Using the DNN models

In this section, we'll use the DNN model based on the ReLU function and trained to leverage the BigQuery ML capabilities to predict the duration of the bike rides for the New York City bike-sharing company.

To test our DNN, we'll use the ML.PREDICT function on the table prediction_table. Let's run the following SQL statement:

SELECT
   tripduration as actual_duration,
   predicted_label as predicted_duration,
   ABS(tripduration - predicted_label) difference_in_min
FROM
  ML.PREDICT(MODEL `11_nyc_bike_sharing_dnn.trip_duration_by_stations_day_age_relu`,
    (
    SELECT
         start_station_name,
          end_station_name,
          is_weekend,
          age,
  ...