Hi Andrew,

actually, I think Amazon Personalize would be a better fit. It seems to me that your problem looks more like recommendation than forecasting. Forecasting is really about predicting a continuous value over time, which I don’t think fits a list of locations.

With Amazon Personalize, all you have to provide is a dataset containing user-item interactions (say, user id, location id, timestamp). Optionally, you can also add a user dataset (say, demographics) and an item dataset (say, metadata on location like average monthly temperature, etc.).

Here’s a complete example: https://aws.amazon.com/blogs/aws/amazon-personalize-real-time-personalization-and-recommendation-for-everyone/