Recommending attractions through SageMaker FMs

FMs are one of the most widely used algorithms for making recommendations when it comes to very sparse input. It is similar to the stochastic gradient descent algorithm we discussed when looking at the model-based matrix factorization methodology. In this section, we will show how to use AWS' built-in algorithm implementation of FMs to get recommendations for our theme park visitors.

Preparing the dataset for learning

In order to use an FM algorithm, we need to prepare our dataset in a certain way. We will pose the recommendation problem as a regression problem in which the input is a pair (a user and an attraction) and the output is the expected level of interest this...