big_query_kmeans.sql

WITH dataset AS (SELECT pm25,
                        pm10,
                        so2,
                        co,
                        o3,
                        temp,
                        press,
                        dewp,
                        rain,
                        timestamp,
                        station
                 FROM `GCP_PROJECT_NAME.DATASET_NAME.measurements`
                 WHERE pm25 IS NOT NULL
                   AND pm10 IS NOT NULL
                   AND so2 IS NOT NULL
                   AND co IS NOT NULL
                   AND o3 IS NOT NULL
                   AND temp IS NOT NULL
                   AND press IS NOT NULL
                   AND dewp IS NOT NULL
                   AND rain IS NOT NULL
                   AND timestamp IS NOT NULL
                   AND station IS NOT NULL
)
SELECT * EXCEPT (nearest_centroids_distance) FROM
  ML.PREDICT(MODEL `GCP_PROJECT_NAME.DATASET_NAME.pollution_kmeans`,
    (
      SELECT timestamp, station, pm25, pm10, so2, co, o3, temp, press, rain, dewp
      FROM dataset
    )
  )
ORDER BY station, timestamp DESC