Model Development & Testing

6 hours of instruction

This practical, hands-on course dives into the next stages in the ML production cycle – model development and testing.

OBJECTIVES

1. By the end of this course, participants will be able to train, test, and evaluate models and pick the best model
2. The students will also set up all the necessary cloud platform services required to perform the above tasks

PREREQUISITES

ML Introduction & Data Preparation

SYLLABUS & TOPICS COVERED

1. Data preprocessing methods
   • ‘Summarize data preprocessing and its importance in model development’
   • ‘Explain the steps for data preprocessing in AWS SageMaker (1st approach)’
   • ‘Perform the initial setup steps for data preprocessing (2nd approach)’
2. Conducting data preprocessing
   • ‘Implement prerequisites for preprocessing’
   • ‘Perform data ingestion for preprocessing’
   • ‘Create features by using feature engineering for preprocessing’
   • ‘Split the dataset’
3. Model-training in SageMaker
   • ‘Explain the possible training approaches in SageMaker’
   • ‘Train a model using the built-in algorithm in SageMaker’
4. Hyperparameter tuning in SageMaker
   • ‘Compare various methods for hyperparameter tuning’
   • ‘Apply the API approach for hyperparameter tuning in SageMaker’
   • ‘Apply the SDK approach for hyperparameter tuning in SageMaker’
5. Model evaluation in SageMaker
   • ‘Explain evaluation techniques’
   • ‘Evaluate a model in SageMaker’
6. Building an ML pipeline
   • Initialize the pipeline
   • Preprocess and store the pipeline model
   • Execute pipeline hyperparameter tuning
   • Evaluate the pipeline model
   • Complete pipeline model register and definition

SOFTWARE REQUIREMENTS

API Gateway, AWS Sagemaker, Access to AWS accounts, CodeBuild, CodePipeline, Lambda, S3