Great post!

Note that BigQuery ML is even easier than exposed, as it will automatically perform one-hot encodings. Instead of transforming strings to `0`, `1`, etc… just input the string!

Instead of:

CREATE OR REPLACE MODEL `xxx.xx`
OPTIONS (model_type='logistic_reg') AS
SELECT Survived AS label
  , CASE WHEN Sex = 'male' THEN 1 ELSE 0 END AS Gender

You could do:

CREATE OR REPLACE MODEL `xxx.xx`
OPTIONS (model_type='logistic_reg') AS
SELECT Survived AS label
  , Sex 

Or in instead of:

CASE WHEN Age BETWEEN 0 AND 12 THEN 1 ELSE 0 END AS Age_Child,CASE WHEN Age BETWEEN 13 AND 18 THEN 1 ELSE 0 END AS Age_Teenager,CASE WHEN Age BETWEEN 19 AND 35 THEN 1 ELSE 0 END AS Age_YoungAdult,CASE WHEN Age BETWEEN 36 AND 55 THEN 1 ELSE 0 END AS Age_Adult,

You could do:

CASE 
WHEN Age BETWEEN 0 AND 12 THEN 'Age_Child'
WHEN Age BETWEEN 13 AND 18 THEN 'Age_Teenager' 
WHEN Age BETWEEN 19 AND 35 THEN 'Age_YoungAdult'
WHEN Age BETWEEN 36 AND 55 THEN 'Age_Adult'
ELSE 'other'
AS age_cat

For other examples, see

https://towardsdatascience.com/when-will-stack-overflow-reply-how-to-predict-with-bigquery-553c24b546a3a