Machine Learning

Demand Forecasting with Amazon Forecast - Ingesting Time Series Data and Improving Prediction Accuracy

Learn how to build time series forecasting models with Forecast, leverage related data, and export prediction results.

約 3 分で読めます

Forecast Overview

Forecast is an ML-based time series forecasting service that predicts future values from up to 5 years of historical data. You input time series data such as past sales data, web traffic, and inventory data, and it automatically builds models to predict future values. It leverages the same ML algorithms used for demand forecasting at Amazon.com, and AutoML automatically selects the optimal model. It provides probabilistic forecasts with P10, P50, and P90 prediction intervals.

Data Preparation and Forecasting

In addition to target time series data (historical data of the forecast target), adding related time series data (weather, promotion status, pricing) and item metadata (category, brand) improves prediction accuracy. AutoML builds models with multiple algorithms including DeepAR+, Prophet, and ETS, then automatically selects the most accurate model through backtesting. Probabilistic forecasts provide three prediction values: P10 (optimistic), P50 (standard), and P90 (pessimistic). For inventory planning, use P90 to reduce the risk of stockouts.

Improving Prediction Accuracy and What-If Analysis

Forecast prediction accuracy depends heavily on input data quality and the richness of related data. Adding related time series data (weather, holidays, promotions) allows the model to learn demand fluctuations caused by external factors, improving accuracy. Item metadata (category, brand, price range) is effective for mitigating the cold start problem (predicting demand for new products). What-If analysis lets you set hypothetical scenarios such as price changes or promotions and simulate their impact on demand. Check the Predictor's backtest results for WAPE (Weighted Absolute Percentage Error) and RMSE to quantitatively evaluate prediction accuracy. For practical techniques related to Forecast, you can also check out related books on Amazon.

Forecast Pricing

Forecast pricing consists of generated forecasts, data storage, and training time. Forecasts cost approximately $0.60 per 1,000 forecasts, and data storage costs approximately $0.088 per GB per month. AutoML trains multiple models for automatic algorithm selection, so training costs are higher than when manually specifying an algorithm. If prediction accuracy is sufficient, reduce costs by changing training frequency from weekly to monthly. Set the forecast horizon (prediction period) to the minimum required by business needs and avoid unnecessarily long prediction periods to optimize costs.

Summary

Forecast is a service that automatically builds ML models from time series data to provide demand forecasts. AutoML automatically selects the optimal algorithm, and adding related time series data (weather, promotions) improves prediction accuracy. It provides probabilistic forecasts with P10/P50/P90 prediction intervals and supports scenario-based decision-making through What-If analysis.

Related Services

Amazon ForecastA fully managed service that uses machine learning to forecast time series dataAmazon S3A scalable object storage serviceAmazon QuickAn agentic digital workspace that handles everything from answering questions to executing actions

Related Articles

Building a Time Series Analytics Platform with Amazon Timestream - IoT Data Ingestion and Query OptimizationIngest IoT sensor data into Timestream and perform time series analysis with SQL-based real-time aggregation and scheduled queries. Covers cost optimization through automatic tiering between memory and magnetic stores.Demand Forecasting - Predicting the Future from Time Series Data with Amazon ForecastInput historical time series data and related variables to automatically build ML-based demand forecasting models. This guide covers forecast accuracy evaluation metrics and patterns for leveraging forecast results through S3 and QuickSight integration.Centralizing SaaS Audit Logs with AWS AppFabric - OCSF Standardization and Security Lake IntegrationLearn how AppFabric collects audit logs from SaaS applications, standardizes them to OCSF format, and builds analysis pipelines.

More on This Topic

GPU-Based Machine Learning Training with AWS Batch - Cost-Efficient Large-Scale TrainingRun GPU training with your existing Docker containers, and cut costs by up to 90% using Spot Instances and checkpointing. Includes guidance on when to choose Batch over SageMaker.Using Claude on Amazon Bedrock - Model Selection, Prompt Design, and Cost OptimizationCompares the Anthropic Claude models available on Amazon Bedrock, provides model selection guidelines by use case, and covers prompt design best practices and cost optimization.Building RAG Applications with Amazon Bedrock Knowledge Bases - Implementing Retrieval-Augmented GenerationAutomatically index documents on S3 and unify search and generation with the RetrieveAndGenerate API. Covers chunking strategy selection and safety enforcement with Guardrails.Getting Started with Quantum Computing on Amazon Braket - Designing and Simulating Quantum CircuitsPrototype for free with local simulators, then run quantum circuits on IonQ and Rigetti hardware. Covers implementing VQE and QAOA with hybrid jobs.Privacy-Preserving ML with AWS Clean Rooms ML - Build Models Without Sharing DataLearn how to build lookalike models with Clean Rooms ML, apply differential privacy, and leverage the results for ad targeting.Implementing Natural Language Processing with Amazon Comprehend - Sentiment Analysis and Entity ExtractionLearn about sentiment analysis, entity extraction, and building custom classification models with Comprehend.Building Conversational Bots - Natural Conversation Interfaces with Amazon Lex and PollyLearn how to build conversational bots using Amazon Lex and Amazon Polly.Demand Forecasting - Predicting the Future from Time Series Data with Amazon ForecastInput historical time series data and related variables to automatically build ML-based demand forecasting models. This guide covers forecast accuracy evaluation metrics and patterns for leveraging forecast results through S3 and QuickSight integration.

Similar Articles and Services

Demand Forecasting - Predicting the Future from Time Series Data with Amazon ForecastInput historical time series data and related variables to automatically build ML-based demand forecasting models. This guide covers forecast accuracy evaluation metrics and patterns for leveraging forecast results through S3 and QuickSight integration.No-Code ML with Amazon SageMaker Canvas - Building Prediction Models with a Visual InterfaceA visual tool for building ML models without writing code. Simply upload a CSV and select a prediction target to build a model, with bidirectional sharing with Studio.Amazon SageMaker CanvasA no-code ML tool for building, training, and deploying machine learning models without writing codeAmazon TimestreamA serverless database purpose-built for storing and querying time series data