adding minor refactors

Author: hejnal

sql/bqml_create_and_evaluate_model.sql

@@ -0,0 +1,73 @@
+#standardSQL
+--demo baseline model
+CREATE
+OR REPLACE MODEL `<model_dataset>.<model_name>` OPTIONS(
+  model_type = "LOGISTIC_REG",
+  data_split_method = "AUTO_SPLIT",
+  input_label_cols = ["will_buy_later"],
+  enable_global_explain = TRUE,
+  auto_class_weights = TRUE
+) AS
+SELECT
+  bounces,
+  time_on_site,
+  page_views,
+  source,
+  medium,
+  channel_grouping,
+  is_mobile,
+  add_to_cart,
+  product_detail_view,
+  will_buy_later
+FROM
+  `<curated_dataset>.<table_name>_ready_for_ml`
+WHERE
+  split_col = 'training';
+
+-- evaluate model
+SELECT
+  *
+FROM
+  ML.EVALUATE(MODEL `<model_dataset>.<model_name>`,
+    (
+    SELECT
+      bounces,
+      time_on_site,
+      page_views,
+      SOURCE,
+      medium,
+      channel_grouping,
+      is_mobile,
+      add_to_cart,
+      product_detail_view,
+      will_buy_later
+    FROM
+     `<curated_dataset>.<table_name>_ready_for_ml`
+    WHERE
+      split_col = 'validation'),
+    STRUCT(0.5 AS threshold));
+
+-- predict results
+SELECT
+  *
+FROM
+  ML.PREDICT(
+    MODEL `<model_dataset>.<model_name>`,
+    (
+      SELECT
+        bounces,
+        time_on_site,
+        page_views,
+        source,
+        medium,
+        channel_grouping,
+        is_mobile,
+        add_to_cart,
+        product_detail_view,
+        will_buy_later
+      FROM
+        `<curated_dataset>.<table_name>_ready_for_ml`
+      WHERE
+        split_col = 'test'
+    ), STRUCT(0.5 AS threshold)
+  );