Gemini Generated

How to Use AWS Glue for Data Prep and SageMaker for ML Workflows

AWS Glue (Data Preparation / Data Cleaning)

AWS Glue is like a data kitchen where before the meal is prepared all the vegetables was washed and chopped.

When you have a raw data such as a customer details or a sales records it is often very messy sometimes names are missing there are spelling error or the data format are inconsistent.

It automatically find all your messy data clean it and converts it into a proper format.

This is also called ETL which means extracting data, transforming it to suit your needs and saving it in your secure location.

Get SULTAN ANSARI’s stories in your inbox

Join Medium for free to get updates from this writer.

Remember me for faster sign in

Suppose you have a file where people ages are written as twenty in some place Glue will automatically standardize everything so that the computer is understand it.

Amazon SageMaker (ML Model Building, Training & Deployment)

Amazon SageMaker is like an AI factory or workshop where you can handle the entire end to end process for machine learning models in one place. It consist of three main steps.

1. Model Building: Here you specify the type of prediction you require for example which customer will purchase a product and one which one will not.

• Sagemaker comes with many pre built algorithms and tools so you don’t have to write anything from scratch.

2.Model Training: Just as a student is taught for an exam using past exam papers, SageMaker provides your clean data to the model and instruct it to learn patterns from it.

• It utilizes powerful computers(GPUs) that can process and train even the largest datasets in a very short amount of time.

3. Model Deployment: Once a model is trained and has learned effectively you can make it live deploy it to production with the click of it button.

How do the two work together

• With AWS Glue, you can clean up your messy and raw data.
• This clean data is fed Amazon SageMaker so that it can train the ML model.
• The SageMaker model is deployed so that the business can obtain real time prediction.