BQML - Weather forecasting demo using ARIMA

BigQuery ML demo created by @elenamatay in November 2022 for demo purposes.

This demo included:

• Preparing the training data in BigQuery
• Training and evaluating a time-series model with BigQuery ML
• Visualising the forecasts in a Looker Studio dashboard
• Scheduling and automate model retraining

The code included in this repo covers the first two points + comparing the model results (average temperature forecast) with the real values (coming from the NOAA GFS public dataset) for later visualisation in LookerStudio

Logical order of the files:

1. create-curated-source-table.sql
2. model-create.sql
3. model-evaluate.sql
4. model-predict.sql
5. create-forecast-vs-real-table.sql