Amazon SageMaker Canvas features

Build highly accurate ML models using a visual interface, no code required

Prepare Data

SageMaker Canvas connects to 50+ data sources or you can upload local files to train your ML model. Tabular, image, or text data is supported. You can also write queries to access data sources using SQL and import data directly in various file formats, such as CSV, Parquet, ORC, and JSON, and database tables.
Through the SageMaker Canvas no-code interface you can explore, visualize, and analyze data using built-in or custom visualizations. With a single click, you can generate the Data Quality and Insight report to verify data quality, such as ensuring the dataset contains no missing values or duplicate rows, and also detect anomalies such as outliers, class imbalance, and data leakage.
SageMaker Canvas offers a selection of over 300 prebuilt, PySpark-based data transformations, so you can transform your data without writing a single line of code. Alternatively, you can use the foundation model-powered chat interface to prepare your data.
SageMaker Canvas makes it easy to launch or schedule a data preparation workflow to quickly process your data and scale data preparation across datasets, export it to a SageMaker Studio notebook, or integrate with SageMaker Pipelines.

Access and Build ML Models

SageMaker Canvas provides access to ready-to-use foundation model (FMs) for content generation, text extraction, and text summarization. You can access FMs such as Claude 2, Amazon Titan, and Jurassic-2 (powered by Amazon Bedrock) as well as publicly available FMs such as Falcon and MPT (powered by SageMaker JumpStart) tune them using your own data.

Through the SageMaker Canvas no-code interface, you can automatically build custom ML models using your company data. Once you select and prepare your data and identify what you want to predict, SageMaker Canvas identifies the problem type, tests hundreds of ML models based on the problem type (using ML techniques such as linear regression, logistic regression, deep learning, time-series forecasting, and gradient boosting), and creates a custom model that makes the most accurate predictions based on your dataset. Alternatively, you can bring your own previously-built model to SageMaker Canvas and deploy the model for inference.

SageMaker Canvas provides access to ready to use tabular, NLP, and CV models for use cases including sentiment analysis, object detection in images, text detection in images, and entities extraction. The ready-to-use models do not require model building, and are powered by AWS AI services, including Amazon Rekognition, Amazon Textract, and Amazon Comprehend.

After you’ve built your model, you can evaluate how well your model performs before deploying it to production using company data. You can easily compare model responses and select the best response for your needs.

Generate ML Predictions

SageMaker Canvas offers visual what-if-analysis so you can change model inputs and then understand how the changes impact individual predictions. You can create automated batch predictions for an entire dataset, and, when the dataset is updated, you ML model is automatically updated. After the ML model is updated, you can review the updated predictions from the SageMaker Canvas no-code interface.

You can deploy your SageMaker Canvas model to SageMaker endpoints for real-time inference.

Share model predictions with Amazon QuickSight to build dashboards that combine traditional business intelligence and predictive data in the same interactive visual. In addition, SageMaker Canvas models can be shared and integrated directly in QuickSight, allowing analysts to generate highly accurate predictions for new data within a QuickSight dashboard.

Leverage MLOps

You can register ML models created in SageMaker Canvas to the SageMaker Model Registry with a single click in order to integrate the model into existing model deployment CI/CD processes.

You can share your SageMaker Canvas models with data scientists who use SageMaker Studio. Then data scientists can review, update, and share updated models with you or deploy your model for inference.