Training the DNN models

Now that we've segmented the dataset into multiple tables to support the different stages of the ML model life cycle, let's train our DNN regression models using different activation functions:

1. First of all, we can start with the training of a DNN model by using the RELU function. Let's execute the following SQL statement to create the machine learning model `11_nyc_bike_sharing_dnn.trip_duration_by_stations_day_age_relu`:

   CREATE OR REPLACE MODEL `11_nyc_bike_sharing_dnn.trip_duration_by_stations_day_age_relu`
   OPTIONS
     (model_type='DNN_REGRESSOR',
           ACTIVATION_FN = 'RELU') AS
   SELECT
     start_station_name,
     end_station_name,
     is_weekend,
     age,
     tripduration as label
   FROM
     `11_nyc_bike_sharing_dnn.training_table`;

   In the SQL statement, we can notice the keywords CREATE OR REPLACE MODEL used to create...