Demand forecasting can have a significant impact on a retailer’s business; factors like supply chain fluctuations and growing global markets can make it challenging to keep inventory in stock. Vertex AI Forecast can ingest datasets of up to 100 million rows from BigQuery or CSV files, covering years of historical data for thousands of product lines. The tool automatically processes the data and evaluates hundreds of different model architectures to create one model that should be relatively easy to manage. Users can include up to 1,000 different demand drivers (like color, brand, promotion schedule, or e-commerce traffic statistics) and set budgets to create the forecast. The tool also offers hierarchical forecast capabilities, which can minimize the challenges created by organizational silos and improve overall accuracy when historical data is sparse. For instance, the capability can tie together demand for an individual item at the store level and at regional levels. When the demand for individual items is too random to forecast, the model can still pick up on patterns at the product category level. Google cited a handful of customers already using Vertex AI Forecast, including Lowe’s, which uses it to create accurate hierarchical models that balance between SKU and store-level forecasts.
