Get data insights from a contribution analysis model using a summable metric

In this tutorial, you use a contribution analysis model to analyze sales changes between 2020 and 2021 in the Iowa liquor sales dataset. This tutorial guides you through performing the following tasks:

• Create an input table based on publicly available Iowa liquor data.
• Create a contribution analysis model that uses a summable metric. This type of model summarizes a given metric for a combination of one or more dimensions in the data, to determine how those dimensions contribute to the metric value.
• Get the metric insights from the model by using the ML.GET_INSIGHTS function.

Before starting this tutorial, you should be familiar with the contribution analysis use case.

Required permissions

• To create the dataset, you need the bigquery.datasets.create Identity and Access Management (IAM) permission.

• To create the model, you need the following permissions:

  • bigquery.jobs.create
  • bigquery.models.create
  • bigquery.models.getData
  • bigquery.models.updateData

• To run inference, you need the following permissions:

  • bigquery.models.getData
  • bigquery.jobs.create

Costs

In this document, you use the following billable components of Google Cloud:

• BigQuery ML: You incur costs for the data that you process in BigQuery.

To generate a cost estimate based on your projected usage, use the pricing calculator. New Google Cloud users might be eligible for a free trial.

For more information about BigQuery pricing, see BigQuery pricing in the BigQuery documentation.

Before you begin

1. In the Google Cloud console, on the project selector page, select or create a Google Cloud project.

   Go to project selector

2. Make sure that billing is enabled for your Google Cloud project.

3. Enable the BigQuery API.

   Enable the API

Create a dataset

Create a BigQuery dataset to store your ML model.

{
  "datasetReference": {
     "datasetId": "bqml_tutorial"
  }
}

Create a table of input data

Create a table that contains test and control data to analyze. The test table contains liquor data from 2021 and the control table contains liquor data from 2020. The following query combines the test and control data into a single input table:

1. In the Google Cloud console, go to the BigQuery page.

   Go to BigQuery

2. In the query editor, run the following statement:

   CREATE OR REPLACE TABLE bqml_tutorial.iowa_liquor_sales_sum_data AS (
     (SELECT
       store_name,
       city,
       vendor_name,
       category_name,
       item_description,
       SUM(sale_dollars) AS total_sales,
       FALSE AS is_test
     FROM `bigquery-public-data.iowa_liquor_sales.sales`
     WHERE EXTRACT(YEAR from date) = 2020
     GROUP BY store_name, city, vendor_name, category_name, item_description, is_test)
     UNION ALL
     (SELECT
       store_name,
       city,
       vendor_name,
       category_name,
       item_description,
       SUM(sale_dollars) AS total_sales,
       TRUE AS is_test
     FROM `bigquery-public-data.iowa_liquor_sales.sales`
     WHERE EXTRACT (YEAR FROM date) = 2021
     GROUP BY store_name, city, vendor_name, category_name, item_description, is_test)
   );

Create the model

Create a contribution analysis model:

1. In the Google Cloud console, go to the BigQuery page.

   Go to BigQuery

2. In the query editor, run the following statement:

   CREATE OR REPLACE MODEL bqml_tutorial.iowa_liquor_sales_sum_model
     OPTIONS(
       model_type='CONTRIBUTION_ANALYSIS',
       contribution_metric = 'sum(total_sales)',
       dimension_id_cols = ['store_name', 'city', 'vendor_name', 'category_name',
         'item_description'],
       is_test_col = 'is_test',
       min_apriori_support=0.05
     ) AS
   SELECT * FROM bqml_tutorial.iowa_liquor_sales_sum_data;

The query takes approximately 60 seconds to complete, after which the model iowa_liquor_sales_sum_model appears in the bqml_tutorial dataset in the Explorer pane. Because the query uses a CREATE MODEL statement to create a model, there are no query results.

Get insights from the model

Get insights generated by the contribution analysis model by using the ML.GET_INSIGHTS function.

1. In the Google Cloud console, go to the BigQuery page.

   Go to BigQuery

2. In the query editor, run the following statement to select columns from the output for a summable metric contribution analysis model:

   SELECT
     contributors,
     metric_test,
     metric_control,
     difference,
     relative_difference,
     unexpected_difference,
     relative_unexpected_difference,
     apriori_support,
     contribution
   FROM
     ML.GET_INSIGHTS(
       MODEL `bqml_tutorial.iowa_liquor_sales_sum_model`);

The first several rows of the output should look similar to the following. The values are truncated to improve readability.

The output is automatically sorted by contribution, or ABS(difference), in descending order. In the all row, the difference column shows there was a $31,595,222 increase in total sales from 2020 to 2021, a 7.9% increase as indicated by the relative_difference column. In the second row, with vendor_name=SAZERAC COMPANY INC, there was an unexpected_difference of $11,491,923, meaning this segment of data grew 28% more than the growth rate of the data as a whole, as seen from the relative_unexpected_difference column. For more information, see the summable metric output columns.

Clean up

1. In the Google Cloud console, go to the Manage resources page.

   Go to Manage resources

2. In the project list, select the project that you want to delete, and then click Delete.
3. In the dialog, type the project ID, and then click Shut down to delete the project.