A clothing company is experimenting with different colors and materials for its products. The company stores the entire sales history of all its products in Amazon S3. The company is using…

Question

Options

• ATrain a custom ETS model.
• BTrain an Amazon SageMaker DeepAR model.
• CTrain an Amazon SageMaker -means clustering model.
• DTrain a custom XGBoost model.

• A
  10% (6)
• B
  85% (53)
• C
  2% (1)
• D
  3% (2)

Explanation

Amazon SageMaker DeepAR (B) is the correct answer because it is specifically designed to handle the 'cold-start' problem - forecasting demand for new items that have no historical sales data. DeepAR is a supervised deep learning algorithm that trains a single global model across all products in the dataset simultaneously. When forecasting a new product, it leverages learned patterns from similar existing products (e.g., similar colors or materials) to generate predictions. A custom ETS model (A) requires historical data for the specific product being forecast and cannot generalize to new items. K-means clustering (C) is an unsupervised algorithm for grouping data points - it does not produce demand forecasts. XGBoost (D) is a gradient boosting algorithm suited for tabular classification/regression tasks, not purpose-built for time-series demand forecasting of new products without history.

No community discussion yet for this question.