Can I use SageMaker Autopilot for my time-series modeling?

Question

I'm working on time-series modeling. I'm comparing battle-of-algorithms against the autopilot machine learning approach to identify the model that best fits my use case.
I understand that Amazon SageMaker Autopilot doesn't work with time series.
Is there an alternative library or algorithm in the AWS ecosystem that implements battle-of-algorithms for time series?

Accepted Answer

Amazon SageMaker Autopilot currently supports regression, binary classification, and multi-class classification. SageMaker supports only tabular data formatted in files with comma-separated values. For more information, see [Automate model development with Amazon SageMaker Autopilot][1].

[1]: https://docs.aws.amazon.com/sagemaker/latest/dg/autopilot-automate-model-development.html

For your use case, you can use [Amazon Forecast][2]. Amazon Forecast includes the [AutoML][3] feature that trains different models with your target time series, related time series, and item metadata. Amazon Forecast then uses the model with the best accuracy metrics.

[2]: https://docs.aws.amazon.com/forecast/latest/dg/what-is-forecast.html
[3]: https://docs.aws.amazon.com/forecast/latest/dg/automl.html

Answer

We introduced timeseries support in SageMaker Autopilot.

Please have a look to [this blog post](https://aws.amazon.com/blogs/machine-learning/amazon-sagemaker-autopilot-now-supports-time-series-data/) for details.

Thanks

Can I use SageMaker Autopilot for my time-series modeling?

相關內容