Real Time Personalization and Recommendation with Amazon Personalize
You want to present a machine learning supported personalization experience according to your users’ actions and make recommendations for them. But you have no idea about machine learning. At this point, you can create a personalization experience according to your users’ actions by using Amazon Personalize service, which Amazon uses in its own e-commerce platform, and make recommendations for them.
According to McKinsey research, the revenue rate of companies that personalize with machine learning increases between 5 to15%, and the conversion rate increases between 10 to 30%. In e-commerce workloads, it is very important to show users appropriate content from thousands or millions of products and to create a personalized experience. In order to provide this experience, you must do plenty of hard work in the background and choose suitable platforms. You may not always have a team with expertise in this area. Just as Amazon does personalization and recommendation in its own platform, you can use this service on AWS and get very successful results. In this article, I will explain you the Amazon Personalize service,
How it Works — Create Data Set Group
First, you have to store your clickstream events on S3. The file format must be CSV. After that, you should create a data set group for yourself on Amazon Personalize and choose which feature to work with. In this example, I proceed through the user-item interaction data model. However, you have specified a schema in json format. Finally, you can complete this step by selecting your S3 bucket.
How it Works — Model Training
In this step, you choose which machine learning algorithm to run your data with. You can choose AWS ready algorithms here. If you do not have much experience in ML, I suggest you choose AutoML.
How it Works — Create Campaign
After choosing the best fit solution, we can start the process. As you can see below, you can test the process by entering any user ID.
As you can see, you can have your users experience Personalization in a very short time. After these processes, you can start making recommendations by using the personalize API.
See you in my next post!