This is a repository with experiments that evaluate a hybrid framework combining active learning and self-labeling strategies for data streams classification. The project includes: source codes for the investigated configurations and experiments and data streams for tests.
- Install Java 1.8
- Unzip data.7z into data directory
- The experiments start in eval/Evaluator.
- Results can be post-processed using output/ResultProcessor
- One can change a classifier in eval/experiment/ExperimentRow class or data streams in eval/experiment/ExperimentStream
Warning: All data streams must be properly unzipped into the mentioned directory. Otherwise the experiments will fail. Also, all experiments must be finished before running the post-processing.
- MOA 17.06 (https://moa.cms.waikato.ac.nz/new-release-of-moa-17-06/) -> moa.jar and sizeofag.jar
- jfreechart 1.0.19 (http://www.jfree.org/jfreechart/download.html) -> jfreechart-1.0.19.jar and jcommon-1.0.23.jar