Challenge 4: Building a Demand Forecast

Challenge 4: Building a Demand Forecast

Previous Challenge Next Challenge

Introduction

BigQuery ML can be used to build and deploy demand forecasting models using the ARIMA_PLUS algorithm. In this section, you use BigQuery ML to build a model to forecast the demand for products in store.

Description

In order to build a model, we need to prepare the data first. Create a new table that includes time_of_sale in hourly granularity, product_name and the total sales quantity for each product for the period between 2021-11-22 and 2021-11-28.

Using this training data create a new ARIMA_PLUS model to predict expect sales quantity for a product.

Once the model is ready use it to forecast the sales for all the products for the next 2 days (48 hours) and store it in a table.

Success Criteria

1. There’s a new table created for training purposes with the correct structure
2. BQML ARIMA model is trained using the created table
3. The predictions (forecasts) are stored in a new table

Learning Resources

• Creating and Using BigQuery Tables
• BQ timestamp functions
• Creating ARIMA_PLUS Model

Previous Challenge Next Challenge