7 @@ TimescaleDB schema is initialized from `db/init/001_schema.sql` and includes:
- `observations_ws90` (hypertable): raw WS90 observations with payload metadata, plus the full JSON payload (`payload_json`).
- `observations_baro` (hypertable): barometric pressure observations from other MQTT topics.
- `forecast_openmeteo_hourly` (hypertable): hourly forecast points keyed by `(site, model, retrieved_at, ts)`.
- `predictions_rain_1h` (hypertable): model probability + decision + realized outcome fields.
- Continuous aggregates:
- `cagg_ws90_1m`: 1‑minute rollups (avg/min/max for temp, humidity, wind, uvi, light, rain).
- `cagg_ws90_5m`: 5‑minute rollups (same metrics as `cagg_ws90_1m`).