No-code Machine Learning and Generative AI on AWS

With Amazon SageMaker Canvas, data and business analysts can prepare data, train, and deploy machine learning (ML) models without any ML experience or writing a single line of code. You will learn to build ML models for tabular and time series data without deep knowledge of ML. You will also learn to use, fine-tune, and evaluate output from foundation models from Amazon and other model providers to support generative AI use cases such as text generation, text summarization, and chat using retrieval augmented generation (RAG). With the help of a guided tutorial consisting of a narrated video, step-by-step instructions, and transcript, you can also try the Canvas service in your own Amazon Web Services (AWS) account.

In lieu of using your own account, you can become an AWS Skill Builder subscriber to unlock all our hands-on, interactive content, including unlimited access to 125-plus AWS Builder Labs. These hands-on, guided labs help you develop practical skills for common cloud scenarios by building in an AWS sandbox environment without the risk of accruing unwanted charges.

You can learn more and subscribe from the Subscriptions page. After subscribing, enroll in the course to take advantage of the AWS Builder Lab experience. After completing this course, you will be able to build and train highly accurate models, and generate predictions using batch inference. You will also be able to share models with data scientists for further analysis and deployment into your company’s ML operation pipelines.

• Course level: Intermediate
• Duration: 5.5 hours

Activities

This course includes eLearning interactions, knowledge checks, and follow-along demonstrations.

Course objectives

In this course, you will learn how to do the following:

• Describe basic machine learning (ML) concepts and techniques
• Identify the ML life cycle and its phases
• Describe the types of problems ML can solve
• Identify the steps to building an ML model
• Describe metrics for measuring the predictive accuracy of a model
• Explain how to use Amazon SageMaker Canvas to transform raw data into a training dataset.
• Describe how to generate data insights and understand data quality
• Identify how to find potential errors and extreme values in data with visualization tools
• Describe the model building capabilities of SageMaker Canvas using AutoML
• Use SageMaker Canvas to launch a model training job and track its progress
• Describe the model quality metrics available in performance reports.
• Deploy a model and make predictions.
• Use the SageMaker Canvas foundational model (FM) user interface (UI) for text generation, text summarization, and model comparison.
• Identify and address challenges with foundation model outputs using RAG and fine-tuning.
• Describe best practices to follow when using Amazon SageMaker Canvas

Intended Audience

This course is intended for the following:

• Data analysts
• Researchers from non-ML domains
• Operations research analysts
• Junior data scientists

Prerequisites

We recommend that attendees of this course have:

• Experience with analysis, cleansing, and transforming tabular or time series data
• Basic understanding of statistical measures and regression
• AWS Technical Essentials course

For those new to generative AI, we recommend the following courses:

• Introduction to Generative AI – Art of the Possible
• Planning a Generative AI Project

Course outline

Module 1: Introduction to Machine Learning

• How to Use This Course
• ML Introduction
  • ML Basics
  • Types of Problems ML Can Solve
  • ML Life Cycle
  • ML Life Cycle
• Building and Evaluating Models
  • Introduction to Model Building
  • Model Evaluation
  • Improving Model Performance
  • Model Training Options
• Wrap-up
  • Knowledge Check
  • Conclusion

Module 2: Data Analysis and Preparation

• How to Use this Course
• Introduction to Amazon SageMaker Canvas
  • Amazon SageMaker Canvas
  • Analyzing Data
  • Quick Model Analysis
• Data Preparation
  • Transforming Data
  • Exporting Data and Data Flows
• Wrap-Up
  • Knowledge Check
  • Conclusion

Demo 1: Amazon SageMaker Canvas Tutorial - Tabular Data Use Case

Demo 2: Amazon SageMaker Canvas Tutorial - Time-Series Dataset Use Case

Module 3: Model Building Using SageMaker Canvas

• How to Use this Course
• Deep Dive on SageMaker Canvas
  • Introduction to Building a Model in SageMaker Canvas
  • Advanced Options for Building Models in SageMaker Canvas
  • Evaluating a Model in SageMaker Canvas
  • Making Predictions and Deploying a Model in SageMaker Canvas
• Wrap-Up
  • Knowledge Check
  • Conclusion

Demo 3: Build a Custom Model Using Amazon SageMaker Canvas Tutorial

Demo 4: No-Code ML Capstone Lab Tutorial

Module 4: Generative AI using SageMaker Canvas

• How to Use this Course
• Foundational Models in SageMaker Canvas
  • Generative AI using Amazon SageMaker Canvas
  • SageMaker Canvas Foundation Models
  • Comparing Foundation Models
• Mitigating Foundation Model Challenges in SageMaker Canvas
  • Model Hallucinations
  • Retrieval Augmented Generation (RAG)
  • Fine-Tuning Foundation Models
• Wrap-Up
  • Knowledge Check
  • Conclusion

Module 5: Best Practices for SageMaker Canvas

How to Use this Course Best Practices

• Getting Access to SageMaker Canvas
• Updating SageMaker Canvas Version
• Saving Costs with SageMaker Canvas

Wrap-Up

• Conclusion

Keywords

• Gen AI
• Generative AI