Predicting Customer Spend with Linear Regression

What BigQuery ML Linear Regression is:

Linear regression (model_type='linear_reg') is a supervised machine learning algorithm used to model the relationship between one or more independent input features and a continuous numerical target (label). In BigQuery ML, you train built-in models using standard SQL through the CREATE MODEL (or CREATE OR REPLACE MODEL) statement.

How it meets the requirements:

• Predicting continuous values (✓): The business goal is to estimate customer spending (total_spend), which is a real-valued, continuous numerical metric. Setting model_type='linear_reg' creates a linear regression model optimized for continuous target variables.
• Designating the target label (✓): The input_label_cols=['total_spend'] parameter explicitly tells BigQuery ML which column in the query output serves as the training label (target) that the model should learn to predict.
• Supplying feature data (✓): The AS SELECT ... query provides both the training features (age, country, previous_purchases) and the ground truth label (total_spend) directly from the table.
• Standard BigQuery ML syntax (✓): Uses the proper CREATE OR REPLACE MODEL ... OPTIONS(...) AS SELECT structure native to BigQuery ML.

Key syntax components in BigQuery ML:

• CREATE MODEL clause: Defines the destination dataset and model name within BigQuery.
• OPTIONS() clause: Configures model hyperparameters, algorithm selection (model_type), and structural requirements such as input_label_cols.
• AS SELECT query: Defines the training dataset, including both input feature columns and target label column(s).
• Automatic feature preprocessing: BigQuery ML automatically handles standardization for numerical columns and one-hot encoding for categorical columns like country.

Why this is optimal:

This statement allows analysts to train and store machine learning models directly inside Google Cloud's data warehouse without extracting data to external environments or writing Python/R code. It accurately defines the correct model family and syntax required for continuous numerical forecasting.