- Lab
- A Cloud Guru
Training Reports Utilized in SageMaker Debugger to Improve Your Models
Imagine you are the data engineer at your company, and your company has just selected AWS as the preferred cloud provider. You have been given a dataset to predict if an individual makes more than $50K in salary. As part of the modeling process, you have been asked to generate a summary of the model training evaluation results, insights into the model performance, and interactive graphs. In this lab, you will fetch the census data and use that as the input dataset. Once the data is split, the data is uploaded to the S3 bucket. Then, the Sagemaker estimator is configured with the debugger hook and Sagemaker built-in rules to generate performance metric reports.
Path Info
Table of Contents
-
Challenge
Launch SageMaker Notebook
- Log in to the AWS console and navigate to AWS SageMaker.
- Load the Jupyter Notebook that has been provided with this hands-on lab.
-
Challenge
Install dependencies and import the libraries
- Load required libraries.
- Fetch the census dataset from shap library.
-
Challenge
Download the data and upload them to S3 bucket
- Convert the downloaded data to a CSV format
- Use the
upload_filefunction and upload the CSV files to S3 bucket
-
Challenge
Set up training and validation data
- Create the inputs for the
fit()function with the training and validation location.
- Create the inputs for the
-
Challenge
Configure and run the estimator
- Fetch the xgboost image according to the region.
- Setup the estimator function with the debugger hook and debugger built-in rules.
- Initiate the training process.
-
Challenge
View the generated reports
- Fetch the training and profile reports from S3 bucket.
- Review the graphs generated by the training report.
- Understand the resource utilization statistics generated by the profiling report.
What's a lab?
Hands-on Labs are real environments created by industry experts to help you learn. These environments help you gain knowledge and experience, practice without compromising your system, test without risk, destroy without fear, and let you learn from your mistakes. Hands-on Labs: practice your skills before delivering in the real world.
Provided environment for hands-on practice
We will provide the credentials and environment necessary for you to practice right within your browser.
Guided walkthrough
Follow along with the author’s guided walkthrough and build something new in your provided environment!
Did you know?
On average, you retain 75% more of your learning if you get time for practice.
