Drawing business conclusions

Using the results that we got from the previous section, Using the multiclass logistic regression model, we'll draw some conclusions about the effectiveness of our ML model.

Enriching the previous query with a parent SELECT COUNT statement, we can count how many predictions are right compared to the total number of records.

Let's execute the following query to calculate how often our BigQuery ML model is able to correctly classify the trees in the classification_table table:

SELECT COUNT(*)
FROM (
      SELECT
        tree_id,
        actual_label,
        predicted_label_probs,
        predicted_label
      FROM
        ML.PREDICT (MODEL `06_nyc_trees.classification_model_version_3...