MongoDB is used as the datastore for unstructured data, e.g. item attributes and user attributes. It's also used as a cache for prediction results, so queries like geospatial search can be performed.
There is no specific reason reason to stick with MongoDB only. It just happens to be the database the team has picked for the first implementation. It is very likely to support other databases in the coming future given the strong community demand.
It describes the real differences among Machine Learning, Predictive Analytics, Data Analytics, Classification, Pattern Recognition and Data Science, Statistical Analysis, Data Warehousing, Data Mining, Knowledge Discovery, Artificial Intelligence and Business Intelligence etc....