In frbl/OnlineSuperLearner: Online SuperLearner package

This guide describes the general architecture of the Online SuperLearner.

Data

In order to be able to deal with both static data (i.e., data that is available upfront, data.tables or data.frames) and to be able to deal with streaming data (i.e., data that comes in as a stream, for example from kafka), a small architecture of wrapping objects is created. This architecture looks as follows:

IMAGE HERE, see the vignettes/onlineSuperLearner-data.pdf file.

The reason for this abstraction is to allow the same methods (getNext and getNextN) to be used for both data.frames (or data.tables) and data streams. What the Online SuperLearner does is treat both instances of data as streaming data.

frbl/OnlineSuperLearner documentation built on Feb. 9, 2020, 9:28 p.m.