QuestionQ61

VerifiedArchitecting low-code AI solutions

You work for a retail company and are responsible for developing an ML model that predicts product sales. The company’s historical sales data is stored in BigQuery and contains features such as date, store location, product category, and promotion details. You need to select the most effective combination of a BigQuery ML model and feature engineering to maximize prediction accuracy. What should you do?

Explanation

Product-sales prediction is a regression problem. A BigQuery ML boosted tree regressor is well suited to tabular data with nonlinear relationships and interactions among promotions, stores, categories, and time. BigQuery ML supports boosted-tree regression and supports label encoding for nonnumeric categorical features; label encoding is the default categorical encoding method for boosted-tree models. Autoencoders are unsupervised, while matrix factorization is not the appropriate general-purpose model for this regression task.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!