AWS Data Science Meetup - Amazon Personalize (03/05/2020)

Conducted as an online meetup due to Coronavirus

Developing Personalization/Recommendation Services with Amazon Personalize

Speaker: Hyunwoo Bae
"Sharing experiences from researching Amazon Personalize."

Things the official documentation doesn't explain
Things non-data experts might be curious about
Errors that occur when using Personalize
Answers received from direct inquiries to AWS

What is Amazon Personalize?

A machine learning service that enables developers to easily create individualized recommendations for customers using their applications

Uses proven machine learning technology that Amazon.com has used for years, so developers without machine learning experience can easily build complex customization features into applications
Using Amazon Personalize, you can provide activity streams from applications such as clicks, page views, sign-ups, purchases, as well as inventory of items you want to recommend such as articles, products, videos, or music
You can provide additional demographic information about users such as age and geographic location
Amazon Personalize processes and reviews data, identifies meaningful patterns, selects the right algorithms, and trains and optimizes customization models tailored to your data
All analyzed data is kept private and secure and used only for custom recommendations
You can start custom recommendations through simple API calls
Pay only for what you use with no minimum fees or upfront commitments

Amazon Personalize is like having Amazon.com's machine learning personalization team available at all times!

Newly Released

Amazon Personalize Can Now Use 10x More Item Attributes to Improve Recommendation Relevance

Overview

Amazon Personalize is a machine learning service that enables users without machine learning experience to personalize websites, apps, ads, emails, etc. using custom machine learning models that can be created in Amazon Personalize. AWS is pleased to announce that Amazon Personalize now supports 10x more item attributes. Previously, you could use up to 5 item attributes while building ML models in Amazon Personalize. This limit has now been increased to 50. You can now use more information about items (e.g., category, brand, price, duration, size, author, release year, etc.) to improve recommendation relevance.

How?

To add item data to Personalize, you first define a schema to tell Personalize the column names in your item dataset and whether you want to send categorical or numerical values. Then, using this schema, you can create a dataset and import items as CSV files through S3. After importing 'user' and 'interaction' datasets following the same steps, you can train custom private personalization models with just a few clicks.

Availability

The increased limit for item metadata is now available in US East (N. Virginia, Ohio), US West (Oregon), Canada (Central), EU (Ireland), and Asia Pacific (Sydney, Tokyo, Mumbai, Singapore, Seoul) AWS regions. For more information, see the Amazon Personalize getting started guide.