Time Series Analysis in Amazon SageMaker¶
Train, tune and deploy state-of-the-art machine learning models for time series in Amazon SageMaker
Overview¶
We provide Amazon SageMaker algorithms for multiple time series tasks, including forecasting, anomaly detection, clustering and classification. Each algorithm implements a state-of-the-art machine learning model designed specifically for time series.
Features¶
- Automated Data Handling
The algorithms work directly on raw time series data in CSV format. All the required data preprocessing and scaling is performed internally by the algorithm’s code.
- Automatic Model Tuning
The algorithms support automatic model tuning for optimizing the model hyperparameters in order to achieve the best possible performance on a given dataset.
- Incremental Training
Most of the algorithms support incremental training to continue training the model on the same dataset or to fine-tune the model on a different dataset.
- Accelerated Training
The algorithms were built by extending the latest deep learning containers and support both CPU and GPU training. Most of the algorithms also support multi-GPU training.
Documentation¶
Each algorithm has a dedicated GitHub repository with detailed documentation and step-by-step tutorials in Jupyter notebook format. Several use cases are also discussed in our blog.
Pricing¶
The algorithms are available on the AWS Marketplace on a usage-based pricing plan. Each algorithm offers a 5 days free trial.
Support¶
For support, contact support@fg-research.com.
Terms and Conditions