Learning Amazon SageMaker (2019)

1h 12mIntermediate2019-04-30

Authors

Martin Kemka

Martin Kemka is the founder of Northraine, a machine learning production house.

Course details

SageMaker is Amazon’s solution for developers who want to deploy predictive machine learning models into a production environment. Programming is done in Python and the results can easily be integrated into cloud-based applications. These lessons review the entire Amazon SageMaker workflow: analysis, build, and final deployment. Instructor Martin Kemka introduces the benefits of Amazon SageMaker and reviews its browser-based interface and toolset. In the second chapter, he shows how to import, investigate, visualize, and summarize your data. The next stage is to use a clean data sample to train a machine learning model to fulfill a basic task. Finally, Martin shows how the model is deployed. Almost every chapter concludes with a challenge that allows you to practice your new SageMaker skills.

Topics include:
Benefits of SageMaker
Importing data
Investigating data
Visualizing data
Cleaning the data
Training the model
Deploying the model
Testing the deployed model

Skills covered

Amazon SageMakerCloud DevelopmentAmazon Web Services (AWS)AmazonCloud ServicesLearningCloud Computing

Concepts

0. Introduction

01 - Machine learning with Amazon SageMaker
02 - What you should know

1. Introduction to SageMaker

03 - What is Amazon SageMaker
04 - How does Amazon SageMaker work
05 - Benefits of Amazon SageMaker
06 - Interacting with Amazon SageMaker

2. Analyze Data

07 - Data analysis tools
08 - Download and import data
09 - Investigate data
10 - Data visualization - Categories
11 - Data visualization - Numerical
12 - Data summary tools
13 - Challenge - Describe a dataset
14 - Solution - Describe a dataset

3. Build Models

15 - Cleaning up the data
16 - Preparing the model training set
17 - Model training
18 - Checking model training results
19 - Challenge - Train a basic model
20 - Solution - Train a basic model

4. Deploy Models

21 - Deploy trained model
22 - Test deployed model for single record
23 - Test deployed model for multiple records
24 - Challenge - Transfer model to server
25 - Solution - Transfer model to server
26 - Review the model for accuracy

Conclusion

27 - Next steps