
Ensembles of Classifiers 163
data streams. The method builds separate classifiers from sequential batches
of training examples. These classifiers are combined into a fixed-size ensemble
using a heuristic replacement strategy.
In a similar way, Wang, Fan, Yu, and Han (2003) propose a general frame-
work for mining concept-drifting data streams using weighted ensemble clas-
sifiers. They train an ensemble of classification models, such as C4.5, RIPPER,
naive Bayes, etc., from sequential batches of the data stream. The classifiers
in the ensemble are judiciously weighted based on their expected classification
accuracy on the test data under the time-evolving environmen ...