This blog post contains some of my codings for the 2019 kaggle BigQuery-Geotab competition.

My submission scored 1st Place in the categorie BigQuery ML Models built in SQL.

The challange was to predict six measures for cars approaching intersections in four US cities.

My objective in the competition was to tryout BigQuery (BQ) including the basic ML features. Therefore the notebooks rely as much as possible on BQ. All features are generated in BQ with varying SQL-techniques. The prediction model is also build in BQ.

Additional resources can be found in my github repo.