Evaluating trained models

Once the BQML model has been trained, you will want to evaluate the key performance statistics, depending on the type of model. You can do so by using the ML.EVALUATE function, as shown here:

ML.EVALUATE(MODEL model_name
           [, {TABLE table_name | (query_statement)}]
           [, STRUCT<threshold FLOAT64,
                     perform_aggregation BOOL,
                     horizon INT64,
                     confidence_level FLOAT64> settings]))])

Let’s look at the options you can specify as part of the evaluation...