Training the time series forecasting model

In this section, we'll train the BigQuery ML time series forecasting model.

Let's start training the machine learning model liquor_forecasting, executing the following SQL statement:

CREATE OR REPLACE MODEL `08_sales_forecasting.liquor_forecasting`
OPTIONS
 (model_type = 'ARIMA',
  time_series_timestamp_col = 'date',
  time_series_data_col = 'total_sold_liters',
  auto_arima = TRUE,
  data_frequency = 'AUTO_FREQUENCY'
) AS
SELECT *
FROM
 `08_sales_forecasting.iowa_liquor_sales`; 

The SQL statement is composed of the following parts:

• The first lines of the query start with the keywords CREATE OR REPLACE MODEL, followed by the identifier of the machine learning model, `08_sales_forecasting.liquor_forecasting`, and by OPTIONS.
• Now let's focus on the options that we've used to train our machine learning model. The selected...